UNION SQUARE VENTURES SESSIONS
October 20, 2005
540 West 21st Street New York, New York 10: 20 A.M.
Reported by Edward Leto
ATTENDEES:
CHRIS ANDERSON
YOCHAI BENKLER
MATT BLUMBERG
BRAD BURNHAM
JOHN BUTTRICK
TOMMY COHEN
DICK COSTOLO
TOM EVSLIN
SHANA FISHER
CHRIS FRALIC
MICHAEL FRUMIN
SETH GODIN
SETH GOLDSTEIN
HEATHER GREEN
UMAIR HAQUE
SCOTT HEIFERMAN
PETER HERSHBERG
MARY HODDER
CYRIL HOURI
JEFF JARVIS
RONY KAHAN
JOSH KOPELMAN
SCOTT KURNIT
HOWARD MORGAN
DAVE MORGAN
MARTIN NISENHOLTZ
CHARLIE O'DONNELL
TIM O'REILLY
MICHAEL PAREKH
NANCY PERETSMAN
JONAH PERETTI
MARC PINCUS
JOSHUA SCHACTER
PETER SEMMELACK
CLAY SHIRKY
BRUCE SPECTOR
JOEL SPOLSKY
JOSHUA STYLMAN
RIKKI TAHTA
MATTIAS TURCK
ALBERT WENGER
FRED WILSON
BOB YOUNG
MR.
BURNHAM: Folks, thank you for coming. This is what I guess is the
first of what we call Sessions, which is just an opportunity to get
some of the smartest people we know together to talk about what's
happening in the market. We are very, very excited about this group
and the set of topics -- I hope everyone's had a chance to see some
of the things on the Wiki that we put up. By the way, since we have
this group here, we can help decide, is it Wiki or Weeki?
THE GROUP: Wiki.
MR.
BURNHAM: It's Wiki. Thank you. There's a whole lot of things that
are going to be made clear to me today that have never been clear
before, like that. So there is a Wiki that most of you have probably
seen that has an agenda on it. There's an agenda in front of you.
We have a lot of things that we want to cover today, and so we're
going to try to move through it pretty quickly. Fred and I are going
to be -- since you know we paid for the green cloth here and that's
about it, but since we did, we're going to be pretty ruthless in
trying to keep to one conversation, which is going to be very hard,
but it's going to be impossible to hear people if we, you know, have
two or three conversations going on. Great opportunity during lunch
and during the breaks to have breakout conversations, but when we're
sitting down here, let's try to keep it to one conversation. There's
a lot of people here who have a lot to offer. With that, I'd just
like you to just dive into the first topic.
MR. WILSON: Brad,
let's hang on. We have some rules, we should go over them, but also,
I'm not sure that the acoustics will be great in here. In fact, I'm
sure that they won't be, so try to speak up so that the people on the
other side of the room can hear you, and if you have something you
want to get out and you can't get it out, feel free to raise your
hand and kind of do this and Brad and I will try to play, you know,
business school professor or whatever the right word is and call on
you. The rules are that we would prefer people not blog or take
transcripts or notes of the event, because we have a stenographer who
is transcribing the whole thing. That's going to be up on our Wiki,
and hopefully that will be a good source of notes and fodder for you
to go grab and blog tonight or tomorrow and over the weekend. We're
also going to post all the pictures that Mark is taking on the same
place, so if you want to grab some pictures, and you know say this
was Jeff Jarvis getting apoplectic, that would be fine. Also, cell
phones, we prefer people not taking cell phone calls during the
session. If you're expecting one, Kerri's here and we'd be happy to
hold your phone and take the call for you and pull you out of the
meeting. The other thing is that we think it's going to be pretty
easy to get cabs on 21st and 10th, but if you're particularly nervous
about getting somewhere, I know a few of you have to get somewhere,
feel free to ask Kerri to schedule a car service to come take you.
If it's something that's really important, we're happy to do that for
you.
MR. BURNHAM: So the other thing we should probably mention
is what our objective is here, and our objective really is only to
try to push the conversations forward. There are a lot of very
interesting things happening in the market. There's a lot of
transformative things that are happening, and we don't have any
proprietary interest in the output of this. We want everybody to
feel free to blog anything. We're going to put the entire transcript
up on the Wiki. That's not promoted, so it's not going to be broadly
publicized or anything, but it's going to be there, so if there's
something that you don't want to be known, you should not introduce
it here, but hopefully, what we're going to do is help clarify some
thinking around the trends that we're seeing, the events that we're
seeing, and help us all think a little bit more clearly about the
kinds of things that'll help move that what I think is generally a
very positive set of developments forward. So that's our goal. The
first topic is really what is peer production. All of us have some
ideas on peer production. We're very fortunate today to have Yochai
-- was I close?
MR. BENKLER: Yes.
MR. BURNHAM: Yochai
Benkler here with us today who coined the term in his paper, Coase's
Penguin, common space peer production. He's thought a lot about it.
Thought a lot about it early. I was wondering if I could ask Yochai,
just by way of introduction, to talk for a minute about what he
thinks it is, because he probably has a better idea than the rest of
us.
MR. BENKLER: Except that for me to talk for a minute is
really hard. I think the critical characteristics, as far as I can
see, is that it is a set of social behaviors that have economic
outputs, but are organized along motivational and organizational
forums that are developed in parts of life that traditionally were on
the periphery of the economy, that we understood ourselves as being
engaged in producing social relations and producing psychological
well-being, and so to the extent we thought about production of
things of value and organizing them in terms of markets and firms,
when we thought of motivations, we thought of material motivations
and we optimize our systems of thinking about how we get people to do
the things we want them to do -- whoever those other people are --
what it is that we expect will emerge out of these centralized
behaviors, all of that was optimized on the motivation dimension on
material well-being, and on the organizational level, either in terms
of the price system or in terms of organizational lines of authority.
I think what peer production is intended to capture is the fact that
a whole set of other behaviors that have grown up in the household,
in friendships, in communities, the motivations that they capture,
the signals that get people to explain what it is that they desire,
how they desire, what they want to do, what they're trying to do, all
of these things are suddenly becoming integrated into the core
economic activities of the most advanced economists, and all of the
players inside of these economies need to begin to think. It's a new
set of social competition. It's a new set of opportunities. It's a
new solution space for ways to solve production problems. And we need
to start learning how to live with, use, provide platforms for, use
the outputs of without undermining this new set of social cultural
practices.
MR. BURNHAM: So, you know, I think in your paper,
Yochai, you talk about Wikipedia, you talk about the NASA project
mapping, the craters on the moon. Are there a couple of other
representative examples that -- I mean, do you think of, as an
example, file sharing, MP3 file sharing as a form of peer production
or do you think of that as something else?
MR. BENKLER: Well, it
depends on what aspect of it. Peer to peer file sharing networks are
peer production of two different things, neither of which is music.
The first thing, more recently in 2004, I wrote a piece called
Sharing Nicely. It's about sharing of material resources,
computation bandwidth and processing, and the economics of that,
those which is the basic economics of those are different, because
these are not public goods. These are not long rival goods. And so
there was some working out to do with that. But the first thing that
P to P networks are is social provisioning of a distributed data
storage and retrieval system. So in 1999, if you were a professor
interested in a robust storage system, or a CEO interested in a
robust storage system and you came to your engineers and said -- in
the middle of 1999 -- build me a data storage and retrieval system
that would be available for something like 100 million users anywhere
in the world, all the time, that would be robust to attacks even to
the level of the central indexing server would be shut down, that
armed guards would know that malicious files would be inserted, and
nonetheless, all the data would be there and would be easily adopted,
again, after each of these attacks. You would say this will cost me
X 10s of millions of dollars and, of course, Shawn Fanning did it
easily because even when NASA was brought down, the data was still
there, all you needed was a new search algorithm. When Major Super
Notes were brought down, there were new Super Notes. So it's peer
production of a distributive data storage and retrieval systems. If
you ignore the copyright issues and you ignore the moral issues and
you just ask "What have we learned about peer production?"
We've learned about peer production that part of the solutions make
the problem. If the problem of production is a robust storage
system, P to P is an answer. And to some extent, the other thing is
accreditation. So the kinds of things that come out of the P to P
networks that are distributedly produced are the aspects of what the
recording industry does in regard to saying "this is good, this
isn't good," in terms of people basically saying, "Here, if
you like this, you'll like this as well."
MR. BURNHAM: Go,
Marc.
MR. PINCUS: Well, one other spin on peer to peer
production that hopefully is relevant to peer to peer is I keep
wondering, every time at the beginning of a new market segment,
industry segment, and I see everyone scurrying around building a
piece of it and building a lot of the same pieces, it's always
frustrating to me that there's various levels of participants in
every new segment, so I started with the social networking segment
and there might be a friend who's relatively far along, there might
be Yahoo who isn't in it but could have a lot of power, and then
there's a lot of companies doing individual pieces of it, like Flickr
was and like Tribe was trying to classify, and what was frustrating
me then and still frustrates me now is that what was clearly in the
interest of an end-user was for us, as service providers, to work
together in a peer to peer fashion so that -- and Tribe tried to
create something called a friend of a friend open standards so there
can be one social network. That was an interest of the end-user, and
in terms of getting the innovation to market, it would have created
better services for each company to provide a layer or interesting
service that interconnected with the whole, and instead, everyone
takes this approach that persists today of their own wall garden for
a while. And now I've been spending time looking at the advertising,
online advertising industry, and it's the same thing again, that so
many of these companies are trying to create their own ad networks,
and I've talked to most of them -- hopefully talk to Dave some more
today -- but I talked to most of them, and it seems like the exact
same thing that I went through in social networking where the
interests of both the advertiser and the publisher would be best
served if there were one open network, meaning in my humble opinion,
and then there were lots of companies that would try to innovate on
top of that open network. But, instead, you have, again, everyone at
various stages of ownership of the industry not wanting to let go --
MR. BURNHAM: You're talking about a subject that I would like to
get into some depth actually a little later today, which is the
relationship between standards and the role that standards play in a
peer produced system, but I want to get Tim in this.
MR.
O'REILLY: It seems to me that there's several different classes of
peer production, you know, and the formative essay for me on the
subject was one written by Dan Bricklin called the Cornucopia of the
Commons, which he published in 2001, and he basically pointed out the
ability to get volunteers to do it or you can architect the system in
such a way that it's produced as a byproduct of people's individual
selfish activity. And I thought that was a really profound insight,
because a lot of peer production examples that we see, like open
source software or Wikipedia are in that second category, people
coming together, network enable to actually build something
consciously contributing to it. Whereas Napster was a great example
of a system that was built in such a way that people were just
pursuing their own activities, but because of the way the system was
designed, you created, as a side-effect, this, you know, this peer
good.
MR. BURNHAM: Del.icio.us would probably be in that second
category.
MR. O'REILLY: That's right. The architecture of the
web, or, for that matter, the fundamental architect of the internet
is such that people can pursue individual activity. People put up
their own web site for their own purposes, and yet, there is a
collective good that was created. And it's kind of interesting,
because if you compare Napster with say Google, they're both
leveraging that same insight, but in the case of Napster, it was
actually built in from the get go as the driver for how the network
got built. In the case of Google, it was somebody coming along later
and saying, "Okay, we can actually leverage this system that's
actually been built without us," and they were able to design
some of the potential in that system and then leverage it. So I
think that's maybe kind of a fourth level where there is an
underlying architecture. Then there's a set of second order
phenomena that are driven by that architecture that were implicit in
it that someone can then capture to create something --
MR.
BURNHAM: Yes. Jeff.
MR. JARVIS: If you give people control,
they'll use it, and if you don't, you'll lose it. It's possible.
That's exactly what Marc's talking about here, and Tim as well, is
that the natural order of things is if I'm doing this as part of my
consumption, and I create my iTunes list, I'm consuming music, but
I'm creating a radio station. When I'm consuming peer to peer, I'm
helping create the network. When I'm tagging something, I'm now
creating a value, but if somebody comes in and tries to control that
in a way that doesn't serve me, then they may win for a short while,
but in the long run, we're all going to lose. But that's so against
our culture. That's so against our culture. It's hard to do the
things that Marc's talking about, but if you do, when those things
happen, if you're smart enough to let them be and then build on top
of them, there's tremendous value to be had. That's the new secret
success here, is to build on what people are doing in the normal
course of their lives.
MR. PINCUS: And the value that would be
achieved there accrues way more to the end user and to all the small
participants in the industry than to the biggest ones and builds the
whole rather than splitting up.
MR. BURNHAM: We're going to come
back later in the afternoon and really talk about what the impact on
the structure of these industries are and how you could potentially
invest or not invest in these industries, where the economic value
is. But let's stay kind of in the definition of what we're talking
about, peer production.
MR. EVSLIN: I think an interesting blog
you wrote about amateurs meeting with professionals, I think the
thing we found out is that anthropologically we have a much greater
urge to cooperate and to do cooperative things than we knew that we
had as a species. Other examples you asked before the help forums
that grow up around every possible service where there's a bunch of
volunteers basically providing export because they want to. I think
what happened is is that we assumed that people only did things that
they got paid for doing, in general, or at least the kind of work
activities, and that was true, but it wasn't only true because they
got a salary, it was true because the employers provided the expense
of facilities which used to be necessary to do those things. So you
couldn't be a writer before unless you had a publisher, because you
didn't have any way to print and distribute what it was that you
wrote. You couldn't be a tech support person off in cyberspace
because you liked helping people, because you just simply had no
access if nobody had any access to you. So all of a sudden, when
everybody becomes accessible to everyone else at very, very small
cost where you no longer need an employer to provide you a platform
for doing those things, we start to do a lot of social things,
because we --
MR. BURNHAM: Tom, you're talking about two things.
You're talking about, first of all, the availability of the
infrastructure, and secondly, the motivation, and one of the
interesting things that Yochai said was that the motivation appears
to be happening in western societies where, you know, the basic
hierarchy of needs is taking care of people. They're not hungry,
they're not worried about feeding their kids, and they have this
infrastructure, and they can begin to think about what else they can
do.
MR. EVSLIN: Right. But I don't think the urge to be
cooperative -- once you're well fed, given -- is a strange one. I
think we just didn't recognize it before because of the
unavailability of the resource, particularly to cooperate with
strangers. We did things in our communities. We did things in our
homes.
MR. BURNHAM: Let's get Michael and then Howard in here.
MR. PAREKH: Another thing, the term peer to peer, one of the
things that's implicit, unfortunately, is that everyone is
participating in an equal way, and the reality is in terms of all
these others people talk about, from Napster to Wikipedia, the number
of people that are actually doing -- contributing content, it's
typically a much lower percentage.
MR. BURNHAM: Well, Wikipedia
is 2,500 to 10 million or so users.
MR. PAREKH: Exactly. If you
look at the number of people who read blogs versus write blogs, the
number of people who put comments in and who actually have their own
blog, I mean, typically you're talking about percentages that are a
fraction of the whole system. And one of the things that I'm trying
to figure out in a lot of these examples is which of the systems that
have lower friction points. Like Del.icio.us, where you would assume
that a lot more people, because it's easier to put tags in, are
participating and contributing to the system as opposed to just
taking things out, and one of the things we need to be mindful about
in all the peer to peer conversation is how do you make these systems
more efficient? How do you get more people to participate, whether
it's trust factors, shyness factors, learning the tools. Whatever
the element is of that system.
MR. BURNHAM: The way you
participate is probably different based on which kind of system it
is, and how you facilitate participation is probably different.
MR.
O'REILLY: Everybody participates in Google, because they figured out
how to leverage what people do implicitly, as well as what they do --
MR. BURNHAM: Howard.
MR. H. MORGAN: One of the things that
comes out of that is utility function for people now is reputational,
so one of the things you're going to get from the peer networks is
reputation, and we see everyone contributing, as you said with
Google, where they're explicitly adjusting search results based on
other user behavior. You're contributing, but you don't know that
you're contributing. And inside pages, you're contributing, but
you're doing it for reputation purposes, because a review from a
friend or a review from someone you know enhances their reputational
value. Ebay is obviously a great example of that. And I think
looking at the utility function in the western world, we can afford
to do that, have a reputation utility function, whereas in other
parts of the world they still --
MR. BURNHAM: Chris and Fred.
MR. ANDERSON: It's interesting, to follow up on that point, the
reputation, of course, is -- rather than money, and it's interesting
that Google got the idea of patroning a citation analysis which was
the way they used to measure the hierarchy of individuals in --
MR.
EVSLIN: I think the point is that makes even lurkers participants,
because they're always voting by the pages that they reference. So
even on any forum, there's probably 50 times as many lurkers as there
are active posters, but everybody is a participant because everybody
is directing attention and where that attention --
MR. BURNHAM:
And the more meta data that's available, the more that becomes
possible. Fred.
MR. PAREKH: And my point is that there are
terms for that. We need to introduce new terms into the language so
we know how everyone should comprehend it, and it's not obvious until
you think about it. When you think about it, it's like "duh."
MR. BURNHAM: Well, we're going to end up talking about terms and
also rights and ownership and things like that at the end of the day,
but Fred.
MR. WILSON: I just was fascinated by the point that
either Michael or Tim made, which is that there are some peer
produced systems where the content is produced automatically, and
I've been fascinated over the past couple of weeks with this service
called last.fm which simply takes your iTunes listening and uploads
it to a social network, so every time I listen to my iPod or my
iTunes, I'm contributing data to social network. I do nothing, but I
know that I'm doing that, and it actually has changed the way I
behave. I actually listen to thing consciously, because I know by
listening, I'm contributing. It's a very odd behavior, but I think
it's something we should focus on.
MR. BURNHAM: Scott.
MR.
HEIFERMAN: As I was in the subway yesterday -- and I'm going to post
a picture of this on my photo log -- I saw this 20 year old guy
wearing a T-shirt that said "Fuck You, I Don't Need Any More
Friends." And so, of course, I had to go up to him and say, "Can
I take a picture?" And he was so friendly. But the point of
this is, you know, I wonder how this whole idea of implicit peer
production, you know, transparent peer production that you don't have
to -- there's a company being incubated at Kleiner Perkins called
Project Edgar, which is about asynchronous voice, and the way it was
pitched in a deck is "Now you can talk without talking."
