Welcome to the first in a very special series of Web 2.0 interviews I’m
conducting on Read/Write Web. My goal is to
interview at least half a dozen people in the Web community who are building or shaping
Web 2.0 – i.e. the Web as
Platform.
My first guest is Lucas Gonze, creator of the
Peer-to-Peer (P2P) web application Webjay. Lucas was an
early developer of P2P applications and back in 2000 he created a peer-to-peer start-up
called World OS. Although it subsequently “morphed into a dot-bomb” (in his own words!), it sowed the seeds for his current
project Webjay – a P2P music-sharing website that has had plenty of
coverage in
The New York Times and other media publications.
It was an absolute pleasure to conduct this email interview with Lucas – I learned a
lot about P2P and the “decentralization of taste”. So in the immortal words of
The Velvet Underground: settle back, pull up your cushions (whatever else you have
with you). Here we go…
About World OS
Richard: The World OS website is no longer on the air, but from what I could gather on
the Wayback Machine archive of it, you were developing a P2P and decentralized network
product called Goa. Can you give me an overview of what you were attempting to build and
why – in semi-layman’s terms if possible 😉
Lucas: WorldOS was the company, Goa was the product. WorldOS was similar enough in both
goals and technology to Jxta, which it preceded by
about a year, that I’d leave the details to documentation on Jxta. In brief this was
infrastructure for P2P applications.
Richard: What kind of “P2P applications” was World OS aiming for – music? business
files? any and everything? What were the main types of files being distributed (or you
wanted to distribute) via World OS and who was your target user?
Lucas: Business files. The idea was that this was a P2P toolkit in the shape of a J2EE
component. It was in Java, the interface was almost exactly like a servlet, there was
authentication, things like that.
Internet as Platform
Richard: There has been a lot of talk recently about the “Internet as Platform”, meaning
decentralized web services and the “network effects” that come of that. The lock-in
strategy to gain users is based on data and content services, rather than software or
operating systems such as Microsoft’s Windows. Google, Amazon and
Flickr are some notable examples of this theory. Was
World OS trying to do a similar thing? I’m interested in why you chose the name
“WorldOS”…
Lucas: Google, Amazon and Flickr are only elements of a larger thing with a coherent
identity if you zoom way out. At that scale, the internet should, assuming the viewpoint
is correct, exist within something like the Gaia hypothesis. The internet OS
idea is a Gaia hypothesis for the internet.
The world OS idea is a Gaia hypothesis for all information processing entities, not
just computers. For example, traffic conditions probably have an impact on internet
weather, and so I prefer a view of information ecology that incorporates real world
systems like rush hour traffic.
The operating environment at internet scale is a different kind of animal than an
operating system. You don’t build it, you observe it, and you don’t write to an API, you
try to take advantage of your observations. So my software was not ever intended to build
an internet OS but rather to work well in the context of the existing internet OS.
Richard: How did World OS fit into this Gaia system?
Lucas: Goa was intended to be radically flexible and lightweight, which seemed to me to
be the defining characteristics of successful software in that environment.
How Goa Worked
Richard: Napster was a centralized database P2P service. I admit I have a scratchy
knowledge of P2P systems, but didn’t decentralized file lists such as Gnutella win out in the end? BitTorrent is the P2P system that I hear
most about these days (given I don’t specifically follow P2P technologies). Where did
World OS fit into all this?
Lucas: WorldOS routed via flooding, which is like Gnutella. However it used preferential
flooding, meaning that it used reputation to learn the most likely paths over time.
Richard: Can you give me an example of how this worked in practice?
Lucas: I only have a hypothetical example, since we never managed to sell the
software.
Let’s say you have a hundred people in an office and one of them, Michael, wants to
get a spreadsheet that his group is working on. The first time he does this his query is
sent by flooding. He has two co-workers, Jesse and Brian. Jesse’s desktop has a lot of
spare capacity, Brian’s laptop does not, so during the first flood it is Jesse’s machine
that returns the query. The next time Michael wants to get a file, Jesse’s machine will
be tried first, so that the extra cost of sending a message to Brian’s machine will be
saved.
