I realize that FoRKing your own bits is gauche, not to mention onanistic, but
sometimes it has to be done. From http://www.worldos.com:
=======
It seems to be inevitable that a corny song will pop into my head and get stuck there
for no better reason than that it's apropos. Which is why I have "Closing Time" by
Semisonic running in a loop in the back of my head today. "You don't have to go home
but you can't stay here" is the line I keep hearing.
Phase 1
I started the project that become WorldOS on February 1, Y2K. The goal was to work on
the, well, world OS, or net OS, or grid OS, or global brain, or whatever you want to
call that wierd unibox forming in the primordial soup of the internet circa right
now. At that time my focus was on mobile code. I figured that if you could wrapper
enough of the environment so that modules could operate on the move then half the
battle would be won. By mid March I was ready for a routing layer, and that's where
the problems came up. Standard routing methods used fixed addresses of one kind or
another, and since I was wrappering resources by mapping fixed addresses to
functional descriptions there was no way to handle the problem.
Gnutella popped up on Slashdot at pretty much the exact moment I needed it. I was
blocked on the routing problem, and reading Slashdot instead of working. I hung
around at gnutella.nerdherd.net a little, checked out slapstick documentation like
Slant> phlypsyde: DUDE! Listen up.
<`deadbeef> phlypsyde: you're stupid/.
, then joined the gnutella dev mailing list at horny.phreedom.net and started working
on a clone.
After a couple weeks the inflexibility of the Gnutella protocol became a problem. The
concept was right but the implementation left no room for extension. I decided to use
an XML protocol instead, but to modify it for Gnutella-style routing, and because
both SOAP and XML-RPC were still bound to HTTP I had to write my own. After I tossed
out the Gnutella cruft the pace of work took off like a shot. WorldOS 0.1 was ready
on May 5, 2000, my 35th birthday.
I spent May filing a patent and then released to Freshmeat and a few other sites on
June 4. At the time there was nothing like it - it was months before any other P2P
infrastructure projects were even announced. Reviewers were intriqued.
Phase 2
The scope of the work was obviously bigger than I could handle alone. I needed help,
and until the code was more complete open source hackers weren't going to start
kicking in time. There was no way around forming a company.
The first person to come on board was Brian Rowe. Brian is way too smart a guy (and
has way too good a sense of self preservation) to join such a risky project. Which
meant that I had to buy a large number of shots before he took me seriously. Next in
was Michael Casey, a StarMedia alumni whose job was to raise money and write the
business plan. The night Michael signed on the three of us went to dinner around the
corner from my apartment and talked about getting filthy rich.
As you'd expect, we weren't the only ones about to get filthy rich. On July 5 the
Wall Street Journal published a story on a tech fad called peer to peer, and by
mid-August there were P2P companies everywhere we looked.
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.
All of which is to say that we are out of money to keep developing the software. You
don't have to go home but you can't stay here.
Phase3
What's next?
We are dropping development of the Goa product and moving full time into P2P
consulting. We are among the few veterans of this field, and if there does exist
consulting business we stand a good chance of landing it.
That doesn't mean that we're not going to continue development. Only that when we do
it, it will be in a clear, strong niche with favorable network effects and a clear
revenue stream.
I hope we can also find time to document and publish our algorithms and analyses,
particularly as related to security and reputation management. This was always our
strength, and nobody else has come close to it as of yet.
Peer to peer is, I am convinced, a real contribution. Long after the hype ballon is
gone it will be a standard tool. Whether there will ever be a peer to peer industry
separate from the larger software industry I don't know - P2P is more like
object-oriented programming (which is a paradigm) than the web (which is a
specification). Changes as a result of peer to peer will stay fast and furious for at
least another year, I think.
Thanks to everyone who contributed. In order of appearance that's Jay, David, Stevan,
Brian, Michael, Petra, Bruno, Jesse, Christian and Bill. I have had an incredible
time. Truly an amazing adventure, and I am grateful.
- Lucas Gonze
Wednesday, January 3, 2001
This archive was generated by hypermail 2b29 : Fri Apr 27 2001 - 23:17:59 PDT