And so it sounds like this whole just idea of peer production where
you don't even know you're involved, where you don't even know you're
participating in some sort of holy grail, and I would just throw out
there that part of the beauty of what we're talking about here is
that -- is that the world's a better place when we all -- when we
realize that we're in something together. We're all contributing to
Google. We're all making something that this transcends the
technology and all that, that you're a part of something bigger, and
so how does what we're talking about -- just to conclude -- how is
what we're talking about going to make it unfashionable to wear a
T-shirt that says "Fuck you, I Don't Need Any More Friends."
MR. BURNHAM: So is there a social benefit that's a natural
byproduct of this evolution that we're seeing.
MR. HEIFERMAN: In
the Napster heyday, I remember -- I mean, there was this certain
subtle euphoria that you felt like, you know, you walked down the
street -- and maybe it's just me -- but you walk down the street and
you said, "You know, these massive strangers, some of them gave
me music last night," and like I want to thank the world for
giving me all this great music. It sort of -- I think that's
something interesting about this notion of peer production, that
you're not just this isolated part of the world, but I'm interested
in how it's less transparent, how it's more -- I'm sorry, I'm
interested in how it's less transparent.
MR. BURNHAM: Tim.
MR.
O'REILLY: What I was going to say, one of the things it seems to me
as a characteristic of system -- this is returning to Marc's question
-- a defining characteristic seems to be how much is the value of the
system you try to capture. You know, if you look at something like
Ebay or Google, even though they're very, very successful companies,
they're actually pretty generous in creating value for people outside
the company, and you know, versus say if you look at sort of the
traditional wall garden kind of company, there's really not that
opportunity for creating value. You know, Pierre likes to point out
how many people make a living on Ebay and how the social goal of
getting people to trust each other is intrinsic to the value of the
system at creating an economic opportunity for people. Ebay gets a
small --
MR. BURNHAM: You know, it's funny, there is a
relationship between the economics of these systems, but there is a
relationship to Marc's point about standards in that it's also the
narrower standards that seem to be the ones that are most successful
that don't try and define the entire system, but our assessment is a
fairly narrow standard. It creates an opportunity for Dick's company
to solve a set of problems around that. It creates an ecosystem
around that. So it's facilitating these ecosystems that --
MR.
PINCUS: But isn't, Tim, I actually in today's -- in last year's
world, Ebay was a great example of an open platform. People could
have businesses on it. In today's world, they're one of the best
examples of a wall garden that doesn't get anymore. Why can't I
export my Ebay rating and use it somewhere else? That should be like
Paypal, a service they provide me that has an open API that let's you
import your Ebay rating and use that in the transactions that has
nothing to do with Ebay.
MR. O'REILLY: Well, I think companies
are learning that. I think it was interesting, at 2.0 Conference,
Yahoo and Google both introduced APIs, and clearly Google got
something Yahoo didn't, because Yahoo's was all about putting stuff
onto Yahoo, Google's was all about syndicating stuff out, and sure
enough, Google is the one that's sort of taking the world by storm.
So I think -- I guess to me the faith I have is that there's sort of
a science to this. I used to say this about open source, and it
either works or it doesn't, and our goal is to find out what works,
and in some of these areas, we're trying to figure out. Do you move
the dial this way and it stops working? Move the dial back that way,
it works better. And our goal should be as sort of the scientists of
the social value and economics to figure out where the boundaries
are, because you move it too far one way and you've created no
economic value for yourself, and therefore, you don't have the
incentive to either. Perhaps on the other hand, you try to capture
too much and you don't create the social beneficial network effects.
MR. BURNHAM: Let's get Mary in here, and then I want to get back
to Yochai.
MS. HODDER: I was just going to comment on what Marc
just said. Actually, you can pull data for reputation for Ebay.
Opinities pulls reputation data from Ebay, but the thing about the
difference between what Tim was talking about and maps and Ebay's
reputation information is that the mapping data makes sense when you
pull it out of the system, whereas the reputation data, because Ebay
is so skewed, it's such a bizarre social environment, everybody is
under tremendous pressure to make this sort of, you know, A+++ best
sale I've ever had, which, I mean, would only exist if the guy who
was selling you the thing drove me the item from Kansas or something,
otherwise it's just probably B+. But everybody wants to either make
something -- the reputation information perfect or it's terrible, and
so when you pull it out of that system and you put it onto Opinity,
it doesn't -- it doesn't match up. It doesn't translate with other
wall garden reputations.
MR. BURNHAM: It sounds like a failure
in the definition of the standard. It's too specific to that
environment.
MS. HODDER: Also, the peer production of that reputation system, it's sort of broken down and it's gotten really weird. Whereas the map data from Google makes sense in lots of different contexts.
MS. GREEN: Does that have anything to do with reputation?
MS. HODDER: Well, but
the interesting part for me is the Google data isn't peer produced.
The map stuff is stuff they make and send out, and the Ebay stuff is
something we're all making. But because they've made these really
weird social parameters, we're making it a very strange way.
MR.
BURNHAM: Let's go to Yochai and then Scott.
MR. BENKLER: Couple
of strengths I think that are worth underscoring from the first sort
of comments. One, I've actually started to use the term social
production to cover the two distinct phenomenon, Tim, that you're
describing, with individual action that's based on social
psychological motivations, that then gets coordinated by some
platform as simply common space production, in the sense that I
neither get the inputs out of the property, nor do I create the
outputs (inaudible), and peer production to describe the more
self-conscious cooperative platforms. And the reason or one reason
why that's important for people who are thinking about building
platforms or using outputs of this social system as inputs into their
own production process is it, again, characterizes, with a little bit
more detail, the set of strategies that are open to interact with the
social system. Sometimes you need affirmative cooperation. This
goes not to the social sense of participation, but to the practical
sense of what kind of platform we need to build. Is this kind of
behavior, that individual socially motivated behavior can cohere into
a useful information product, and if so, what's the intervention, as
a separate design question from no, this is something that actually
requires affirmative contribution and I need to build a cooperation
platform. So I think that's a very important distinction to separate
out, even though it's all part of the broader phenomenon of social
production where social motivation and organization, of course, are
pushing things that are economically valuable. The other thing,
though, just with regard to, Howard, when you mentioned issue of
reputation, I think you need (inaudible), which is reputation is one
social mechanism. It's true that it's very strong in academia, but I
think people also participate for a sense of -- as Scott was
suggesting -- of being part of a community. And sometimes that
actually requires understating the reputation, and sometimes
reputation isn't actually enough, and we don't have actually very
good research on the extent to which -- we do have very good research
on how adding money undermines social motivations depending on
contents. We don't have very good research, because there's never
been a research question that was important to people in designing
economic systems of whether or not you build a very strong reputation
system, you're actually undermining some of the solid -- because it
becomes "I'm better than you" or whether they work with
each other. But the design question is how you build a set of
motivations that's outside of the material in a way that's mutually
reenforceable. Might be reputation. Might be less reputation. It's
not homogeneous.
MR. BURNHAM: Scott was going to jump in.
MR.
KURNIT: I just want to go back to Google. One of the reasons that
Google is happy to -- they're not just saying "use our APIs and
have that because we're good guys." They built a better outbound
monetization in their sense than Yahoo has -- so what in fact what
Google has done is expanded the wall garden to be the size of the
internet, whereas Yahoo's wall garden because they (inaudible).
MR.
O'REILLY: I just want to pick up on something on that Yochai said
that triggered a thought for me about reputation in the context of
open source and I wonder if it's direction for some further research.
It seems to me that in open source communities, you actually see two
kinds of reputation. One is somebody like Linus Torvalds or Larry
Wall creates a system and effectively an architecture by which people
can add to their work, and people get secondary reputation by
extending what they've done. There are other cases, though, where it
is actually rivalrous reputation. So Perl versus Python versus Ruby.
And I'm just sort of wondering, it would be kind of interesting to
understand the economics and the reputation -- not just of the
monetary economics, but the reputation economics of rivalrous systems
that are using the same, you know, all the same principles, but, you
know, versus ones that are building purely on what was done before.
And again, I don't have any answers there.
MR. BURNHAM: Let's
get Martin in here.
MR. NISENHOLTZ: I guess I have a question
which has to do with what I sort of call the lottery syndrome. There
was a Powerball lottery yesterday. Tons of people entered it. We
know that someone won in Oregon. I don't know who quite it was yet,
but we also know that the chances of winning were one in 164 million,
and we know that because we can certainly measure that. I guess what
I'm struggling with is how we measure the number of peer production
efforts that get started versus Wikipedia, which has become the
poster child, the lottery, the one in 164 million actually works.
Now it may not be one in 164 million. It may be one in 10. It may
be one in 50, but I think that groups of people like this -- and I
count myself among you, so guilty -- tend to create the lottery
winner and hold the lottery winner up as the norm.
MR. O'REILLY:
There's some data for that.
MR. NISENHOLTZ: That's what I'm
asking for.
MR. O'REILLY: Look at Source Forge, there's
something like 104,000 projects on Source Forge. You can actually do
a long tail distribution and figure out how many of them -- but, just
to my knowledge, I would guess that one in like, you know, whatever
154 million there are probably out of those 100,000 projects, there
are probably, you know, at least 5,000 who have made significant
reputation gains as a result of their work. Maybe more. But, again,
somebody should go out and measure that.
MR. NISENHOLTZ: I'm not
suggesting -- what is success?
MR. JARVIS: That's what's changing
now. The definition of success, big enough, is different now. I
don't need to -- I can publish a book however I want. I don't need
to be big anymore to be successful. And so that's what has
essentially changed the economics the way Peter recognized. We're no
longer a blockbuster economy. We can be. But if you want to go in
and earn a living now, by whatever value you get, plus reputation,
plus good heartedness, plus return on my (inaudible), you can now do
that in a far smaller scale. However, Marc, when you add all those
little things up, I would wager that they're bigger than the old big
blockbuster and more sustainable in the long run.
MR. PINCUS:
And, Martin, you are in the position you don't have to play the
Lotto, because this is really a gift economy, and the more that you
give, the more you're going to get back, and that might sound very
California new age, but.
MR. NISENHOLTZ: Fuck you.
MR.
PINCUS: I did have an experience with a friend I brought and I tried
to explain that to her and she just didn't get that, gift economy.
But you don't have to wonder whether you'll throw a party and they'll
show up. Somebody starting with nothing says, "Look, folks,
here's Wikipedia," but you could, the New York Times or AOL,
someone who has something of value, could put it out there and give
it -- you could give something whether it's like Google with Maps,
whether it's some database that you've been closed and hoarding
because it's been a value to limit access, you could give it out to
the world. You can give people the ability to publish something in
your newspaper. There's so many pieces of value you can give out.
MR. H. MORGAN: How about AOL IM. IM is the classic case where
we had so many different wall gardens, and, Brad, I have the scars
that can prove that. That could have been much better for everyone
if it was really open up.
MR. PINCUS: Including AOL.
MR.
BURNHAM: We're going to have to break in a couple minutes. What I'd
like to do, Scott's got something important to say, and then I would
like to hear from anybody who hasn't yet been heard from, and if you
haven't spoken yet in this hour long or 50 minute session, we'll give
you a chance to speak and then we'll go break. Scott.
MR.
HEIFERMAN: Isn't it interesting that in the Powerball lottery, the
more people that play, the better the prize, but the more people that
play, the less likely for you to win. Whereas network effects, you
know, network effects of Napster and Del.icio.us or Ebay or Meet Up,
it's the more people that play, the better it is for everyone.
MR.
BURNHAM: It's a difference between a rival good and non-rival good in
a way.
MR. WILSON: So, Matt, you had your hand up. You got
something to add?
MR. BLUMBERG: Just a couple points about
reputation. One is I think reputation is absolutely critical to this
environment, and it's critical on two different levels. One is about
the reputation of whoever is generating the content, which may or may
not matter for things like Del.icio.us or for clicking on something
in Google, but it clearly matters as you go up the food chain as the
investment of consuming the content gets greater. So I'm more likely
to read Tom's book because I know Tom has a good reputation because
I'm actually going to spend five minutes a day (inaudible). So
reputation is important for that, but it's also important to start
developing some standards around reputation of greatness, because
that meta data is critical, at least going to be so much reputation
data, there ought to be a way of sifting through that, and I think to
Mary's point, what's interesting to think about is hey, if my
reputation on Ebay is kind of portable, but who gives. I think what
you'll start to see emerge is this aggregation of reputation data.
So to look at an offline analog, look at your credit score, and you
may hate how your credit score is compiled because you have nothing
to do with it, and it's not transparent, but there will be things
like that that emerge in this economy as well where my Ebay
reputation is one thing, but my reputation across 15 different
services says a lot about me.
MR. BURNHAM: But there are a lot
of systems that don't depend on reputation, like the music system
that Fred was talking about, and there are probably a number of ways
that you can navigate through this without knowing anything about --
MR. JARVIS: Problem is the reputation.
MR. BURNHAM: Right.
There's a number of other solutions. Dick.
MR. COSTOLO: I was
just going to say that it's interesting that we like to assume that
there must be some selfish motivation for participation, but
something that Tom said is interesting, right, that there does seem
to be anthropological propensity to cooperate or participate without
anything in return. We get comments on our feedback forums where
people post very detailed replies to technical questions as
anonymous. No one's linking back to them. They don't -- why
wouldn't you at least log in and say who you are so someone could
post a follow-up question. Nothing accrues to you in return, other
than you cooperating. So I think there's an interesting distinction
to be drawn between creating companies that try to take advantage of
reputation or build on reputation and just building companies or
things or peer produced goods that just assume that people will
cooperate for the purpose of cooperating.
MR. BURNHAM: Dave.
MR. D. MORGAN: I think one of the disconnects that we're talking
a lot about the wall garden (inaudible), but I think that there's a
big difference between the whole philosophy of peer production, and,
you know, low risk potentially higher reward, but low risk low reward
is fine. You can put out lots of things and if they work, that's
great, and if they don't, I don't know the answer to the 164 million
to one, but you can do that. But for people to have to invest cash,
particularly if it's corporate cash, the $1 that's spent badly can
sometimes overwhelm the 99 that's spent well. Particularly in the
media world, it's the one ad that shows up on the Yahoo of the
pedophile site that potentially jeopardizes a $40,000,000 Pepsi
relationship with Yahoo. And so that leads to a lot of protection,
because there's a difference between throwing somebody else's money
you're the agent for as opposed to throwing your ideas, whether they
be anonymous or not.
MR. BURNHAM: Peter.
MR. SEMMELACK: One
of the things that fascinates me is how many people here have a
Blackberry or ever used one? There's a lot. And if you think about
that, that's a completely closed system. End to end, they make
hardware. They make the software. They make everything, but the
value of that is pretty good. People pay a lot of money for that
sort of thing. And so where the two worlds meet is actually where --
I think we're going to talk about this a little later -- is where
pure reputation based systems that are altruistic versus the profit
motive where we got to please our investors. We got to make money at
this. Where these two things come together I think is going to be a
very interesting discussion, because Apple, again, iPod, totally
closed system, but rules the world.
MR. BURNHAM: We are going to
talk about that in the next session or earlier this afternoon. We're
going to talk about it both from the perspective of what does it mean
for young companies getting started and what does it mean for the
companies that Dave is talking about that are kind of playing defense
in this world trying to figure out how to handle that. There's some
other people that haven't had a chance to say anything. Josh.
MR.
SCHACTER: Just two comments. We're talking about reputation systems,
but things like Ebay are not just a reputation, it's also an
authority system. So that there's a transaction log and you can look
at it. So I sort of wonder, what does it mean to take the reputation
and move it somewhere else? Ebay continues to provide the authority
or is it something free floating and what does that mean? How does
that even work? Here's a file that says I'm really great. You should
believe it. I'm not really sure.
MR. BURNHAM: Anybody else want
to jump in before we break? Let's break. We're going to talk about
the idea of an open data architecture, which is really picking up on
what Marc was saying about standards and the gift economy and what it
means to be open or closed with your data architecture. 15 minutes
we'll be right back. (A break was taken.)
MR. WILSON: Since
we're talking about peer production here, I don't know if many of you
use the peer to peer files sharing systems, but there's a concept in
a lot of those systems of contributors and leechers, and leechers are
people who take content out of those systems, but never contribute.
So we don't want anyone leaving here today being a leecher.
MR.
BURNHAM: So the risk is that you will actually be called on if you're
not raising your hand.
MR. WILSON: There's a few people, this
guy over here in particular has not said anything. Anyway, Brad is
going to introduce the next topic.
MR. BURNHAM: So what we want
to talk about now is the idea of an open data architecture and what
that means. And so over the years, we've become familiar with what
an open hardware architecture is, that's the PC, and what an open
software architecture is, Linux, an open network architecture, the
internet, and we are beginning to try to understand what an open data
architecture is, and there have obviously been examples of Wikipedia.
You can go in and pull the stuff out. Del.icio.us, it's fairly easy
to pull the stuff out. Interestingly, Craigslist, in the last couple
days, has made some -- they've been very open to date, maybe they're
not going to be open, but what I'd like to do is ask Tim to just
introduce the idea of whether or not that concept even makes sense.
Is there such a thing as an open data architecture, and if so, what
does it look like?
MR. O'REILLY: Well, Brad just hit me up to
offer some thoughts about this three minutes ago, so I don't know
that I have anything terribly profound to offer, other than on a very
superficial level, we've had a variety of open data architectures and
they've been intrinsic to the success of the other open architectures
we've been talking about. For example, in the original design of
Linux, the idea of pipes and filters was based on the idea
(inaudible). And programs cooperating through an open data format.
It was simple. I think in the era of the web, obviously HTML open
data format, and it drove a lot of the success of the web, but as
we've gotten toward more complex systems, it's been pretty hard to
figure out. We typically have had attempts to standardize way too
much. What we do see is that the data architectures that are simpler
tend to win out, and so we've seen this played out -- starting to see
this played out in web services, but clearly in a lot of the systems
that we think about are heavily data driven. And data in the
aggregate is the source of value such that I've tended to use the
phrase data is the next Intel Inside. And clearly there are
companies whose lock on natural databases, it gives them enormous
power, and so I've posited that we will -- there'll be a Richard
Stallman among us who will start the Free Data Foundation somewhere
about now, because they're going to recognize that we have a problem
that somebody owns data, they're not -- again, we heard Bob talking
about it already today -- somebody owns some critical piece of data,
and it's not portable, it's not modifiable, it's not exportable, and
we're going to need to solve that problem. I don't actually know
when or how we'll come to a crisis. I do believe that there will be
some sort of signal events where somebody basically takes a database,
private -- we saw it in small scale with CDDB -- where they'll become
peer produced data source that some company will say, "You know,
we created this wonderful environment. You built this for us. Thank
you very much."