How Webjay Works
Richard: This question leads on from the previous… You’ve said on the Webjay website that you don’t
consider Webjay to be a file sharing network. It seems like a very grey area. I guess I
think of it as a link-sharing network that just happens to have media files on the
end of each link. But then every time I click on a Webjay link, the media files – mostly
songs – are automatically downloaded to my computer (to the ‘My Music’ folder on my
Windows PC). So essentially I’m downloading files, whether I mean to or not. So I’m
confused 🙂 Where does WebJay fit into the ‘P2P system ecosystem’, in your
opinion?
Lucas: Webjay decentralizes taste. This seemed to me to be the next frontier after
decentralized network connectivity was fully colonized by the filesharing people, because
the decentralization of network connectivity created more centralization of taste, not
less.
The first reason is that you traverse filesharing networks by search — search-driven
navigation relies on memorable identifiers to search for, for an identifier to become
memorable requires marketing, and marketing is a tool only available to large centralized
entities like major labels. The second reason is that, when demand drives supply as it
does on filesharing networks, being known is a condition of becoming more known. The
expense to break into this system is currently covered by marketing dollars.
To decentralize taste I needed to break that cycle. I chose to stick strictly to above
ground networks because unauthorized material is cleaned out by DMCA requests and lack of
bandwidth for consumer ISP accounts. The more marketing dollars are going into an artist,
the more DMCA takedowns are issued and the more downloads there are to blow through
upload bandwidth. If a rights holder has a problem with a URL, I don’t want the URL, so
it’s convenient that such rights holders will knock down those URLs for me. Everything I
do is out in the open because open networks are, for now, naturally inhospitable to
centralized taste.
Development path from World OS to Webjay
Richard: In the Wayback
archives, you describe how the World OS project began and how eventually you stopped
development on Goa and moved into P2P consulting instead:
“Writing now six months later, while the P2P hype balloon has been
growing, the dot-com hype balloon has been shrinking. In that time we grew to eight
people, released a steady stream of updates, worked an unbelievable number of hours and
talked to more investors than I can count. We had serious deals on the table, but never
one with plausible terms.
[…]
We are dropping development of the Goa product and moving full
time into P2P consulting.”
That’s from January 2001. Looking at it now, 3.5 years later, is Webjay a natural
progression for you from World OS – i.e. is it on the same developmental path you started
down with World OS, a path which has thrown up legal and money obstacles for
everyone?
Lucas: At the time the legal issues made a big difference because they scared away
investors and customers. My colleagues in other companies doing P2P for business will
tell you the same thing — the RIAA successfully irradiated that turf, at least for a few
years.
So what’s the developmental path from WorldOS to Webjay?
WorldOS’ budget was ridiculous. Webjay is ultra lean — one guy, me, plus a lot of
help from my friends. All it takes for Webjay to exist is a server and my rent money.
WorldOS was all vegetables and no dessert. Webjay has very little delayed
gratification, it gets straight to dessert without stopping for dinner. The concept is
that, where you normally have to download and listen to songs one by one, with Webjay you
do it all with one click. It’s about saving clicks.
What about the legal issues that Webjay is designed to finesse? Honestly, if I wanted
to go for unauthorized music it would be no problem as long as I was willing to live in
an underground style. Put the server in Russia, get a PO Box in Jenin, you’re all set.
But that’s not the point — authorized (but freely downloadable) music has compelling
advantages.
Some P2P History and Decentralization Theory
Lucas: But let’s go back a bit, change the question a little, ask things differently,
because I have better stories than these to tell. Specifically I want to say how it is
that the idea of decentralization is now so common.