MR. BURNHAM: Michael.
MR. PAREKH: The
one point I would ask from a definitional point of view on open data
architecture is if we were doing this 10 years ago, it would be fine
if everything for the most part was in English. One of the questions
I have is look at the next 10 years where over 60 percent of the
content out there is going to be not in English, is what open data
mechanisms are being created from a translation point of view so that
you can take stuff from China or European countries, etcetera and
make it malleable, fungible instantly. I know there are some
translation efforts going on with Google or Yahoo, but is there an
opportunity here to introduce language, translation elements from an
open data piece of it that allows more people across countries and
cultures to get in on the conversation.
MR. BURNHAM: So there's
a couple layers here. We're talking about sort of the gist of
philosophical practice of whether your data is accessible or not, and
that's irrespective of the standard of the data. So, for instance,
do I just allow my site to be crawled. Do I allow the site to pull
data out. Then there's the question of are there efficient ways to
define interchange of data so you can avoid that crawl and make it
more accessible, and then finally there's the question of within that
data, there's the problem of languages and formats and things like
that. I'd like to try to stay for a minute with that first problem
and just say who owns this? You know, who do we think should own it,
and particularly of interest in a peer produced system, where the
peers at least feel as if they're contributing this to something,
like CDDB.
MR. O'REILLY: Or Del.icio.us, to bring it close to
home.
MR. BURNHAM: Right. And then how do you -- so what should
your philosophy be, and maybe we should just let Joshua tell us what
the philosophy should be, and then I want to get to Mary.
MR.
SCHACTER: Well, in Del.icio.us, in my belief, is that the users own
the data, and this actually brings a couple interesting and technical
challenges to the field. Basically the user can say, you know, they
put their stuff in. They can take their data and go home. They can
remove themselves. They can delete themselves from the system. How
many systems actually let you delete your account typically? That's
very rare I think. You know, when the user says "delete this
item," we really delete the item. We don't mark it as deleted.
We actually purge it. So I believe that that data's theirs, and that
we have the right to use some of it in some way, but not, you know,
personally, identifiably that kind of stuff. This is a tough line to
walk in general.
MR. BURNHAM: So if the data is theirs, then do
you believe that they have to choose whether or not that data is
accessible to a crawler as an example?
MR. SCHACTER: It's
interesting. Right now we don't let any crawlers in. This is mostly
for, you know, load reasons, because there is a lot of pages in
Del.icio.us because of all the tags and tag combinations.
MR.
JARVIS: Should you?
MR. SCHACTER: Should I? Well, it's
interesting, the question is -- the question is, for example, in this
particular case, are the crawlers going to get value they're not
going to pass back to us, right? If they certainly keep asking --
MR. O'REILLY: So you've made a decision, though, that therefore
you own this aggregate data in some way with respect to other
aggregate data.
MR. SCHACTER: Yes. The user owns their data and
we own the aggregate.
MR. JARVIS: But the value of the whole
community together is greater, the greater the community, right.
MR. SCHACTER: Indeed.
MR. JARVIS: So that if the wisdom of
the (inaudible).
MR. SCHACTER: Maybe. I'm talking about the
crawlers, like, for example, just the search units.
MR. JARVIS:
Right.
MR. SCHACTER: I'm not quite sure what that looks like.
MR. JARVIS: It's not that the individual user owns data, the
community owns the data combined. So in that morale thing, what
happens?
MR. SCHACTER: Actually, I sort of refute the notion of
the community. There are many communities. So that's an interesting
question. I mean, a lot of it is dominated by technical. I don't
have much of a choice right now, and then there are business
concerns. For example, Google keeps asking for the tech data, and
should I just hand that over to help them or --
MR. O'REILLY: No.
You should make a deal with them.
MR. WILSON: Let's go to Mary,
because I think that Mary's involved with something called attention
trust, and you know, I think the model behind attention trust is that
the user owns their data.
MR. COSTOLO: The user owns a copy of
their data.
MS. HODDER: There's
more to it that than. Dick is actually also involved and Seth
Goldstein, who's not here. Seth and I think basically -- and Dick as
well -- I don't know how early you were involved -- but so basically
the idea is that users own their data, right, and the trust puts that
forward and says that companies own the aggregate data, and a copy of
the user data, and then the user owns a copy of their data as well.
And just to sort of reference what Jeff and Josh were talking about
in terms of what skew, what's of value. Is it the entire aggregate
of the data or is it my data at Del.icio.us, plus my data at Amazon,
plus my data somewhere else, and that can I make a trade for that
somewhere, and if we set up an economic system where that information
has value on the other side, does it then sort of change the
marketplace so that users have some power to negotiate that
information away from different companies because those companies
won't remain competitive, and if their competitors are giving users
access to their data -- I think that, I mean, one of the things that
I see potentially coming out of a system like that, and I think the
attention trust, which is the trust side of this, there's sort of the
nonprofit that advocates for the user using the data, has to be
really careful about is companies could then use that, you know,
knowledge that they can get that information from users against them,
right. So when my Mastercard company sends me the annual change in
the terms, they say "if you don't like them, you can discontinue
your service with us in 30 days," but I like all of the
privileges that I get from having a credit card, so, of course, I do
not refuse the service. So if Mastercard and Visa decided to say
part of our news of service this year is that you have to give us all
of your online click stream data, we want everything you collect out
of our attention recorder, am I sort of helping the bureau in having
to give that information away? So I think that there's, you know,
there's the positive of that they get, and then there's the negative
of can this be used against me, and can the attention trust kind of
be advocate for users, but also advocate for users being to keep that
data private and having users truly stay in control of that
information.
MR. BURNHAM: Mike.
MR. FRUMIN: Just along
those lines, the concept of, you know, different people owning
whatever your data is and people abusing it, like the Mastercard
example, I think that if you think about the fact that Mastercard or
American Express or whoever, a lot of people, their business is
having your data, and so if you found ways to make your information
available to you and selectively to other people, a lot of that
opportunity goes away. I mean, you wouldn't get spam about
refinancing your house if there was a way for you to tell everyone
who was interested in offering you a mortgage that you wanted a
mortgage, but there's not really. So in terms of owning your data
and abuse, the more you are in control, the more you share, frankly,
the less room for abuse there is.
MR. BURNHAM: Peter.
MR.
SEMMELACK: I think data ownership is an entirely different question
than this.
MR. BURNHAM: Data architecture.
MR. SEMMELACK:
That's full of all kinds of privacy issues, but open data
architectures, for me, I feel as if like a public library is a great
example, and because who owns the public library? The public library
is owned by the public, and if you look at the internet, who owns
that? Nobody. So isn't -- would it be true to say that the best
owner of a standard is a body that is not driven by profits? ASCII is
a good example. You have ASCII versus EBCDIC. ASCII obviously won
that battle. You have standards bodies who are not the holder, you
can argue that, but there's no profit motive per se that is driving
-- you know, there's two kinds of standards, there's standards by
bodies like that and there's de facto standards like Microsoft. I
would argue that the standards bodies are the best owner of something
such as this, because you avoid a lot of the strange behaviors that
come out of the chase for dollars.
MR. BURNHAM: Tom.
MR.
EVSLIN: That's the first thing I've heard here today that I totally
disagree with. I think standards bodies is wrong. I've been a part
of them. I've been on the outside of them. I think that they do a
very good job of slowing down progress, for good and for bad reasons.
I think that people who spend their time on standards bodies are
being paid by somebody to do that, and that somebody has interest,
and I'm not putting down those interests, but they're real. I think
that if we're going to have a data standard that's as fast and
flexible as we need it to be, and if we give up the idea of 100
percent solution, because if we wait for 100 percent solution, then
we'll never have the 90 percent solution, that it almost has to be
Wiki in the sense of both the mechanism and of it being fast. If
there's a Wiki with perhaps some kind of voting, perhaps not
mechanism for defining tags for various kinds of data and formats in
that, from a technical point of view, we can afford to tag each piece
of data with what it is because storage is so cheap. You used to be
able to do that. So I haven't thought about this much, but I think
we need something that as flexible as the people who participate in
it and as fast as participants are willing to make it and open to
all.
MR. BURNHAM: Let's get Chris and then go to Marc.
MR.
ANDERSON: Talking about de facto standards versus de juris standards,
de facto standards are somebody does something, grass roots. It
bubbles up. The marketplace votes on it and it wins. De juris
standards is what standards bodies do, and you think of a few
successes, but a zillion failures. So my sense is there is no
process. There is no body that can establish standards. What you
have is a zillion people who are trying things and good ones quickly
percolate to the top.
MR. BURNHAM: I'd like to actually see if
we can move back up away from sort of the technical process of
establishing a standard for the interchange of -- the technical
interchange of data and get back to the question which isn't an
ownership question, a data ownership question, but what are the
issues around allowing yourself to be crawled? I mean, you've got
Craigslist, you've got Google Maps, you've got other people who are
just pushing the data out there. Does Google Maps have an open data
standard or an open set of principles? Is Craigslist changing? Can we
talk about that for a minute? Marc was next and then we'll try to get
everybody else.
MR. PINCUS: Which is my point, and back to Tim's
point I think was spot on for this conversation, I think that all our
-- we don't matter in this. Whatever it is we figured out won't
matter, and it's really just going to be a continuation of trial and
error, and so Craigslist tried being open and crawled, and now
they're saying Oodle and whoever, "you can't crawl us," and
all of these job databases and other classified databases have so far
chosen to not work together and share data because it's competitive.
And I think what we're going to find is that people are just going to
keep trying different models and out of greed and business interests,
the open -- I think the open database standard will eventually win,
we just don't know how. And so Hotjobs has said they're going to be
kind of open or I think what they mean is they'll be one way open.
They'll crawl other people and present it and not allow others to
crawl. But I think that we'll see people and see some of the more
enlightened or -- that's a biased way of putting it -- we'll see some
of the big companies and big databases experiment earlier and some
later. AOL, I think it's been painfully obvious for five, six, seven
years that they're just slowly dying by being closed, and now they're
making open AOL or something. So I think it would be interesting to
watch, and I think that it'll be what's in the best interest of the
end-user to then how is the end-user best served, and Craigslist
right now is a good experimental case, because right now it doesn't
look like it's really important to end-users for Craigslist to stop
Oodle or others from crawling them, but maybe it will, because maybe
other people in the market will open up.
MR. JARVIS: My
contention is that Craigslist and Monster and Hotjobs are just way
stations. They are cheaper marketplaces than newspapers were. So
one version of control is a cheaper version of control. The cheaper
version of control has destroyed millions of dollars of revenue
money. That's not the end of the game. Somebody's got to come --
MR. BURNHAM: I got to interrupt you just because I don't like
that phrasing, because I've heard that applied to Craig a lot by the
newspaper industry, that Craig has destroyed value.
MR. JARVIS:
Fair. I love Craig.
MR. KURTNIT: What's the term you used?
MS. FISHER: I call it
creative destruction.
MR. BURNHAM: Well, if you think about
value as being the most efficient delivery of a service, delivering
the service more efficiently doesn't destroy value, it destroys
pricing.
MR. WILSON: It's wealth transfer, not wealth
destruction.
MR. JARVIS: The real point is, we're still thinking
centralized. We're still thinking one dot world. So in a
distributed world, what can happen? Somebody comes along and can do a
micro format to say, "Here's how you put your resume up and
here's how you put it on the Wiki. And you know what, you can use
Google to search that." And we'll put together a buyer and
seller of jobs with no friction and no value at all, except for the
people who kept their money in their pockets because they didn't buy
the ads anymore on Craig or on newspaper. So if you start looking at
it that way, is that Craigs definition -- I'll still say destroy,
because Craig is the coal mine here who says "Watch out."
Someone can come along. Take the newspaper industry and say, "Bing."
Guess what? Well, somebody can now do that to Craig too. So Craig is
now -- there's a path that says "I'm open. I'm closed."
Somebody else can come along and be more open, and if the user, the
job seekers and sellers benefit more, that will win out. That will
win, if it can. Some day real estate brokers will get destroyed.
Someday that will happen.
MR. BURNHAM: Let me get to Joshua here
because he hasn't been too vocal, and he usually is.
MR.
STYLMAN: Thank you for that. I was afraid of being called on. It
seems like this is a conversation about control and leverage. So
initially I guess you didn't like the phrasing of that, and a prime
example is look at Blockbuster and now look at Netflix, and obviously
Netflix has all the control and all the leverage until the next thing
comes along. And what I think will be really interesting is when you
got someone like Indeed, and Indeed gets really, really big. Well,
at that stage, do they get the control and leverage to say, "You
know what, Monster, we're not even letting you into our audience,"
and ultimately, is that the next power shift?
MR. BURNHAM:
There's a question about -- to come back to something Yochai said
very early -- is that there may be a sustainable -- if you think
about all of these, they're, in a sense, a form of peer production.
Craig is able to deliver a service in 190 cities in 35 countries, you
know, with about 18 employees, not because -- because he's figured
out a way to use the technology really efficiently and get everybody
to contribute their own content. So it is a form of peer production.
So the question is, is there some sustainable evolution of this that
maybe -- I would argue that Jeff isn't correct entirely, because you
still have the problem of finding things. So Google has a relevancy
ranking algorithm that will bubble up the appropriate jobs. Somebody
is going to create some probably peer produced indexing relevancy
ranking and there needs to be some centralization of that in order to
be able to do that.
MR. JARVIS: Google Jobs, at some point, that
won't work, you can't get all the jobs. So you figure my job is
stuck in Craig now and I don't want it in Craig. I won't put it
there.
MR. BURNHAM: I think there is a pressure.
MR. H.
MORGAN: Whether or not there's physical good or whether it's just
information. At Heartsearch, we're happy to have our database
crawled deeply. We have -- Amazon has it available, because it can
eventually lead to a transaction of a physical good, which is money.
But if the transaction is purely informational, that's when it gets
driven down to how do we get that information out there, and it's
cheapest if the person who's providing and wants it out there can get
it out there for free, whether it's individually peer produced or
aggregated in other ways.
MR. WILSON: Scott hasn't spoken on
this topic yet.
MR. HEIFERMAN: I would just add that there's,
you know, I'm a huge fan of disruptive, distributed bottom up stuff,
but, you know, there really is something about the potential
combustion that can happen when these things are sort of cultivated
by -- I'm sorry, I'm not speaking well. If Web 2.0 was in the year
2000 and not 2005 or 2006, I would argue that there wouldn't be such
a great, wonderful, efficient market called Craigslist today, because
you need that place where all the aggregation was happening. That if
Del.icio.us was just sort of a completely distributed notion without
this sort of a level of protective centralization there, then that
little explosion of wonderful stuff wouldn't happen. And just, you
know, lastly, I think that there is still the reality, which is 90x
percent of the world has no idea of the benefits of Craigslist or
Del.icio.us or any of this stuff, and it needs these really well
cultivated cultivators to make them available. That, you know, my
ability to find something on Craigslist would be dissipated if it was
so distributed for some period.
MR. BURNHAM: Tim.
MR.
O'REILLY: Couple points. First of all, your question originally was
about an open data architecture, and so one of the -- which is not
the same as an open data format. First of all, it seems to me there
are many, many formats for many, many specialized classes of data,
and the question that we have to ask ourselves is what are
characteristics of data formats that make them susceptible to being
used in an open data architecture. And I don't think any of us
really know the answer. I mean, this really is the frontier of the
industry right now. But I do think that there are a couple of
lessons, and one is an open data architecture is one that is
potentially syndicatable, ie. can appear in many places at once.
And you know, if you think about the fundamental nature of the
hyperlink, you link to many, many -- from many sites. You don't --
nobody's keeping count. I think one of the things that Katarina
talks about in Flickr is that sort of she hates sort of two-way
links. And that was originally hypertext theory was we need two-way
links, and they thought that was appropriate. And Katarina's comment
about social network, social networking is broken because it requires
two-way links. She said basically just say you set a watch list. So
this idea is it one way or is it two way? Actually, one way may be
one of the characteristics of an open data architecture, ie. you
don't need acknowledgment. There may be some other things like that.
So I guess what we need to figure out is looking at the
characteristics of systems that seem to have network effects around
their data. They're syndicated. They're one way. They're outbound.
And, you know, so, again, Google Maps versus Yahoo Maps?
MR.
BURNHAM: I wanted to jump on that one-way aspect, because, for
instance, one of the things that if you syndicate a piece of data,
you know, do you want to have a string tied to that, at least be able
to capture the meta data about how it's being used, and does that
make it then two way, because if you -- is it the same to syndicate
and just let it go and never have no feedback at all about where it's
being used, how it's being used, or is the ultimate architecture
going to have something to do with --
MR. O'REILLY: I'm making a
guess here, but I'm making a guess that if you try, actually quote
Blake, catch the joy as it flies, the wing or something or other
destroys. Anyway, basically, you have to let -- I think you have to
let whatever you do go. If you try to kind of say, "Oh, I have
the instrument so I captured it," you're not going to get it.
MR. BURNHAM: That's a pretty significant departure from just
saying, "Okay, I'm not going to let it go at all. You can't
crawl me. I'm not going to syndicate my data. It's in a box,"
to "I'm going to syndicate it, but I'd like to be able to at
least capture the value that is associated with how people use it,"
to saying "I'm just going to let it go and I have no idea what's
going to happen."
MR. O'REILLY: I'm going to make a leap
here, and I don't know if it's actually true, if I look at the web
and what happened, it wasn't until the web got to a certain scale
that it was possible to discover how to recapture the aggregation.
So I think that something happens when you syndicate outwards. When
you let things go, you create network effects. If you try to
optimize prematurely so that you capture the value and you build the
aggregate system too early, you limit the network growth. And that's
why often the actual value is captured by a second party, right,
because you cannot build it into the system at the outset. You have
to figure out how to, you know, how it's actually implicit somewhere
in the system. Somebody discovers later where that implicit way --
MR. ANDERSON: Google was formed as an experiment to try to -- it
was linking. It was called back rub, and it was the very nature --
in fact, the one-way links created an opportunity for a third party
to out of band answer the questions "Who's linking back?"
This was the experiment. In doing that, it then created a search
engine which was a completely unexpected outcome, which added value
in a whole new way, simply because they hadn't predetermined that
back turn.
MR. O'REILLY: So the value actually needs to be out
of band for the open data system I think is maybe the principle that
we were trying to get at here.
MR. WILSON: Let's hear from
Bruce, because we haven't heard from Bruce yet. Then we'll go to
Josh.
MR. SPECTOR: Thanks. I think that a lot of what we've
been talking about, we're really kind of talking around here, and
part of my question here goes to I think the afternoon's session, but
we're talking really around it now, and that is, what are the, you
know, what are the models, what are the economic monetization models
that are driving all of the conversation that we're doing right now.
And we really haven't been talking much about it. It seems to me
that the notion of open data and commoditized data is a done deal. I
mean, the web has essentially established the ability to commoditize
lots and lots of data in a variety of ways, and many of us sitting at
this table have been, you know, struggling over how to earn value,
whether it's social value or whether it's economic value, out of the
proposition that exists. We're in a commoditized data universe right
now, and Michael Bloomberg made millions of dollars by selling
proprietary data.
MR. O'REILLY: By selling commoditized data.
MR. SPECTOR: Right. But he found his way, but it was his
proprietary version of that commoditized data. So we're now talking
about a world in which all the data may be free and open, and I
believe that, in fact, it will continue to develop along those lines,
then what happens then? Literally, what happens then? If the data is
commoditized, whether it's through Del.icio.us, whether it's through
web crawling, whether it's through an open source database, then
what? I think that's really the question I'd like to --
MR.
BURNHAM: Well, let's --
MR. SPECTOR: -- put on the table.
MR.
BURNHAM: Let's save the question about the economic models that
underlay that, but I think that one of the interesting points that
Tim is making is that, you know, it may be, you know, that somebody
creates -- sets the table for others. You know, they're going to
create an environment in which the data is going to get out there and
then others are going to figure out how to do the re-aggregation
after the fact. After the models begin to emerge. So maybe we
almost can't answer that, but at least we can talk about this
afternoon, how one might approach tentatively a couple of those
opportunities. Joshua, you want to say something and then we'll come
back. Ah, Seth. Let's do Joshua and then Seth.
MR. SCHACTER:
It's interesting in that we were previously talking about peer
production, which is largely a system where we're able to diffuse the
value of the system such that more people benefit, right.
Aggregates, aggregators, which we're doing, which is the opposite of
a syndication, right? If someone is syndicating someone, someone else
is going to be aggregating those things. So I wanted to call
bullshit on Jeff Jarvis before who said that when everyone goes off
and its micro formats and, you know, everyone transacts for free, but
at some point those things have to be found, right. We're stumbling
towards markets of information, right, and you say market, you think,
you know, is that the (inaudible) or whatever, but that's really the
place where everyone gets together, the market square or whatever,
right. And the people have to come together to transact. There's no
-- the fact that it's distributed and makes a peer produced system or
whatever adds a lot of value on the whole, but also, the fact that
everything comes together and the liquidity is increased, because
we're not talking content in general, right. We were talking about
open data where a piece of data is just sort of given away and
someone else gets value from that, but these are job listings, right,
which is not actually the entire piece of data, right, and that's not
the value of the thing. If the pointer's the value of the thing,
which is the job, at some point, people have to go somewhere to
transact, right. So it's not an "either or" thing. There
is tension in the system such that people want to be open, but they
also want to gather value from it. So there's the spreading and then
the collapsing. It has to find some happy medium, right, and that
these -- both of these properties are going to be places that people
are going to be able to compete on. Being more open is going to be a
place to take a point of competition. Different sorts of ways of
aggregating and listing data are all going to be ways that people
figure out -- so I think it's going to be not just one person
generating and one person aggregating, and the question is do we fall
into some system. I think it's going to be continually going through
cycles and moving around.
MR. JARVIS: I agree, but I think
that's where you get down to you or Jimbo Wales or somebody who is --
what Scott was saying earlier -- the benefactor of that or Craig, as
long as they are the benefactor. So it's about trust.
MR.
SCHACTER: I agree.
MR. JARVIS: It's like you trust that.
MR.
SCHACTER: Right. At some point there's a lot of value. I think that
an incredibly distributed system with no central point of authority
are very elegant from a bunch of conceptual reasons, but centralized
systems also add different values, and that at some point, what we're
assembling toward is that the ownership and publication of the data
is distributed, but the central point of aggregation of finding this
attention router, you type in what you want and it goes to where you
want. It gives you something that was at the top of -- some way of
finding these things, because we're now transacting in millions of
different commodities instead of hundreds or dozens. So the value
shifts. So the question is --
MR. O'REILLY: What is the access
of aggregation?
MS. GREEN: But you could crawl, you know,
Del.icio.us, and we don't want that to happen, because that's why
Del.icio.us was created, because Google's flat.
MR. PINCUS:
We're all kind of real estate developers. We're living in a world
where all the land will be free, and there will be some new kind of
land created and maybe whole new building levels that you can build
on top of the land, and Del.icio.us might become another kind of land
created that buildings can be built on top of, and then the buildings
that are going to have value -- the land will have more value because
better buildings are built on it, and one way or other Google came
and took free land and then stuck a building on it that everybody
liked, and now, all of a sudden, through people -- so many people
going through their building, there can be new buildings built on top
of Google.
MR. BURNHAM: Let's get Seth in here, if we can.
MS. GREEN: Actually, where we're going is very interesting, because we would have a lot of stuff being created by people and (inaudible).
MS. HODDER: So,
actually, to tie in what Tim said at the very beginning of this
session about open source and making free data, making a free data
foundation, I think it's a really interesting idea, because then the
question is what's the GPL for free data, right, because the issue is
it's not to me. It's not -- I don't see it as --
MR. BURNHAM:
What's GPL? Give me a public license?
MS. HODDER: Right. So
the idea is, you know, when I use the GPL for stuff that I'm working
on, I create --
MR. HEIFERMAN: Define GPL.
MR. O'REILLY: The
specifics of that license is that it says if I give this to you, you
can only -- basically you can only incorporate into your products if
you give it away under the same terms under which you've received it.
MR. PINCUS: Give away under the GPL for data.
MR. O'REILLY:
The Berkeley style license, which is I give it to you. You can do
whatever you want with it, as long as you acknowledge that I gave it
to you.
MR. PINCUS: Why would I let Oodle crawl Craigslist when
Oodle won't let anyone crawl Oodle?
MS. HODDER: But there's
a couple of properties and data and it's covered under copyright and
not under software licensing terms in terms of the GPL. But the
point that I was going to make, I mean, there is a way to do that.
Craigslist could make a deal where they would say "Well, you
know, our data is available for free noncommercial use. But when you
go to use it commercially, you have to make a deal with us."
MR. PINCUS: Or you have to make it also public to other people.
MS. HODDER: But that's
part of the deal that you make. So the thing that I'm suggesting is
that that actually could work here in terms of "here" being
the US, because we have a particular structure of laws and what not,
under copyright, or some kind of thing where you say for
noncommercial use, that people can take things with APIs and they can
play around with them. But at the point where you're ready to do a
commercial service, you have to come and make a deal. So to me, the
difference occurs not when Oodle wants to take Craigslist data right
not here in the US, but it's will other places observe, you know,
those kinds of I don't know.
MR. BURNHAM: What happens in China
is the question.
MS. HODDER: Yes.
Exactly.
MR. BURNHAM: Another interesting question is could you
actually, in some way, enforce it in an API? Unfortunately, it heads
you sort of towards some kind of weird DRM system, that is the GPL
DRM system, which says here is the data. As long as it's
noncommercial, but maybe you can't. Or maybe there is some way to do
it within the API. Seth, you've been trying to get in here.
MR.
GOLDSTEIN: I want to try to connect some of the data architecture
conversation with what we've been trying to do with attention trust,
and I think what I learned from Josh about a year ago, what is
Del.icio.us? Now it's going to be or it is certain mass market
Del.icio.us is social bookmarks or whatever it might be, but I
remember asking Josh, and Del.icio.us is, according to Josh, and
correct me if I'm wrong, crystalized attention, and that was what he
said, and it sort of stuck with me throughout this past year as I
tried to think through all sort of Web 1, Web 2 services, and if you
think about that, consumers create content with their attention. So
where they pay attention to passively in terms of their click stream
or various proxies or actively in terms of tagging or searching for
things, they're creating content. Some of that content might be
explicit content that you see and some of it may be hidden in
databases, but it's still there. One of the things that has been
helpful for me related to this is China solved for influence, right.
Because influence is really kind of Seth's working on it with lenses.
So we defined influence as influence equals the amount of attention
you get, divided by the amount of attention you give. So I would
argue that -- so you can think about all the different ways that, you
know, a service gets attention from people and all the different ways
it might provide it out, Del.icio.us is extremely influential,
because people can add tags in so many different ways and it controls
the attention that goes out of it in terms of it doesn't have access.
Access is a huge I guess a deflationary force in the attention,
maybe because people that put add tags on their pages just basically
-- as much attention as they're getting, they're just sending it out
the other way, and Josh and Del.icio.us have been really good at
maximizing the influence of the Del.icio.us. So as you start to
think about -- I think a lot about what Dick is doing with the feed
burner around awareness and the whole web analytic community and the
behavioral targeting community is about trying to help publishers
understand their awareness, right. And help publishers keep track of
subscribers and try to help web sites figure out who's going to web
sites. And that's a pretty big industry. And it's getting bigger.
And I think to this point, there is no industry for consumers to
understand the value of their attention.
MR. BURNHAM: Let's
bring Dick in here. What you're basically saying is that when data
becomes flat, commoditized, freely available, then the value moves to
attention, and then ultimately to influence, which I think is a
really interesting model.
MR. GOLDSTEIN: It needs to get priced.
MR. BURNHAM: And that's different than -- and that may make the
whole discussion of an open data architecture irrelevant, because it
may mean that, you know, the real conversation should be at the next
level.
MR. COSTOLO: Well, what I was going to say is so I don't
think it makes it irrelevant. I think the thing that it implies for
me is that an open data architecture will be one that is purely an
API. I think you'll see architectures emerge that have no
destination site, they're just APIs into the attention data or
content out there, and enormous value of some sort will accrue to the
developers and creators of that architecture that have made this API
generally available as content flows through that API, and it will be
interesting to see what sorts of economics and organizations emerge
around that. But I think that what the implication, the bottom line
implication is that an open data architecture will be one that is
purely API based and not destination based.
MR. H. MORGAN:
Semantic Web.
MR. EVSLIN: We're using open in two senses,
obviously. One that the architecture's open and one the data's open
and aggregated. On the second point, because you were talking about
the first, I think we've seen this game before, and we've seen the
movie, and it's a cycle, and it'll probably go on forever. If you're
the first mover in anything that has a network effect and you create
a large enough network, which you partly do by being open for a
while, or more open than anyone else, you accrue a great value in
that network, and there's no value to you in letting somebody be your
N+1 competitor. If you say, "Well, you can crawl me,"
somebody who has only 10 percent of content you have, but it's
disjoint, now has 110 percent of the content that you have, and
you've been able to N+1. We're seeing that with Skype. Skype has no
reason to be open. Skype has no reason to be open. They have too
many subscribers to want to do that until they're trying to add value
to Ebay, but then what you have, once somebody gets into this
commanding position, and it happens over and over again, is there are
two ways to get out of that position, and it's almost never
voluntary, or it is never voluntary. One is that they may be the
dominant player, but they may only have minority of the market share,
and everybody from the second to last has a reason to want to be open
and to share, because it's the only way to break the position of the
dominant player. And so, if all of us little guys together have 60
percent and the other guy has 40 percent, we have every incentive to
be open. The only one who doesn't is the hoarder at the top. So
that's one they get broken. Sometimes they don't get broken. Take
AOL early on. We had screen names. We had e-mail addresses. Things
that were very sticky and had network effect that eventually it's
like trees don't grow to the sky, and the wall garden, even if it's
dominant in its area, cuts itself off from innovation, because it is
a wall garden, and innovation happens on the other platforms, and
then soon there's a service of greater value, which N+1 is value
rather than content.
MR. PINCUS: It could just die, Tom, because
of economics.
MR. EVSLIN: And it dies. But I think we'll see
that cycle forever and ever.
MR. WILSON: Let's get Seth Godin in
here. We haven't heard from him today.
MR. GODIN: I want to
first thank you. I thought that was very insightful. I wanted to
just jump in with three Fs, leaving out your F from earlier in the
day. The first F is about Friction. If we think about Malcolm
Gladwell's book The Tipping Point, how many people in America at this
table paid money to read the book? A big number. How many read it in
The New Yorker? A smaller number. How many read it for free on
gladwell.com? Almost none of us. And if you look at the value
accrued to Malcolm, he made no money if you read it on gladwell.com.
He made no incremental money if you read it in the New Yorker, and he
made a million dollars if you read it in book form. So friction is
interesting, because consumers like friction and creators of content
like friction, and I don't see any time soon that a desire for
friction is going to disappear from most people's -- not the Berkeley
community --
MR. BURNHAM: Why do consumers like friction?
MR.
GODIN: People are voting with their dollar. They'd rather read the
best selling book that everyone else is reading than seek out, figure
out a way to look at all the different choices, which leads to the
second one, which is Focus. The TV Guide, when it sold at its peak,
was worth more than the TV Guides that were selling at the same time,
because the directory feature, the directory function of pointing --
I mean, one of the things that open data really is is a billion TV
networks. A billion book publishing houses. A billion magazines.
And anyone who wants to write now can write. So the TV Guide
function becomes really important. And then the third is really a U,
is Unfair, not Fair, which is if people create stuff, want more than
their fair share. And the feedback that I'm getting more than any
other feedback with regard to data is people want more than their
fair share of traffic. More than their fair share of people looking
at what they're creating. So I think when we put those three things
together, just to piggyback on Tom, it's nothing that new. What's
happening here is that human nature isn't going to go away, and that
this desire to have friction, have focus, and get more than their
competition provides a platform for people who want to make a living
in the middle, because it's just -- I don't see how if it's
completely open and dynamic with no friction and no focus, that
people are going to be satisfied. The average person.
MR. D.
MORGAN: I would say, following that point, you can look at Walmart as
a great example. When it was the number 10 retailer in the country,
it had an open database. It participated in IRI. At the point it
started to accelerate a lot and became number two or number three
like six or seven years ago, they closed down the database, because
there's more predictability, and you can grow faster if you control
the data, because you can do that with engine if it's growing and its
pricing, and that's what's happening here. You play by the open
rules to get moving, but as the markets mature, competitive issues
become more important than exploiting just the general open
opportunity, and you use the data to control that growth better.
MR. BURNHAM: Let me just follow up on that, do you think that
cycles is -- and to Tom Evslin -- do you think that cycle's going to
happen faster today in this next --
MR. EVSLIN: Yes.
MR.
BURNHAM: Why would it happen faster?
MR. EVSLIN: Because
everything happens faster. There's less friction and more force. If
you asked me that about anything, I'd say yes.
MR. BURNHAM:
Let's go to Jeff and then Tim and come back to Peter.
MR.
JARVIS: Rephrase what you rephrased before. If you create content,
content is data, data is now free, so you can never own enough
content. Marc will shoot me across the table, but that's fine, the
value of content increases. What increases, the value of trust. So
that's what you're saying. When you say it's from data to attention
to authority, what you're talking about is trust. What you're
talking about in terms of Del.icio.us and what Heather wants is that
I choose to swim in the pond of Del.icio.us and I trust that pond and
I trust Josh to man that pond, but I also want the value of being in
the ocean once in a awhile. And so the content isn't what's
valuable. It's the trust and relationship that's valuable, and that
to me, in a post scarcity -- what the internet does is it takes away
scarcity in terms of both content and distribution, and it changes
the value essentially to trust. So the friction is still there. The
friction isn't "I own the content and you don't." The
friction is, "You're going to keep good content? Okay, then
let's talk." It's a different friction, a different value, but
that's essentially what we're going towards. So we forget that -- if
you're a magazine, your value is that we picked good stuff and you
trust us to do that, and the readers will say probably like you,
maybe they have value (inaudible). So relationships and trust
becomes a new structure.
MR. BURNHAM: Yes. Tim.
MR.
JARVIS: I also want to hear Umair on that, too.
MR. BURNHAM: Go
ahead, Umair. We haven't heard from you.
MR. HAQUE: Yes. I
think fundamentally what -- the way that I approach this is through
the lens of simple economics, and what's happening here is we think
of a --
MR. WILSON: Speak up.
MR. HAQUE: If we think of a
simple value chain with data in the middle and stuff on either side,
what's happening is that the middle of this value chain is exploding.
Data is commoditizing, as Bruce said. So what happens? Value shifts
to the edges essentially. This is not a new concept. Herbert Simon
said it in 1971, which is that "What does an abundance of
information create?" A scarcity of attention basically, right?
So when we talk about data, if we're thinking in e-com terms, what we
need to do is understand how data becomes information, and
information to an economist means information about people's
expectations and their preferences and their utility. That I think
is really the heart of this argument, and when Jeff says that
relationships and trust are important, certainly they are. Why are
they so important? I think the reason is because they give us a very
efficient way to reveal other people's expectations and preferences
very cheaply, and that allows us to allocate attention fairly
efficiently at a cheap price.
MR. BURNHAM: Tim.
MR.
O'REILLY: I wanted to pick up on what Tom said. First of all, I
really agree that there's a sign wave through the economics of all of
these things where we have this aggregation and the like, and I want
to just suggest a specific mechanism by which some of this will play
out, and I reason a lot by analogy and I've looked a lot at what's
happening in software, in particular with Microsoft, and if you think
about how developers made their deal with the devil so to speak, it
was around the solution to the device driver problem. So what
happened was that everybody had to write their own device drivers.
Every software company had to deal with many, many problems. Every
software developer had to deal with all the possible devices, right,
and write the drivers. And basically Microsoft offered a deal. They
said, "Wait a minute. We have this great idea. We will deal
with all the developers and we'll give you a uniform API." And I
think we're in very much of the same situation right now where we
have a whole lot of data format, some of them closed, some of them
open. And but that doesn't really matter, because what's going to
happen is somebody's going to basically figure out that the way to do
the interchange between all these data formats, and the aggregation
of all these data formats is for there to be some kind of aggregation
point where somebody says, "I'll give you a uniform interface."
That's what people like Grand Central have tried to do. I don't know
who will do it or how it will come about, but I think it's going to
be through the transformation of data types. And, again, we see it
maybe with Google Maps in that here they closed data providers,
Navteq, who meets up with Google, who's just a licensee, but then
because of the way Google implements, people say, "Wow, I can
mix this with other data types," and it's at the transformation
point that the place where multiple data types connect that there is
the opportunity for a new sort of economic leverage. So Google, you
know, I was told a story of how Navteq was the Intel Inside of all
the mapping services. That was until Google Maps came along, and it
was by upping the anti to become a conversion point and an
aggregation point of multiple data types of data that they became the
new power broker.