It’s New Years, 2000. I’m running a little web consulting company and we’re doing
well. I’ve got the money to do something else for a while, so I let the main contract
lapse without renewal, let the subcontractors go off to fend for themselves, and sit down
to do my thing. I’m just fooling around on the code that’s going to be Goa, though it’s
not that well defined. In early March Gnutella appears. I get interested in it as a
solution to the problem of ad-hoc discovery. I start working on a clone, in Java, which
gets incorporated into the rest of my code. The Napster/Gnutella/Seti@home thing starts
to break big. On June 2 I posted an
announcement of a pre-alpha Goa release, along with a tarball of source:
“WorldOS is a framework for distributed applications similar to Freenet or
Gnutella. The recent announcement of a portal based on Gnutella, Infrasearch, shows that
there are a number of useful tools that can be created using this new technology. This
framework enables the creation of many more such tools.”
I get invited to talk about my related work at an academic conference called Twist 2000, which is at UC
Irvine. The UC Irvine guys are mainly W3C affiliates; WebDAV and REST (the thesis, not
necessarily the concept) are from there. This is July 2000. About ten days before the
conference the P2P term took off via a column by Lee Gomes in the WSJ, so there is now a
word.
There’s a colloquium on what this new stuff is about. Now, back in those days we were
calling this new stuff distributed computing, not decentralized. The question came
up: what’s the difference between this new thing and DNS? Somebody, I don’t remember who,
suggested that this new thing was decentralized.
I came home from the conference. To follow up on the conversations there I founded a
mailing list called “decentralization” on eGroups, the topic of which was this new stuff.
The list became a community center for people interested in peer to peer. It took off
with the punditocracy and pretty soon
that word become the conventional wisdom as to the value
of P2P:
“All this was envisioned by our common teacher, Tim Berners-Lee, who was willing to
design a system built on links that can break. This is the key philosophy to
decentralization, a lovely term brought to us by Lucas Gonze. Don’t wait for the chaos to
end, embrace it, move on and do it again. The world will take care of itself.”
On the Legal Hassles of P2P
Richard: With Webjay (and I think with World OS too?) you’ve been careful to avoid any
of the legal trouble that plagued the likes of Napster and Kazaa.
On the Webjay website you say that Webjay is
“specifically crafted for both legality and common courtesy in a crazy environment” and
you are at pains to encourage your users to “stick to authorized music”. Is this strictly
a business decision for you, in that you don’t want lawyers to come down on you like a
ton of bricks. Or were there other factors in the ‘play it safe’ strategy? e.g. a moral
duty??
Lucas: It’s true that I can’t afford to go to court. Webjay will be history the instant
somebody sues, no matter how stupid and wrong the suit is. Obviously.
But it’s more important that the music I want to promote is music that I can share
(whether through a URL or a direct copy). Webjay is ultimately a promotional tool — it
fills the same kind of role as the radio. I don’t want to promote unauthorized music
because it forces me to choose between the golden rule and the law. I don’t listen to
unauthorized music, so I need Webjay to find stuff to listen to.
I don’t believe there is a moral duty to stick to authorized music. I do believe that
politeness is the only path to a political solution. If somebody wants me to stand on my
head while listening to their music, I will either stand on my head or find other music.
If somebody wants me to listen to their music, they will have to make it available under
terms that I can accept.
Politeness is a winner tactic. It forces the crappy businessmen in the recording
industry to stop hiding behind piracy. It makes the good guys smell serious. It’s a
dignified way of living. It helps musicians who respect listeners get popular at the
expense of musicians who don’t. The sole problem with politeness is that the technology
and culture to filter up the best music libre is still immature.
The Future of Webjay
Richard: Lastly, what’s the future of Webjay do you think? Given your experience with
World OS and the lessons you learned from that, where would you like to go with Webjay in
the next 2-3 years?
Lucas: Webjay will probably take on new features via spinoff projects, so that I don’t
break the existing community. The site does need a major makeover for usability and
attractiveness; I don’t know yet whether I’ll call that new version Webjay or something
else.
This web app is a beautiful machine. Over the next 2-3 years I will try to make more
beautiful machines. I’d really like to make a better living, but that’s secondary.