MR. BURNHAM: It's a powerful thought, because
to this point, I've been thinking that data really behaves
differently than code in an API, and that data flows much more easily
between systems than code does, and what you're pointing out is that
-- and for that reason, I was thinking that the Microsoft opportunity
doesn't exist in an open data world, but what you're pointing out is
there are still some wrappers around that data and some ways of
presenting the data and making it available that are unique.
MR.
O'REILLY: I still believe that there is a choice ahead of us, and the
choice is because of in solving the many, many problems, there really
are two solutions that have been proven. And one is I refer to the
Microsoft solution as the one ring to rule them all model, and the
UNIX and internet solution is the small pieces loosely joined model,
and I don't know which of those people will buy into. I think there
really is a race. And I think if you value open data, you want to
figure out the small pieces loosely joining model before someone
figures out the one ring to rule them all model. And on the other
hand, if you're the next Bill Gates, you want to figure out the one
ring to rule them all --
MR. BURNHAM: You want to find that ring.
Let's get Tom in here. We haven't heard from him, and then Peter.
MR. COHEN: I think in listening to the conversation and going
back to what Marc said, I think the real estate analogy is a good
analogy, because the developer chooses the architecture based on what
he perceives to be the need of the audience that's going to go into
the development of that construct. The open or closed has to do with
how he markets the development. So there are many cases cyclically
which we've seen, the AOL example, etcetera, where as a marketing
tool, if the end goal of the architect -- of the developer is to make
money, and that is what skews the equation, then he might choose to
have an open architecture as part of his initial marketing goal, and
as for the trust factor, he might choose to give away five houses to
influencers so you trust the neighborhood, but then once he has you
in his grasp, he closes the architecture. And so I think it really
is dictated by what the needs and the goals of the developer are, if
end goal is to make money. If it's not to make money, it's a
different equation.
MR. HERSHBERG: I was going to make a
completely different point, but I guess I'll just follow up on that
instead. I think that's a really fair point and I think the real
estate analogy is an interesting one, and in thinking about the
Craigslist model for instance, I may choose to participate in
Craigslist because I like what it stands for and I like everything it
represents, and so for that reason I give them my -- and I contribute
to the community. Then the question becomes -- and again, going back
to this real estate question, and then you're a specific development,
do I not want my development put on an 18 wheeler, driven down the
LIE and dropped off some place else where I may have not decided to
participate in the first place. And the question becomes do I in
fact want to end up in a place without my consent, which goes back to
this larger question of who owns the data. But I think that really
ties into the whole economic question, and Jeff made the point
earlier that Craigslist did something to the newspapers that now
something like (inaudible). And I think there's some distinction to
be made there, because Craigslist didn't take anything from the
newspapers. They created an entirely new data. So, at some point, I
think you have to ask the question when do the -- I mean, how much
intermediation is there among the intermediators, and at what point
does it stop, and when can you say all right, Craigslist created
this, somebody else is now taking this, and Craigslist should
recognize some benefits from it.
MR. BURNHAM: What you're
picking up on there, Peter, is I think related to the question of how
small pieces loosely joined or one ring that rules them all plays
into this sign wave of the transition into this open data
architecture. And what I'd like to do here is it is 12:30 now and
we're going to break for lunch, but before we break for lunch, I'd
love to just bring Yochai in for a second, and because the question
I'd like to put to you is, you know, do you think that either the
small pieces loosely joined or the one ring to rule them all, is
either one of those inevitable, and/or do you think it's possible to
create something that defies this sign wave, a platform of peer
production in an open data architecture that's sustainable and that
has sort of a consensual value proposition in that trust or
navigation space that keeps it going?
MR. BENKLER: I'll partly
answer that, but partly also characterize a little bit of what I
think I'm hearing. You characterized the problem as one of open data
architecture, but I think most of the conversation has been around
that question as one instance of the broader question of what is the
interface between commercial enterprises that are collecting -- that
are getting financing and building with the expectation of extracting
commercial value, but from a phenomenon that is very much based on
peer production. So in the context of data, it strikes me from this
conversation that the fact about the world is that lots and lots and
lots of value is being created by lots and lots of people, and
captured locally by them in the form of content creation and content
use without a transaction. And this creates a problem of management,
filtering, relevance, accreditation, which is something that
Del.icio.us allows people to do, and that function itself gets done
not through authority, but through distributed peer production of
claims of relevance and claims of things that are like tagging, this
is like this, and there are several dozen people that think this is
like this, so that becomes value. And then just take the example of
crawling Del.icio.us or not crawling Del.icio.us. There is -- I'll
take one step back. So the question for the commercial developer of
a platform is how do I get the most people producing the most value
that I'm providing -- in this case, relevance and accreditation --
while at the same time retaining enough control to be able to extract
value to justify my investment, my investor's investment, and this
goes to what Tim said in the last session about dialing up and
dialing down the level of control of how much do you open things up
in order to bring people in and how much do you close it down in
order to collect value. What you're describing in the question of do
you allow crawling and the constraints on that is if I don't allow
crawling, the thing that I have is my group of users now, which
somebody else can't then use to get their N+1, as Tom described, and
I'm growing it at a certain rate, and then the question was raised
well, are you growing it quickly enough or will you grow it faster if
you open, but if you grow it faster if you open, are you going to end
up loosing that advantage and somebody else is going to come in as a
second generation and actually find your wonderful platform. "Thanks
for creating this. I now found this extra level." And then when
Jeff pushes back and says, "No, it's not the control over the
relevance judgement of the people who are inside that are giving you
your value, it's the fact that people will associate that value with
Del.icio.us." So even though someone else -- lots of other
places -- and now it's not money, it's reputation -- lots of other
places supposedly create the N+1, and yet that doesn't happen. And
instead, what you get is more people seeing Wikipedia is valuable and
coming back. So the question in a sense on the table is whether
given that you accept the source of insight and rapid information
growth is radically distributed interfaces that give a lot of value,
all of the value essentially to the individual participants, and then
that creates a problem with creating control, is the control then
having better technology to sift through it? Is it being the first
mover and being understood as the place where you go? Does that mean,
for example, that you allow crawling, but you require reference back,
so that essentially each one of these users then becomes a replicator
of your end, whatever else they have, but then everybody else knows
that your end is the place to actually go and that's the place -- but
that seems to me to be the interface. The question of the interface
is what is the way in which you build your platform most quickly, and
I think the answer there is opening it, and do you have a mechanism
of capturing the reputation in the marketplace you contribute rather
than closing up and treating your currently, relatively slow growing
base of users and essentially co-creators of the value of the
relevance that you're creating as the thing that's preventing
competition. And if it's all true -- sorry for going on so long --
if it's all true that you've got hundreds of millions of producers of
content and the way to solve the problem with what's relevant is to
get millions of producers, then it's the many distributed loosely
company that will grow faster than the one ring, and so that is the
way and that forces you to actually solve that problem.
MR.
BURNHAM: Boy, why don't we just break for lunch. Josh, we'll pick
you up --
MR. SCHACTER: That was sort of directed toward me, so I
would like to respond. The interesting question here is an
observation that we're providing markets. Del.icio.us is a kind of
information market and that market itself has some level of trust to
the users which we're talking about and I agree, I think that's a
good analysis. I think the issue on the side, however, is that
markets compete and that the people who -- if the crawlers were just
crawling and providing that stuff, I mean, I'm well aware that I
would bring much more attention to the system, because these are big
directors of attention, but at the same time, they also are going to
compete, and that their end goal is to reduce the size of this market
that I have. So it's not just a simple as that analysis in that they
are actually not going to just do that. That is the complication.
If they were just doing that, then I would be okay with it. That's
the issue. Does that make sense?
MR. BURNHAM: Well, they are
going to -- markets do compete. If somebody creates an N+1 market,
it will compete. What Yochai I think is suggesting is the
possibility that there is a frame work within which you can allow
them to crawl that identifies you as the source of the value and that
what you're building is reputational value that, you know, that may
be something that's leverageable broadly.
MR. WILSON: Well,
we're going to get to business models after lunch, so I think we're
going to try to figure out where the rubber meets the road in all of
this after we eat. (A break was taken.)
MR. WILSON: So we have
lost a few people. Clay Shirky couldn't make it, so we have a couple
of spots around the table that I would encourage the people sitting
in the back to take, and I'll identify which ones those are.
MR.
BURNHAM: Let's be more ruthless here just so people don't have to
feel uncomfortable about it. Peter go sit in that place that says
Clay Shirky.
MR. WILSON: Okay. I think the tension that I felt
in the room this morning was everybody trying to stay on topic on
these sort of more academic I'd say things. We were talking about
peer production and open data which are the foundation for the
conversation this afternoon, which is, can we make money in this
world, and I think a lot of the entrepreneurs and people who run
businesses sitting around the table I think found the conversation
interesting, but were kind of eager to talk about how do I put money
in my pocket or maybe my investor's pockets. So, anyway, that's what
we're going to get to this afternoon. And I'm going to turn it over
to Brad to get more specific.
MR. BURNHAM: To see if I can start
that fire? Well, you know, we have touched on a number of ideas about
how one would begin to think about making money, and, actually, maybe
I'll just pick on Rikki, because we had an interesting conversation
at lunch, and Rikki Tahta has been working in the financial markets
for a long time, and he makes the point that that's an open data
architecture. That the pricing is actually very broadly distributed,
and there have been a number of efforts to create value and sometimes
proprietary value around that, and some successful ones. So maybe
you can talk for a minute about that.
MR. TAHTA: Just a quick
analogy from the financial services industry of data within that.
The last 20 years they've been struggling or fighting for open
databases, and now, the vast majority of financial markets data is
pretty much peer produced or commoditized data, produced centrally
and distributed, and yet there are still I can think of five
multi-billion dollar companies that make money out of data, run
businesses out of data adding value to the data. Now those are
traditional aggregation models, which can be cleaning the data,
normalizing data, but there's also been a history of financial
markets some quite interesting fuse as to where the value of adding
to data comes from. In 1987, Thomson's first call of business
demanded from investment banks, research organizations the exclusive
right to commingle their information. So they took data that was
being distributed freely, but they captured value around effectively
the commingling of that in a central source. In 1990, disclosure --
this is pre-Edgar -- took data out of financial statements and
copyrighted the organization of the data. Not just the meta data,
but literally how the column sat within your spreadsheet exactly
where they think that value adds to the end-user came from. And what
excites me this morning, I found I was listening to a lot of ideas
that I had seen in the financial markets occur, but one of the
exciting things that I feel being touched on, but my question now is
is the peer production somehow and this peer network somehow create
value, other than what a traditional aggregator does in cleaning,
normalizing, commingling and copyrighting some organization. So is
there something where -- I think Scott kind of put it -- where one
piece of data -- I produce one piece of data, the guy next to me
produces one piece of data, and somehow between those two pieces of
data, there's a third piece of value that comes out. That's what
strikes me as the holy grail.
MR. H. MORGAN: Certainly
Bloomberg did that with analytics. They did it by adding lots of
analytics that people didn't have easy access to.
MR. TAHTA: But
I view that still as part of the aggregator model. I got all this
data. I'm putting it in one place. I'm organizing it. I'm
providing some centralized value to it. It's not the producers
themselves.
MR. H. MORGAN: But the meta data provides that.
MR. BURNHAM: Jeff.
MR. JARVIS: Example on Del.icio.us is
person A tag, person B tag, I have a content, but together, you
created more value because you've created (inaudible). Where, in
turn, I get the value in other ways. The consumer gets to buy more
stuff, but I as the owner of that content get to target better
(inaudible). So there's new volume created, or even out of Web 2.1
(inaudible). If I take all the tags and the content associated with
that here and here and then analyze, then I can renew this content
and say what (inaudible). So there's an exponential growth in the
value of it, and it wasn't just the old fax machine. Metcalf's Law,
and I won't get into it because I'll get killed in two seconds, but
trust did have an impact, because there was a trust issue. There
was, in fact, wisdom. It was more than just keying it. It was truly
adding intelligence to it.
MR. PINCUS: What I add to that is I
think what I heard this morning and the pattern that I think we've
seen as all these markets grow is whether it's the financial markets
or these new kinds of markets is there's some enormous fixed cost
initially to making a market where there was none. Whether all
fragmented. Whether it's online dating or the web or ads or
whatever, and so someone goes out and creates a market, and for a
long time, they get paid a lot of money just because they created
that market, but eventually, it gets to the point where anyone can
bring that same group together, those commodities, and someone else
goes and does it cheaper, openly or whatever, but the one factor that
I think -- the formula that runs through all this from its starting
close to ending open is consumers are driven by time per lead. So
consumers, whether they're going to Craigslist or Hotjobs or Match,
they are always going to be viciously following the least time per
lead, and Del.icio.us, for a lot of people now, is reducing their
time per good lead versus someone else. You have to use a certain
community now. Advertisers are driven by dollars per lead, so
they're going to viciously go where they can spend the least dollars
per good lead they get, and I think the world that we're all entering
into is the one where community and identity and trust start to
enable new, better models for time per lead and dollars per lead, and
I think that it's an inevitable model where the Match.coms and the
Hotjobs and the Monsters, they're all going to -- and Ebay will all
be put out of business just because of these vicious market
phenomena, unless they go and play on the same curve. They're off
curve. So Ebay, by making more and more money every quarter, is
eventually off the curve, because someone can eventually find -- the
advertiser can find a buyer for their good for less money somewhere
else.
MR. BURNHAM: So I think the time per lead and the dollars
per lead is a very good way to think about the potential for hyper
deflation in a flat, connected, you know, information economy. But
what does that mean in terms of the ability -- I mean, you almost
think about, you know, enterprise value as being associated with
trying to create some disruption in that flat economy, some pyramid
or some peak in that economy. Is it possible for us to do that?
MR.
D. MORGAN: This came up from John at lunch, but if you take a flat,
if you commoditize everything, then the brand becomes the
differential. Sort of the peanut butter principle where you go into
the grocery store, you have six apparently similar, the same chunky
peanut butters and all you really have to differentiate them is the
brand. Well, what makes the one jump off that the others didn't? And
maybe there are some mixed ones that start getting better, but it
takes a long time until they move to the center of your attention and
can buy the shelf space and all the other things that go with it. So
I think that's one of the big things why the wall gardens have a lot
of defensibility. Arguably, reason AOL is flat is not just because
it didn't open up its content, it's probably because that wasn't the
only thing. Maybe that wasn't just -- not just opening up any
database, it didn't open up the most important database to open up at
the right time, and Ebay figured that out.
MR. WILSON: I wonder
if let's take Craigslist for an example, they're right here at the
top, Craigslist makes money, could they make more. Today, the
Craigslist model is you list for free, except if you're listing for
employment jobs and real estate in certain markets; is that correct?
MR. BURNHAM: It's really employment in three markets and they're
just beginning to experiment with real estate.
MR. PINCUS:
Commercial real estate in New York.
MR. WILSON: And so they
really haven't monetized that business very much.
MR. BURNHAM:
Well, this is the point that Tim made earlier this morning when he
said that many of these businesses, the relationship -- the ratio
between the value provided to the consumer and the amount of
monetization taking place is different in these businesses.
MR.
WILSON: So I want to propose an idea of how Craigslist can become
more open and potentially make more money at the same time. I'm
curious of what people think of this idea. So if Craig recognizes
that eventually these listings are going to get distributed, that
whether it's Oodle or Indeed or whoever is going to crawl them and
they're going to get outed, and it's too bad Matt Turck's not here,
he had to leave, Matt started this service called Word of Blog, which
I like, but if Craig simply said okay, I'm going to be the place that
people come to create the listing, and on top of that, I'm going to
let them provide a rule that stays with that listing, and the rule
would be if this listing gets clicked on, I will pay you 50 cents, or
whatever that rule needs to be, and then the listings get everywhere,
but then any time anybody interacts with that listing, there's a
business rule that delivers value, and because Craig was the place
that that rule was created, he might also be able to be the place
where that rule gets sort of transacts and the value gets
distributed, and I wonder if that's where these large brands need to
go.
MR. HEIFERMAN: Isn't that AdWords?
MR. WILSON: That is
what AdWords is. That's exactly what AdWords is. AdWords to me is
the defining model of what an open database web service -- the only
problem is that AdWords won't let Marc or Dick take their feed and
run it wherever they want. And that's why everybody else in that
business is probably going to go and do that to them.
MR.
COSTOLO: Right. So I think the definitive answer to your question is
yes, Craigslist can make more money doing that, and that gets to --
and I may be putting words in Tim's mouth -- what Tim was talking
about this morning in creating just an architecture for providing
transformations and APIs around transformations is a much more
compelling way of outflanking your competitor, because once
Craigslist does that, there becomes no way to out aggregate them.
They've just already said anything can aggregate me. I'm just going
to provide some APIs or rules for how to, what happens to the concept
when it gets aggregated. So it becomes much harder to outflank them
when they do that. And they then also defined a market for how they
make money.
MR. PAREKH: And pushed out the walls to their
garden.
MR. COSTOLO: Right.
MR. O'REILLY: I'm not sure I
agree. Here's the issue I see. I think you ultimately, in the long
wrong, you get money from doing things that are hard, right, and if
you think about the network effects, and that's, you know, that's
wonderful. You build this nice, you know, you can get a great
network effect doing something that's easy, but to extract money, you
have to do something that's hard. And so, for example, take Google
AdWords, it's like when you talk about a business rule, the question
is, well, there's a lot of in that -- you know, what is that business
rule? Well, the business rule of AdWords is I will actually find a
relevant advertisement for you, and that's hard, right. It's not
just something that they just kind of put a rule that "Thou
shalt get thy ads through Google." They're good ads, and it's
not just that they happen to aggregate the advertisers to do this
sort of self-service opportunity, they've also figured out somehow to
track and they've done a lot of stuff that really works. And I guess
what I would say is I don't think there's any magic that's going to
just -- somebody can just sort of syndicate it out. To me there's
network effects in building a big aggregate, and you can definitely
-- there's sort of a set of rules for how you can set something like
that in motion. As Flickrs demonstrated. As Del.icio.us has
demonstrated. And that's I think step one. The web demonstrated it.
(Inaudible) Tim figured out the simple things that had network
scaling effects. So I think for us, the question as investors has
got to be what is hard about what, you know, someone is doing.
MR.
BURNHAM: The only thing that doesn't ring through there --
MS.
GREEN: You don't think the graphic interface is kind of hard?
MR.
O'REILLY: The graphic interface for what? (Inaudible). He didn't
actually invent the graphic interface. He made quite a lot of money.
MS. GREEN: But wasn't he actually --
MR. O'REILLY: It is
debatable.
MS. GREEN: When he released it as open, right, Tim?
MR. O'REILLY: Tim did. Right.
MS. GREEN: They did with it
what they wanted.
MR. O'REILLY: Oh, yes. I guess all I'm saying
is the extraction of value, if the barriers are -- well, first of
all, I really do believe that you get network effects. You know, if
you look at Ebay or Amazon or Google, they're all network effects
businesses where they've built some critical mass where the user is
adding to the service in some way gives them critical advantage. But
I also think that they've each added some things which mean they
can't simply be displaced by someone else who just kind of says, "Oh,
yeah, this is easy."
MR. BURNHAM: Let's get Tom in here.
MR. EVSLIN: Let me try gaming a little bit. I'm sorry, if we
think of Craigslist's strategy, it's a brilliant anti-competitive
strategy and I don't mean a value judgment by that. How do you
compete with them? If you're not in the areas that they charge for,
then you can't make any money because you can't charge for those
areas either. So you can say, "Okay, well, I'll go compete in
San Francisco by giving the ads away," but you don't have the
readers in San Francisco. So what they've done is they've said that
I've got the network effects from all the places that I give the ads
away that I use to make the ads more valuable in the places that I
sell them. And so I don't think it's just negligence that they
haven't got around to charging for the other places. It's
deliberately focusing the value on where they're extracting the
money. It's like the Microsoft strategy of always pricing under
Lotus, because you just don't want to leave any room for a
competitor. And so here's the brilliant I'll price it free in most
places so you can't go around me and get in anywhere, and I'll expect
value from a few sweet places which you can't beat me there because
the value I'm giving away for all the free places --
MR.
O'REILLY: So how does the value in some other city add to the value?
MR. EVSLIN: Because of the job seeker. Remember the network
effect, right. So the job seeker is often not as localized. They
don't necessarily know they want to go to San Francisco. So there
are some ads where it adds no value, obviously, Tim.
MR.
BURNHAM: I hear lots of people saying I find my life on Craigslist.
So it's not I just found my job. It's I found my apartment. So
there's a whole bunch of other services that keep people coming back
to Craigslist, and therefore, aggregate the audience around jobs in
that market.
MR. EVSLIN: But it's really all about creating that
network effect and not really leaving room around the edge.
MR.
PINCUS: Why does Craigslist grow in a market like Milwaukee where
they don't have crap on there? They still grow. There's something
that's missing. Users are not driven. They're not going to
Craigslist because it has the most of anything.
MR. HEIFERMAN:
No city is a vacuum. People in what did you say, Milwaukee? People
in Milwaukee hear about it from their friends in Chicago, but I think
we're sort of -- the question on the table is kind of being asked in
the world view of sort of the one ring to rule them all in a certain
way, when here you've got -- one of the most fascinating things I
think in recent history is how much of a weird dude Craig is, and
that's how -- and how, you know, there's a giant body pile, a giant
body pile of all the MBAs and marketing people that pat Craig on the
head in 1997 and '98, '99, 2000, 2001, 2002 that said, "That's
nice, Craig. You do your little thing. We're going to build a real
business here." And they absolutely didn't expect to come out of
Craigslist what it is, and that is because he is such a weird dude.
And he's not sort of thinking this way. But really connected to that
is --
MR. GOLDSTEIN: Not weird enough.
MR. WILSON: Can you
get weirder.
MR. HEIFERMAN: What were we just talking about,
because I wanted to make --
MR. JARVIS: That was a weird moment.
MR. BURNHAM: We'll come back. Michael.
MR. PAREKH:
Craigslist has been talked about as the poster child. I'd be curious
as to what this group thinks what the other poster child is for
Wikipedia. How would you --
MR. PINCUS: What, Michael?
MR.
PAREKH: The question is Craigslist has become Craigslist because "he
was a weird guy." He let it happen for free. Maybe there was
tactical elements as Tom mentioned, but there was ultimately a lot
more value given out for free essentially for him to build a network,
and one of the points that was brought up earlier is you got to let
this thing grow to a certain point before you get the network effect
before you can reveal the big model.
MR. BURNHAM: Well, Jeff,
you were talking at lunch about some of the other initiatives.
MR.
JARVIS: He's got Wikibooks, which Wikibooks you got the textbooks
competing with the textbook world and maybe (inaudible). So my 13
year old needs to learn how to do more about Linux and stuff and I
can go to Wikibooks now and find things for him there.
MR.
PAREKH: So there would be Wiki For Dummies on any --
MR. JARVIS:
Well, yes. It's about capturing knowledge. And so could you put ads
on that? Yes. Could you sell those ads for a price? Yes. It becomes
a gathering point of something. And he doesn't make money at that,
right. You can instead use that as the beginning of a tutoring
network. You know, who knows what the value will come out of it, but
I want to hear what Tim thinks about it.
MR. O'REILLY: You know,
it's hard to know. I don't know that it depends on which market. My
basic thinking about online books is that books don't exist, you
know. Books do various jobs, and the answers are very different
depending on what jobs we're looking at. Already, you know, Google
and the web have wiped out a whole pile of reference books. Other
books are about teaching, right, and so he's setting aim at
textbooks, and it's certainly true that it's quite possible the
product can be developed. That the question, though, for example in
the academic system is the distribution system for books is fairly
resistant actually to the quality and content from outside the
system, because the way that the textbooks, you know, for example in
K through 12, they're actually adopted by committee, and it's
typically in a few states, and then other states follow suit. In
Higher Ed, it's actually one professor who adopts his buddy's book
because he expects the same thing to happen in reverse. So there's
often information about channels that's kind of missing from. So
reference is pretty clear that the web is kicking the butt of
traditional publishing. In other categories, entertainment is a
third category, and already I tend to think the book has been
bypassed in many cases by other forms of online entertainment. And,
you know, in the entertainment world, it's pretty clear that while
peer production definitely, you know, can play a role, high
production values really matter, use that small games, and we're
getting back to that with casual gaming, but the costs of now
approach the cost of the developing movies. And so people basically
find a way to up the anti again.
MR. H. MORGAN: One of the
business models, by the way, is hardware. Apple obviously with the
iPod shows the way to make money with hardware. And once we get a
really effective electronic book platform, that will destroy, you
know.
MR. EVSLIN: That suggests a possible monetization if you
charge (inaudible).
MR. BURNHAM: At a slightly higher level of
abstraction. What we're saying is look for the intersection between
hardware and software. Look for the intersection between the
information and the transaction for the area in which you're going to
begin to develop that value; is that right?
MR. EVSLIN: I think
lesson one is don't try to build a business and network at the same
time. Don't try to build a business and a network at the same time,
and the reason I'm saying that is because of the --
MR. WILSON:
Because you didn't take your own advice when you launched your book?
MR. EVSLIN: I'm giving the book away now, right. Let's forget
Reed's Law for a moment because Metcalf's Law is steep enough, that
everyone knows that Metcalf's Law says use networks that increase
value, and everybody forgets the converse, which is small networks
have no value. And so what value is there to the first few users in
joining a small network. Almost none. And so you can't extract
anything in return. You can't put friction in the way of people
joining a small network. You have to make it incredibly attractive
and easy. That's the secret of Skype's success. They only had
distribution experience. Then they used that, and they didn't
introduced Skype in and Skype out because it would have been a
distraction until they reached critical network mass.
MR.
BURNHAM: Let's keep going down the list. Google, I mean, launched in
'96. Revenue in 2000 for the first time. Something like that.
MR.
PAREKH: In fact, they discounted heavily the distribution deal with
AOL and others relative to what they could have monetized that for.
MR. BURNHAM: So there's a long history of people who have
successfully built networks and invested. From an investor's
perspective, there was an example of someone who was willing to
invest for a long time before they took the risk of putting friction
in the networks.
MR. EVSLIN: Yossi Vardi, who was the poster
child for that with ICQ. I said to him, "Yossi, how are you
going to make money?" He said, "I'm probably not going to
let myself be distracted by that. That's not what I'm doing."
MR. BURNHAM: And Skype.
MR. EVSLIN: And Skype is another
example.
MR. BURNHAM: We didn't hear the joke.
MR. PINCUS:
Seth said revenue is a distraction.
MR. GOLDSTEIN: I was quoting
Yossi.
MR. PINCUS: I said investor's are the biggest
distraction.
MR. WILSON: You better remember who paid for your
lunch.
MR. BURNHAM: All right. Let's come back to --
MR.
EVSLIN: So you have this enormous network value, but it's not until
you get there that you can worry much about how you're going to
monetize it. Maybe don't monetize it directly. Maybe do what Skype
did. You sell it to somebody else who has an infrastructure for
which that network, you know, that that network can add value to,
even though it shouldn't have grown to value by itself.
MR.
HEIFERMAN: Skype has growing revenue, and I would just say, revenue
let's you be confident in sustainability and growth and support of
the network.
MR. BURNHAM: But let's challenge that, because if
it is a peer produced network, it could actually compromise the
sustainability and the growth of it. So if you introduce Google
AdSense to Wikipedia, it might in fact compromise that, or if Craig
became blatantly commercial, it might in fact compromise that.
MR.
D. MORGAN: It would also depend who got the revenues from AdSense or
Wikipedia. I don't think it would bother me one bit if I thought it
was going to the foundation.
MR. JARVIS: What you're saying too
is sometimes the infrastructure has to resist for flowers to bloom on
top of it. So the fact that more press being an open source, it
exists, and flowers are being let (inaudible). But that's where you
try to reach value. We'll all help build it because I know --
MR.
BURNHAM: Well, this is a point that Tim was making earlier, which is
you may not be the one that capitalizes on the platform you create,
but what Tom is suggesting is that Google did create the platform and
then capitalized.
MR. JARVIS: It's the advantage of creating the
network without revenue that you're the first guy to capitalize on
it, and others will too, but you will benefit first and most because
of that, not because you created the network.
MR. O'REILLY:
Tom's point is really about timing.
MR. EVSLIN: This is just
about investment strategy.
MR. BURNHAM: Let's get Michael in
here. Then Josh.
MR. FRUMIN: A question I wonder is if as
people, as all the peers doing the producing begin to realize that
they're creating all this value for other people and they're used to
getting some kind of value back for themselves -- you make a web
page. You create a value on Google. But once more, the people who
are doing the producing are realizing that they're creating economic
value for other people, and there's a way to share that economic
value more explicitly with the people. Can that contribute to the
growth of the network. So not having destruction of growing the
network and then creating value that --
MR. O'REILLY: Isn't that
what Google does with AdSense?
MR. FRUMIN: Well, it started to,
right. But the question is --
MR. BURNHAM: Well, Yochai would
make the point that introduce financial incentives into something
that is otherwise driven by something else, you can potentially
destroy it.
MR. WILSON: Is there an example of that? Yochai, is
there an example of that where someone had a nice flowering peer to
peer business, they introduced too much commercialness into it and it
just blew up?
MR. FRUMIN: But not just commercialness, but
commercialness for the producers, for the peers doing the producing,
if they can get value.
MR. BENKLER: I mean, Heather said
something here on the side that I think is important as a
disciplinary measure for this conversation which is Bubble 2.0. this
needs to be a standard -- a standard question that we're all asking,
because the idea of don't worry about it, let's grab the eyeballs is
familiar.
MR. EVSLIN: We've heard that.
MR. BENKLER: So
that's important. The other thing that's happening here -- by the
way, you'll notice I say lots of things and I won't answer them. The
other thing is that there are two design questions now on the table,
not one, with regard to the relationship between our one,
understanding the source of value being peer production in the social
relationship, and number two, understanding this conversation among
people who are interested in making money. And one has to do with
getting the social cultural framing of the behavior right so that
people -- and this goes to Scott's point about Craig being a weird
dude. That is to say, having the characteristics of yeah, you know,
we'll do it for free, and we'll then figure it out and having being
trusted to do that, which has to do with people being willing to put
in more, and that's actually where I would say that if you are
looking at introducing economic value and people begin to see the
monetized value of their contribution, and the only -- well, the only
thing that at least comes to my mind at the moment is the difference
between the free distributive computing systems and the one or two
efforts to create paid per cycle or paid per month of cycle
distributive systems, and if you look at all of the successful large
scale distributive computing systems, they all give people a sense of
what they're for. A sense of social mission. They render reasonably
transparently the success of the common project. They provide a
particular platform for people to express how much they're giving by
saying who's the user of the month, the user of the week. They
actually do a lot of mix and match, but when you compare them -- I
don't remember their names now -- there are two or three efforts on
the web who actually pay people X dollars a month for your -- and all
of those have been very small and very flat, even though they're
trying to do the same thing. There are some complicated factors,
because they're actually not doing exactly the same thing, but that's
one place.
MR. BURNHAM: Let's give Martin a chance to get in
here.
MR. NISENHOLTZ: I think at some level, what you're trying
to do at this meeting, Brad, is reduce risk. I mean, that's what I'm
hearing. In other words, what I'm hearing is there's this notion of
peer production. Most of the companies that evolved out of this peer
production idea have been created almost -- goes back to my lottery
point, although it wasn't a perfect metaphor -- but almost out of
lightening striking with weird dudes in the middle of it. Brad, what
I'm saying is, as investors, what you're trying to do is you're
trying to provide a systematic overlay to this conversation in order
to reduce your risk, and that's a perfectly credible thing to do. I
just don't think it's possible, and I think at the end of the day,
the success of this meeting is going to be measured in some part by
you or perhaps should be as to whether we as a group can figure out
how to identify and then in some ways reduce the risk around
investing in these very weird entities.
MR. BURNHAM: Yes. I
would say whether it's a personal investment in --
MR.
NISENHOLTZ: No. No. I'm not accusing you of anything.
MR.
BURNHAM: But one way to think about this is Heather's point of Bubble
2.0, and basically is there something different between Bubble 2.0
and Bubble 1.0, and is there anything that we've learned. It's
smaller, right. If we can, let's get Bruce in here.
MR.
SPECTOR: I think it needs to be said that in the early days of the
"commercial internet," what have now become some of the
very most successful players, Yahoo, Ebay, Google, with, of course,
some exceptions, like Amazon, were not really started as businesses
as all. They just were not. They were started as passionate
enterprises.
MR. PAREKH: They were the Craigslists of the time.
MR. SPECTOR: Yes.
MR. NISENHOLTZ: And a ton of them failed.
MR. SPECTOR: They shared as the impetus for their creation the
passion of the founders, and that's what they were. They were that
and they were only that. It was David and Jerry's directory that
became Yahoo, and it was a better search algorithm. It became
Google, and it was some passion to create Ebay. But what's
interesting is that there appears to be a cycle at work. So you need
the innovative passion. You need to be able to simply have creative
people, artists if you like, creating whatever the social networks
are, creating whatever the network effects are by whatever means, and
then at the point that those things are engaged, then the MBAs come
in, the investors come in, and various monetization schemers are
effectively applied with more or less success. And I think that's
the cycle that we've seen in Web 1.0. and I think it's likely we're
going to see a variant of that same cycle in this new cycle.
MR.
WILSON: But part of the question we want answered is when do you
bring the suits in, right?
MR. SPECTOR: I think to a certain
degree, we're answering some of those questions, what Yochai has been
saying and others, Tom, that the efficacy that the web has brought as
this new network environment is in fact that network effect. The
ability to, by virtue of aggregation of data, if that's what it is,
by virtue of aggregation of eyeballs, if that's what it was, for
Yahoo in 1998, then by virtue of that kind of effect, you gather
enough interested parties in whatever that network environment is to
generate revenue opportunities. And then that's the time presumably
to go to your question, "when do the suits come in." It's a
little bit like, if you will, it's a little bit like creating
Seinfeld. Seinfeld wasn't successful from day one. Seinfeld needed
time to gather an audience that became, you know, a very, very large
and successful audience. So these things take that kind of time in
cycle, and it's hard to know exactly how much time there is.
MR.
BURNHAM: There's one thing --
MR. SPECTOR: You're going to see
more successful ones, but the notion of so much the environment, I
think what it really is is it's a very low barrier of entry to create
network effects. That's the cool part. The cool part is it doesn't
take a lot of capital to create a quite large network effect
relatively quickly. Del.icio.us is a recent example. Passed that,
then you need to attach, if you're going to go after money, then you
got to attach a variety of really quite well-known models for how you
make money. Whether it's an advertising based model. Whether it's a
subscription based model. Whether it's a brokerage based model.
MR. BURNHAM: One thing that we've noticed is that a lot of things
that we really like are up and running and working on the web,
aggregating passionate group of users before the person who created
it quits their day job, and that's not something that you saw, for
instance, in the enterprise software development world or in the
hardware world or the chip world, you couldn't do that. So there is
a difference in the capital requirements for taking that first step
and building that first passionate audience. The point that you're
making, Bruce, is that it may be that taking that from there to some
kind of business, then will require a certain amount of investment.
It may not be --
MR. JARVIS: But it may be less.
MR.
BURNHAM: It may be less.
MR. D. MORGAN: We need to see a
different part of strategy, though, which says because the cost of
creating networks is so low now, you have all sorts of people
creating networks and you can sit back and wait until those few that
make it out and you can go pick them, but I don't think that goes
against the fact that there may be some of the most powerful networks
that we haven't seen because it's still so immature that are
commercial from the get go. And, actually, may be much more
successful because they recognize that they're building up around an
actual customer or problem that people will take on and not the
accidental 164 million that can be extraordinary.
MR. BURNHAM:
Josh. Then Tim.
MR. STYLMAN: I was actually going to make a
similar point to Dave and I think something important to keep in mind
is that where the network is built is not always where the revenue is
derived from. So we were talking about credit earlier and sort of
giving everything away for free. I make the same case about Tim
mentioned that Google kicks the reference world's ass -- well, my
words, not his. So Google does not make money from reference
generally. Even when people are monetizing, not a lot of people can
pay a lot of money for reference keywords. However, it gets people
coming back to Google, and presumably some percentage are going to be
more commercial in nature, thereby building those things.
MR.
BURNHAM: Tim.
MR. O'REILLY: Actually, couple points since I
first raised my hand. I keep hearing it's easy to build networks,
and I would just say it's easy to build networks early in a new
market, right. I mean, I built the first commercial web site, GNN.
And it was really easy to do, all right, but we sold it because we
realized that it was going to take capital to grow it, and I wanted
to stay private, so I sold it. And the fact was there comes an
inflexion point where you've got the organic growth and the next
stage does take money. So there's definitely I think still an
opportunity for capital. So the second point that I wanted to bring
to this whole line of discussion came off what Yochai said about the
distributive computation start-ups and it brings us back to peer
production and what drives peer production. And I was part of an
event that Macro Media held the other day, and they had a speaker
(inaudible). He was talking about why do people pay so much for
coffee. Coffee is a commodity. So put that thought in one spot, and
then I want you to jump to a comment probably of all the things that
happened at the Web 2.0 conference, the thing that struck me the most
is a comment that Michael Powell made. He said, "My kids
basically think the music should be free. Yet, they have $60 a month
ring tone budget. And why is that?" He said, "Because
music is something you consume privately. Ring tones is something
that you -- they're part of your display. Kids are basically used to
paying for this." And there's something in terms of all those
thoughts about why do people contribute to folding at home, but not
to a paid start-up. Why do kids pay for ring tones but not music.
And this whole idea of the experienced economy. What is it that
we're increasingly paying for in an economy of abundance?
MR.
WILSON: Bling.
MR. O'REILLY: It's basically bling.
MS.
GREEN: (Inaudible).
MR. O'REILLY: Absolutely. Yeah. From the
provider point of view, you get money because it is closed and that's
the only way you can do it, but they've been able to sell it, I think
partly because -- well, first of all, you know, they have produced
pricing that people thought was reasonable early in the market. But
I do think that an interesting concept in that is what do we value in
today's economy.
MR. PINCUS: But, Tim, I think it's more basic
than that. Anything that can be free will be free. So to Heather's
point, the only reason they pay for ring tones is they can't get them
for free. Kids paid for music when they had to, and they did, and
they'd save up for it, and now they don't, so they won't. I just
think anything that can be free will be free, and then there's
questions of laws that may stop it and that's it.
MR. BURNHAM:
There is a start-up that's like an open source ring tone start-up.
MR. WILSON: We were talking about it at lunch. It's a service
that allows you to upload an MP3, create a ring tone and have it
SMS'd to you and you use it. And Shana said she thinks that that
will get shut down.
MS. FISHER: A lot of labels are looking at
it. And that constitutes use for them.
MR. SPECTOR: It doesn't
have to be a service. There's software available that allows you to
take any MP3 file and turn it into a ring tone.
MR. BURNHAM:
Well, it's interesting, because even with the existence of those free
services, you know, at three bucks,
you know, from the
carrier, I'll do it. So there's a point at which people will pay
maybe for display, you know, even at a price.
MR. WILSON: Seth
was going to say something about five or 10 minutes ago and you never
got the opportunity.
MR. GODIN: Well, the topic went away and I
thought it would be selfish to bring it back. Well, I was just going
to try to recast some of what we're here for, and say what if we
think of this entire topic of the whole day as just another way of
talking about marketing technique, but it's a marketing technique
that's extremely counterintuitive and that's almost never attempted
by anyone who has an alternative. Meaning, that if you got investors
or cash, you want to control certain things. You want to spend. You
don't want to treat the people who are providing the stuff you are
controlling with the same level of respect that you would someone who
you're paying. And so what we see is that these very low-funded,
garage-type start-ups use these marketing techniques to do all these
counterintuitive things and we notice the ones that work, and the act
of interacting with Craigslist makes you talk about it. The act of
having auctions on Ebay makes you talk about it. People notice them.
So, in essence, it's a marketing technique, and the reason to date
that traditional companies haven't done it is you'd have to make all
your decisions along the way and always do the opposite of what your
instincts tell you to make it go. So I want to give Lens Masters as
much as I can, but advisors say well, you can get away with paying
them less. Well, if you're viewing this as a marketing technique as
something that is cheaper than running ads in a newspaper, then it's
totally appropriate to pay people's budgets. You treat these people
with respect and generosity, they're going to come and not go
somewhere else. And the reason why I want to say this to the group
is if we're doing this for Brad and Fred, which we are, and they're
looking at the world to say, "What should we do and how should
we help our existing companies do this better or find a reason for
doing it?" I think it would be interesting to look at it from a
marketing point.
MR. HEIFERMAN: Actually, I want -- I'll bypass
and throw something right back. Why do you use the word co-op and
how does that -- Seth is using the word co-op. Why that word, and
how does it relate to peer production?
MR. GODIN: When I think of
co-op, I think of two places, the Harvard Cou-op and the grocery
store where you have to pack your granola in a bag in order to get a
discount, and so we're not legally a co-op, and we make it clear to
people we're not legally. That's a structure you could have. What
we're saying is that if money comes in, money goes out. We're
sharing with people who contribute the most to the community, the
same way IRI or the Harvard Co-op sends you a check if you shop there
a lot. But part of the story we're telling, which says you should
join this and contribute because you're going to get something back,
which you can give to charity or keep, but it's different than most
organizations. Wikipedia says there is no money, and other
organizations, Google says, "We're going to keep it."
MR.
HEIFERMAN: I just want to respond real quick. You sparked me to
really study up -- when I started using that word, you sparked me to
study up on co-ops, the co-ops recently, and maybe most people are
aware of this, I wasn't aware of this, did you know that, you know,
many world famous brands are co-ops? Sunkist, Ocean Spray, AP, Best
Western. It really is a fascinating thing to look at in the context
of what we're talking about here. These are not hippy commie
concepts. But if I could just adjust one quick note about this whole
past hour, which is that the, you know, at Meet Up, Fred asked the
question can you come up with an example of something that had
momentum, a network that had momentum that introduced commercial
interest and then failed. I think that Meet Up is out of the woods,
but the jury's still out. We had 50 percent organic growth on
basically a free service in the first two months of this year, after
people thought we would be dead after the election. Then we
introduced a mandatory fee, and we lost half of our active Meet Ups.
Now we're at about 80 percent of where we were before, but, you know,
there is always that nagging thought of if we grew 50 percent in the
first two months, what would we be here in November or October if we
hadn't introduced the fee in April. But the big point there is is
that this notion of like, well, first you build a network and then
you introduce the monetization, I think with all respect, that's very
conflicting. The point is, for Ebay or for us or for, you know, like
the question is how does the monetization make the product better. I
would argue that AdWords makes Google a better product, not just is
this sort of like you have to sift through the commercial messages.
MR. GODIN: And your fee makes your service better. Also, Ebay
and AOL were monetizing from day one.
MR. GOLDSTEIN: Just this
point of co-op is I think a useful segue to something which hasn't
come up, which is related to monetization. So USAA, anyone belong to
USAA? It's military families. One of the largest financial
institutions in the world. 80, 90 billion dollars at least. It's a
co-op. It's essentially owned and operated by the families insofar
as they all -- sort of think about this from an open data frame work
perspective. They're aggregating this audience and they're kicking
back a lot of the savings to the users. The users get some kind of
dividends at the end of the years because you're owners. And it sort
of works. So when I think about -- we talk a lot about well, we got
to use organic network e-business and you have to figure out
monetization, somewhere in between is price. Price discovery. Price
discovery is not monetization. Price discovery is what's the data
worth. And that comes before it gets sold. You can't go to the
selling of the data before you know what it's worth, and I think
we're all trying to grapple through A, who owns the data, and B is,
what is it worth. What is a click worth. What is a weed worth.
What is my click stream worth. What is my attention worth. What is
a tag worth. And you feel, at least in the beginning of figuring out
a valuation around the consumer as opposed to the publisher. And
until we get there, you're going to be flailing for business models.
Not to complicate things, but there's a lack of price discovery in
Google. They're a price higher, not a price sharer.
MR.
O'REILLY: Right. But they figured out the price discovery. There's
a big variation in the price -- in fact, that's what I was going to
say. One of the things that Google teaches us -- and this is maybe a
bit of an answer to Martin as well -- there are different business
model for different classes of data. The price of a Google AdWords
for Java programming is 11 cents. For some other category where
there are products for sale is 50 bucks, and the fact is that, you
know, that $50 cost per click in one area will support a very
different type of publishing business. I mean, Matt's blog works
because he's in a high cost per click area, whereas our stuff in
technical information, we have to have a subscription business model
because you can't make it on cost per click because the click's not
worth very much. And so I think that one of the things that we need
to learn, I think your point about price discovery is a good one.
It's not one size fits all.
MR. STYLMAN: Well, the network
decides the price. So the business model is completely aligned with
the participation of the community.
MR. BURNHAM: So this is a
peer produced price discovery mechanism that Google has created.
MR. GOLDSTEIN: But there is no price discovery. It's their
price. Google's exchange.
MR. EVSLIN: There's a whole industry
of search engines.
MR. GOLDSTEIN: The point with Google is there
is one broker set up to sell access to that exchange, and it's called
AdWords, and you basically either trust the prices you're getting
from AdWords, or you trust the price you're getting from AdWords.
MR. BURNHAM: So you're talking about the cost per share to trade,
not the actual --
MR. STYLMAN: Well, there's a difference between
price of discovery and ultimately --
MR. GOLDSTEIN: But so much
less.
MR. KURTNIT: It's just you don't have (inaudible).
MR.
H. MORGAN: In Overture, we built a model. We actually showed the
price per click on every search and --
MR. BURNHAM: Speak up,
Howard.
MR. PINCUS: You're assuming that the end-user cares.
MR. BURNHAM: That's the interesting lesson.
MR. PINCUS: I
don't think anyone gives a shit. I don't think they care about the
value.
MR. BURNHAM: Wait. Let Howard finish the lesson.
MR.
H. MORGAN: The lesson was that the end user didn't care. The
advertiser cares, as long as they controlled and could see it. Once
Google came in and closed it and Overture (inaudible).
MR.
WILSON: Tom, we're going to give you the last word on this, and then
I want to change direction a little bit with this.
MR. EVSLIN:
Go back to what we were talking about is how Bubble 2.0 is different
than Bubble 1.0 it's not different in that prices aren't going to get
overinflated. It's not different in that it's not going to collapse
because that's what bubbles do. It is just like Bubble 1.0 in that
it will raise huge amounts of money in what is essentially a lottery
that will fund a discontinuous advance in infrastructure. But one
big difference and I think a constructive one when we're talking
about investment is in the exit strategy, and the exit strategy, the
IPO is the dominant exit strategy was one of the real weaknesses of
Bubble 1.0 and a lot of good companies were born like premature
babies -- and I'm not making personal statements -- but you can't not
go public if your competitors are going public. And so once
companies with no profits, with no revenue, with no predictability
could go public, the companies themselves were hurt. I'm not
worrying about the investors now, but the companies were then unable
to perform the way that they could have.
MR. PAREKH: But in
today's Web 2.0, the IPO market is not there as much as it was in the
first one, but the exit strategy of choice is selling it to an
incumbent.
MR. EVSLIN: I absolutely agree with you.
MR.
PAREKH: But when they do that, that company then is changed. Myspace
is forever changed now that it's part of -- their priorities are
going to be changed.
MR. EVSLIN: And some of them will certainly
be ruined, but I think that the companies run in expectation of
acquisition, because they still can be run privately -- can be run
better up to the point where they're acquired, and there's better
chance that post-acquisition -- not 100 percent, you know
acquisitions fail plenty -- but post-acquisition is probably better
than post-IPO, particular for a class company, and Fred, you wrote
about some of this, that we don't know if some of the companies that
we admire now are really companies or technologies or networks, you
know, can they really turn into a business. I question that about
Skype. I don't want to reopen that argument, but the answer doesn't
matter if there's a market that can fold them into something that's
money making.
MR. PINCUS: You're talking about rationality, Tom.
You're saying what you think is more rationality driven in this
market --
MR. EVSLIN: I think from an entrepreneur's point of
view, and I'm not a player this time around, but I think some
entrepreneurs will be able to do a better job of running their
companies, not that lots of monies are going to be lost, because the
IPO was almost victory through death, and this is a better exit
strategy both to plan for and to effectuate.
MR. JARVIS: Isn't
the exit strategy also to risk high, in that because you start things
so light -- I don't know. (Inaudible) I had lunch with Ed Siv
yesterday about this topic where you say all right, you're going to
start right. You get up while you still have your job. You learn
you didn't need anybody. So then you come along and say, "Okay,
it's great." Certainly smaller cheaper, but how do you manage
many more investments to get to the same level of investment. And
how do you find those things. Your entry is almost harder now than
your exit.
MR. EVSLIN: I think Jeff has a very good point. You
become angels.
MR. BURNHAM: Do we become talent agents?
MR.
COSTOLO: Angels.
MR. BURNHAM: Angels.
MR. JARVIS: It's
about at some point you have money and you have trust and you know
smart people, and that maybe that it's more about trying to bring
businesses together.
MR. BURNHAM: One of the possibilities,
there's room for small pieces loosely joined -- and this sounds
really bubbly -- we can get back to the CGI or CMGI, but is there
some room, and obviously there are some lessons learned about the way
you build synergies and things like that, but if you think about very
like -- so there's -- wait. So there's no synergies. So then you're
making argument that there is no value, but I'm trying to follow
where you were.
MR. JARVIS: I'm trying to say what happens if
you're a VC fund, do you have to come along and say shit, for me to
really play, I can't make that bid, just like an add buyer
(inaudible), you can't find that one big --
MR. BURNHAM: Right.
But one of your proposals was there's a possibility of kind of a
social network around the capital that creates value.
MR.
JARVIS: You get your fingers into more pies, you share a risk and you
share knowledge.
MR. O'REILLY: He's going to send you a T-shirt
that says "Fuck you, I don't need a VC."
MS. FISHER:
When I was at 2.0, these are young people who learned from Google's
success. They said the big thing they took away is that don't work
for VCs. It's just not a good thing. I think it's easier to start
VC companies that way. I think it goes to something which I was
going to tell John that there were -- I feel like there is a lack of
heart in data, and I don't know if that's just there's a lot of easy
things to be done. It's just judgment. I wonder where the hard
problem being solved is.
MR. HEIFERMAN: There is no long tail of
household words. Meaning -- like out of the whole boom, there's a
few words that 80 percent of the world knows. Ebay and Yahoo and a
whole big chunk of them know about dotcom. That's the thing about
you look at the list of Web 2.0 companies and projects and sites, and
it's all this explosion of words and things and things that no one
will ever remember, and no chance that more than one percent of the
population will ever be able to digest and make a part of their
vernacular. And this isn't about sort of brand and trust, it's
literally about how can human beings process this stuff. And this
may not sound exactly like it has precisely to do with what this
conversation is, but I actually think it does, because there is no
long tail for new household words, and household words are like the
currency of real value, because you have to be the go to place.
Craigslist value is driven by the fact there's a go to place called
Craigslist. If it was just that you could go to Google and see this
smattering of listings of real estate, that if Google is the ultimate
go to place for everything, then the system doesn't function well,
but there are these go to places.
MR. PAREKH: But Skype is a
virtual peer to peer go to place. There was no Skype --
MR.
HEIFERMAN: When I say "go to place," I mean that people are
starting to say I'll --
MR. BURNHAM: Let's get Heather in here
and wrap this up.
MS. GREEN: I wish Clay was here. I was
talking to him yesterday, and I was trying to find out what's going
on with kids, right. Myspace. What the heck is going on with
Myspace? I went to Clay and I said to Clay, "Clay, you work with
kids, do you know what's going on?" He said, "I can't know
what's going on." He said, "When you were in like '99,
2000, you could. You could know Ebay. You could know all these
companies because there's a finite number of people and you could
find stuff." What he talks about now is there's so many people
online and there's so many different kinds of places they go, there's
not going to be household names in the way we thought about in the
past. Craigslist comes up over time. We know it now, but maybe over
time that's what happens, but his point is that there's all these
different kinds of things that users of the VCs that can probably
find a lot faster than anyone else.
MR. BURNHAM: I do want to
wrap up this kind of part of the discussion because we have one more
topic that we want to touch on, and we are actually rapidly
approaching the time that we committed to break, and I would like to
keep to that commitment. Get people back to the real world here. I
think if I hear what Scott's saying and what Heather's saying and
come back to something that Dave Morgan said is that ultimately, that
there's a limited amount of attention, and that that attention is
going to be directed to a limited set of household names. Those
names are going to represent brands, and the tricky part is if we go
back to 1.0, everybody spent money building brands. I think we now
figured out that particularly when the brand needs to have something
to do with authenticity, you can't just buy it. You have to grow it
in some other way. So it's not a slam dunk. But I think there
probably will be a way of managing attention around concepts that are
linked to household names, and that there will be -- back to Martin's
164 million to one scenario -- we will see a very steep curve in
terms of what's successful and what's not. The one other thing that
we wanted to talk about briefly, and I want to just try to get
everybody's ideas on this, is does the current -- this is the last
topic in the agenda -- but does the current regulatory frame work
help or hurt the evolution of I would say the consumer -- the
potential consumer benefit from the services that we're talking
about, the peer produced services?
MR. EVSLIN: Did you mean that
to be a US centric question?
MR. BURNHAM: My mind is mush if I
try to get outside of the US, but I think thinking about -- I think
it was Mary early on who introduced the idea of China as being a
place that you can't control, therefore, the regulatory frame work is
irrelevant, as long as there is a place in the planet where different
models are emerging. So I don't mean it to be US centric. Just is
there anything that we can be doing? Do we as a body of people who
have an interest in this have an obligation to be voicing an opinion
about, you know, about the way copyright works in this scenario, the
way patents work in this scenario, the way any other legal frame work
impacts the development of this market.
MR. WILSON: Brad, let me
take a shot at that. Earlier this week I went down to the 11th floor
where Joshua runs his business, and the only guy in the office at
that time was this guy Dan, who I don't know, how old is Dan, 24, 25?
MR. SCHACTER: I think so.
MR. WILSON: I said, "Hey, how
you doing, Dan? He said, "Good." He said, "I got a
question for you." He said, "When you post those MP3s on
your blog, is that legal?" I said, "No." He said, "Why
do you do it?" I said, "It's a political statement."
And this is the political statement I'm making. I've posted my
policy on how I post MP3s to my blog, and the policy is this, I
encode them at 96 kilobits a second. I link to Amazon or some other
place that you can purchase that music legally, and I basically do it
to Amazon, because through Amazon, I can actually get a report back
on actually how many people do link through and buy it. And if the
RIAA ever comes after me, I will elevate this as high in the law of
the land as possible, because I believe I'm doing the artist a favor
by that policy, and I believe I'm selling more music, and I think
that's what needs to happen in this country, is people need to take a
stand, and fuck what the law says, let's create a new law.
MR.
O'REILLY: This is very related to the Google library. I really
believe that, you know, opt out that's the only viable option. If
you look at the music industry, it said, "We cannot clear the
permission." They are waiting to be rescued from the problem of
too many property rights. Book publishers are waiting to be rescued.
They just don't know. The fact is, we have built so many rights
into the system, that you cannot act unless somebody cuts the
guardian off.
MR. JARVIS: The big guy at the last lunch -- the
last lunch I went to --
MR. WILSON: There may have been some
since.
MR. JARVIS: The big guys can't do anything. Somebody's
got them on hold. It's either the MSO or the talent agent or the
talent. So it's the little guys or the outlaws who are the avengers.
MR. D. MORGAN: I think you can take the risk and have the one
out of the 164 million and be socially legally disruptive. The
problem is as a business, you can't. I mean, at the point you decide
this is going to be the business, I mean, Michael Roberts and others
have to say, "Okay, I've got to play by a certain set of rules."
And then we'll just have to look to regulatory structure and say it's
really problematic, and unfortunately, it's also skewed towards those
that don't want to be disruptive, because they control the regulatory
structure. They control the legislative structure. The ones with
the money can create inertia problems. And I don't know that it
necessarily helps our industry to be the ones that are going against
the policies all the time, because it actually feeds the forces that
are trying to stop it because it plays well in Washington. You can't
trust these people because they don't obey laws.
MR. WILSON:
But, Dave, the mistake we can make as business owners and business
operators is to pay too much attention to those entrenched players in
the marketplace and try to serve their interest, because at the end
of the day, they'll be an anchor that will sink your ship. What you
have to do is you have to at least recognize the only wind in
anybody's sail is coming from the people, you know, the ants. This
is a revolution of the ants. That's the people that need to take you
forward. And the big dinosaurs are going to sink your ship.
MR.
JARVIS: When are you running for the Senate?
MR. WILSON: No.
MR. MORGAN: But, Fred, your blog's not a business.
MR.
WILSON: I'm not talking about my blog, though. I'm just talking
about you as a business might be better off -- I don't want to argue
your business -- but you might be better off providing services to me
and 1,000 other people like me who have the wind in my sails than
Martin and his crew.
MR. D. MORGAN: I don't disagree with that.
I just said that I think -- and this is maybe more personal policy
thing, which is you get to the point where there are clear laws, and
I think that -- I don't think you can say, A, I'm going to directly
violate those laws and still say I'm operating a business because I
want people -- or to do things where I know that people are going to
break laws, but I'm not really going to ask the questions. I do
think that, though, that you got to work with the small pieces,
because I do think that's where the wind is.
MR. BURNHAM: I
think we're talking about something that isn't clear. That isn't
black and white. Yochai.
MR. BENKLER: I want to push Fred a
little bit in an awkward situation where I will stand as a law
professor for the first time in my life. I'm not sure that your
marginal contribution is best being putting on the MP3 with a very
clear policy as opposed to using the fact that unlike the millions of
teenagers who can do exactly what you're doing, you're also a venture
capitalist who knows lots of people who are in the business, and can
speak a language that the political system understands, which is
here's the value we're preserving, a whole 11 billion dollars
recording industry, that's the whole recording industry, 11 billion
dollars a year, that's it, versus let's take a look at the IT
industry's somewhere on the order of an order of magnitude or more
larger, and say where's the direct? We're saying the new value is
coming from people interacting with and around the information as
freely as possible, creating new sources of value, which is where our
relative advantage is, and in this regard, I agree that in the sense
national is the wrong way perhaps to think about it, but the
political economy is such that Congress thinks of US jobs and the EU
thinks of European jobs, and an argument about relative advantages of
various economies is important, and in this regard, I think, putting
together a coalition of people who are investing in this area and
intervening in the legislative process, rather than saying the ants
will win, is much more powerful, because there are lots of ants that
don't have the advantage you have, which is speaking credibly in the
name of business.
MS. GREEN: You have to run for Senate.
MR.
BENKLER: You don't have to speak.
MR. O'REILLY: I have a comment
on that, having done some lobbying in Washington. Unless you're
prepared to make that a full-time job, you're spitten to win, because
there are people who have that full-time job, and you know the people
in Washington know that you're visiting, you're going to be gone.
MR. JARVIS: And we organize an association that pays for --
MR.
O'REILLY: Absolutely.
MR. BENKLER: So, today, the Freedom Law
Center is being launched by and funded as essentially a standing
public interest law firm in favor of the free software with funding
from some very big players who understand that that's important.
Clearly you have to be there, but let's remember, the recording
industry and Hollywood are two industries that are optimized around a
particular regulatory system called copyright that they have
specialized for close to a century in lobbying around that, and of
course they're going to be pulling against, because everything we're
talking about kicks them in the business model. And so the point of
actually getting together and saying, well, they're using a
particular set of regulatory systems to undermine our business, one
of the things we need to invest not individually one firm, but to
begin to understand that the whole cost of people who are looking at
these kinds of technologies need to be able to generate is something
like the Freedom Law Center. Something like a standing body that
represents the interests of those people who as a business want --
MR. HEIFERMAN: That is both effecting and actually directly about
what I think we're talking about here. This concept of peer
production, production -- I think production isn't just content and
web pages and stuff like that. I'm making one of these long bets
that says that one of the biggest impacts of technology over the next
10 years will be something that really doesn't exist yet, which is
collective power, collective action. I'm betting that half of the
Fortune 500 10 years from now will have their business drastically
affected by the ability for consumers and people to organize via the
net. And that's going to be, you know, if five million Walmart
customers signed an active pledge that they would reduce their
Walmart spending by 10 percent if Walmart does not give their
employees good health insurance, Walmart will listen. And because
they'll listen -- well, they will listen, and you'll say, "Well,
why would five million people sign that pledge?" Because the
social network of the million Walmart employees is easily five
million. Point, bottom line here is that peer production isn't just
production, it's peer power, and so this kind of stuff, this ant
stuff is really like the net itself is going to fuel an incredible
amount of people power that we've never seen before I believe.
MR.
WILSON: Marc, this is your soap box, so stand on it for a second.
MR. PINCUS: I tried to start something at the same time
(inaudible). And the idea was to create a web-based platform that
would enable revolutionary ants, and the idea was that it would be
like the people's lobby, and if you had five million people who all
agreed to put money or votes into things they cared about, you have
the power, and I learned quickly, people care about issues, not big
esoteric things like their own power and move on. You've heard of an
e-party cabinet. But I do think totally believe it'll happen, but
what's interesting me, I was one of the founding investors in Napster
and one of the ideas I pushed to Hank Berry or before Hank Berry that
they rejected and went the legal route was let's say "fuck you."
Instead of trying to work with the RIAA and everybody, why don't you
just put something out on the front page of Napster and tell
everybody if you want to -- here's all the list of people who are
being bought by RIAA, and if you want to contribute to all their
competitors click here, and just go to your 80 million people and
just ask them to donate money to make what they're doing legal, and I
do think if any of these guys now built that in, I think you would
get an incredible amount of money, and I do think that money is what
buys politics more than votes or anything else. So, I think it would
be -- I agree with you, and I'm interested to hear what your long
term thought is.
MR. WILSON: Mary, you had something you wanted
to say.
MS. HODDER: Oh, I'm building a remix community for
video, but anything that can get plugged into video.
MR. JARVIS:
She likes VCs.
MS. HODDER: I do like VCs. I can build for
1/10th of what it would cost five years ago, so. Anyway, so I think
that there's some place in between what Tim was talking about and
Yochai was talking about. You know, Viacom has 14 people full time
in Washington, Viacom Digital. We can't compete with that. And I
also don't think that -- I mean, I have blogged for years about how
all these people with top down contents should be doing something
different than what they are doing. Or they shouldn't be suing
whoever they are suing. I use MP3s that are copyrighted in the
videos that I remix and put on my blog, and probably will remove
those before I go live because I don't want to spend two days a week
at depositions like Hank Berry did. Bloggers often made their own
content, and so many eyeballs followed along behind them, that
newspapers had to pay attention. I think we can do this with rich
media. So if we make it easy for people to upload their digital
cameras and, you know, cell phones to upload snapshots of video to
the web and play with each other's stuff under creative commons
licensing, so that it's for noncommercial use and figure out the
issues around remixing for that kind of content, if enough eyeballs
go away from top down content, then those top down producers are
going to pay attention, because they're going to realize that they're
losing a percentage of their audience to something else, and then
they're going to have to figure out how to play with that world.
MS. GREEN: They might do that. (Inaudible).
MS. HODDER: No.
No. But the point is, if users are remixing each other's stuff,
they're completely out of the loop.
MS. GREEN: I agree with what
you're saying. I think you should do what you're doing. I think
that's great (inaudible).
MS. HODDER: But, I mean, New York
Times started paying attention to blogs when enough people -- when
they were having enough traffic driven to them through links, and
also, when they realized that enough people were reading other stuff
and were just going to simply be lost if they didn't figure out --
MR. WILSON: So, Cyril, you were going to add something.
MR.
HOURI: Yes. I was wondering why there hasn't been a way to work both
sides of lobbying. We're talking about companies like the 15
lobbyists, and you know, I wonder if people's votes are not more
powerful than the lobbyists making cases in Washington. And I
haven't seen really that happening. Like people saying to like
congressmen or senators or whatever, "Look, we're against this.
Even if AOL Viacom tells you that it's a thing you should do, if you
do it, we won't vote for you." And I'm wondering why.
MR.
BURNHAM: So we need a peer produced lobby.
MR. EVSLIN: I think
the answer to that is there isn't any mass movement that thinks
Viacom is doing anything wrong, because consumers don't understand
the intricacies that we understand. There is a Web 2.0 effect.
Every campaign that is getting going now understands the value of the
internet. The Bush administration is very effective in using the
internet to reach its faithful. So it's not that the technology
don't have any affect, but it doesn't split along ideological lines
at the level that you're talking about by the people who are the
consumers. I think Marc's over optimistic in thinking that Napster
could have raised a lot of donations to be channeled to the opponents
of the enemy, because people are disassociative, and it has to do
with what you believe is credible where. You know, much as Jeff
Jarvis' point. So you think Napster's great when it comes to music,
but it's not associated with politics. So you see that you ought to
contribute to this campaign, and I'm not talking about us now. I'm
talking about the downloaders, and you just turn it off. I don't
want music. I don't want politics. So I don't think we have as much
power there as we think we have.
MR. JARVIS: So given the three
choices, competition, votes or lobbyists, what's the best path, those
three things?
MR. WILSON: Competition. Get in the market place
-- I mean, I'm with Mary. Get in the marketplace and just beat the
crap out of them.
MR. EVSLIN: If Chris Anderson were here, he'd
point out that he did speak for the Churchill Club where he talked
about how he's being competed with by the long tail, and he was the
head, and he's got wires and he's got two million subscribers. He's
got Cande Nast behind him. And there are amateurs who can do things
better than he can do in specific areas. Since there's much less
competitiveness now for those amateurs doing their things, and the
amateurs doing it out of love or what many of the reasons we talked
about, I think what happens in the end, that's why Mary's directly on
point, that there's enough talent that was buried by the hit system
-- not the copyright system, the hit system -- and the hit system was
a function of distribution, and as the hit system disappears and the
distribution gets better, then we get effective contribution from the
tail, and in the end, the copyright holders can lose by winning.
They can win their regulatory battle and be left with a lot of
worthless property, if you want to be an optimist, and I am, because
Mary makes it possible for the tail to produce the content that
people want, and the issue simply doesn't exist anymore.
MR.
BURNHAM: Doesn't look like you're optimistic.
MR. BENKLER: No.
No. It's not a question about optimist. The sense that these are
either/ors as opposed to what are the set of strategies and what are
the relevant competencies, right. Clearly, you compete -- you start
with competing. That's the existence proof, and that's the source of
funding, but then the question is, you know, I don't think that
there's no social movement around IT. I think creative commons is
first and foremost a social movement and only secondarily a set of
licenses and a set of practices. I think what we're beginning to see
are the international trade and IP regimes. You're suddenly seeing a
very interesting alliance between developing nations, some of the
NGOs, and players like Cisco, who are concerned about excessive IP
coming together and beginning to lobby. So the question that I
thought I was answering was what is the relative advantage of people
who are concerned with the business side of the emergence of peer
production who understand the excessive -- who understand the drag
that the regulatory system called copyright exerts on the whole
amount of value. What is your relative advantage in this system.
And one is clearly actually competing, but the other is also speaking
with a particular voice that the volunteers and the peer producers
out there don't have. Sure, they have the potential votes, but what
they don't have -- and this plays an important role in DC -- is the
credibility of this is a business position as opposed to an anarchist
position, and that's an important voice that needs to be articulated.
And in the area of spectrum, there are all sorts of activists
talking about free spectrum, and then around 2001 or so, Intel and
Cisco and Microsoft suddenly stepped in and told the Commission, "We
can make a lot of money if you make more unlicensed spectrum
available." And as it turns out, that changed the way the
Commission saw things, and we got a lot more spectrum and a lot more
focus on open spectrum policy. And I think crystalizing that as a
pro business deregulation, right. Scaling back copyright from
businesses is a pro business form of deregulation would be an
enormously important means.
MR. WILSON: That's very good. We
need to wrap. I want to just sort of tie this back to where we
started. Rikki said that he was looking for the one plus one equals
three. In the old model, if you aggregated, you got a certain
aggregate set of data. If you had two people producing content, you
know, you'd get the same aggregated set of data. Was there some
magic that came out of the fact that the content was peer produced
versus aggregated in some other way, and Seth Godin said that it's
really a marketing technique, and that, you know, engaging the peer
producers in the production of the content turns them into advocates
of the content, and they then have equity -- he didn't say it this
way, but I interpreted it -- they have equity in that product, and
they become promoters of it, which is free marketing. He then went
on to say that this event was for me and Brad, and I'd hope that he's
wrong about that in the same way that I hope that one plus one does
equal three, and what I'm hoping is that since this was a peer
produced event, that each person who came and contributed walks away
with more than what they contributed. I certainly have, and I hope
all of you have too.
MR. BURNHAM: We have one more housekeeping
detail, and that is to thank our hosts. And I'd like to have Jonah
just say, very briefly, who Eyebeam is and why you invited us.
MR.
PERETTI: Eyebeam is a nonprofit hardware technology organization, and
one of the main reasons we were interested in having this here is
because we are launching a new lab, we just got funding, and the idea
behind the lab is that there are certain things that businesses can
do better than a nonprofit and there's certain things that work
better in a nonprofit environment. Instead of just thinking of
nonprofit work as being something like, you know, does good things
that the market won't take care of, we wanted to develop this idea of
doing actual productive work that adds value, and that's partly why
this concept of peer production is interesting to us. So this lab
through the glass there is the new Eyebeam open lab, and starting in
November 4, we have four fellows who will come and they'll develop
work for a year, and all the work they will do will be under open
licenses. Some of it will be media based. Some of it will be
software based. Some of it will be hardware and physical objects.
It says reading, printer, laser cutter, and the idea is that the work
that happens in this R and D lab will be public domain R and D, so
it's something that people can develop. Take with them. Say here,
it can be shared. It can be developed and distributed by a network
of people beyond Eyebeam. So we're starting to look into these
issues and how to do creative R and D.
MR. BURNHAM: Thank you
for your space.
MR. GOLDSTEIN: I think it's appropriate that we
had it here too, because Del.icio.us is the Web 2 poster child for
New York, and I think it's a great contribution to the Union Square's
portfolio. I met Josh who I introduced to Fred and Brad through Mike
Frumin.
MR. PERETTI: And Mike Frumin is the technical director
of the new lab.
MR. BURNHAM: And, also, for those of you who
don't know Mike, Mike is the guy that created the election database
by Geography and put that up in the rafts, so it's another interface
on top of the data set.
MR. WILSON: I've got one more thing. We
do want this to be more open than just what happened in this room,
and Eddie has been busily transcribing the entire thing, and I've
been watching. It's just amazing. I could never do what you do,
Eddie. This is going to be on our Wiki at usv.jot.com. All of you,
because you were invited to this event, actually have that URL
somewhere in your databases. Just click on the public site of the
Wiki. There will be a link to Sessions, and everything will be in
there. Mark, who's been photographing, we're doing it all digitally.
Hopefully by sometime tonight or tomorrow by the latest, we will
have taken those photos and put them up on Flickr. They're there? On
Flickr? What's the tag? Linked to it from the Union Square
Ventures.com web site -- blog, correct. And what we're going to do
is if anybody blogs anything about this event, we are going to tag it
in Del.icio.us with a tag USV Sessions. There's already a pretty
nice group of links there with Yochai's work and some of Tim's stuff
and some of Jeff's stuff. Umair, who's been very quiet -- put a plug
in for Umair -- he is doing some of the best thinking about this
stuff that we've come across. There's a bunch of links to some of
Umair's stuff. Tom's debunking of Reed's Law is on there. So
anyway, contribute. Blog. Keep the conversation going. We'll try
to pull it all together and keep people focused on it, and thanks for
coming.
MR. EVSLIN: Can I ask one question, in the interest of
opt out, is there anyone who doesn't want anything they said
attributed to them?
MR. O'REILLY: I think Scott probably doesn't
want to be quoted as saying Craig's a weird guy.
MR. EVSLIN: I'm
happy to have anything attributed to me, but just in case somebody
doesn't.
MR. WILSON: If anything's on the transcript that you
find objectionable, let us know and we will censor. Thank you.
(TIME NOTED: 3: 10 P.M.)
CERTIFICATION
I, Edward Leto, a Notary Public in and for the State of New York, do hereby certify: THAT the witness(es) whose testimony is herein before set forth, was duly sworn by me; and THAT the within transcript is a true and accurate record of the testimony given by said witness(es). I further certify that I am not related either by blood or marriage, to any of the parties to this action; and THAT I am in no way interested in the outcome of this matter. IN WITNESS WHEREOF, I have here unto set my hand this 21st day of October, 2005.--------------------------- EDWARD LETO