--- In decentralization@egroups.com, "Dave Winer" <dave@u...> wrote:

>
> TCP IS ALREADY P2P.
>
> There's nothing more to do.
>

The first statement is definitely true. The historical net has all
the facilities for p2p built-in, and it's why p2p keeps emerging.
But there's a *lot* more to do. For example, as Larry Lessig has
pointed out in many public talks (e.g. at PC Forum), the old internet
"end to end" architecture is morphing under a lot of pressures, from
firewalls and NATs and dynamic addressing to web architectures that
treat the web as a passive medium.

If you look at the hyperlink, one of the fundamental elements of the
web architecture, you would conclude that the web is a p2p toolset.
Anyone can link to anyone else. But there was a fundamental flaw in
the p2p web as it was actually deployed: browsers and servers were
decoupled, and the power to publish was concentrated in the hands of
a relatively small group of people. This is why at O'Reilly, we
developed and released our WebSite product line in 1995 with the
message that "everyone who has a web browser ought to have a web
server." WebSite was the first windows based web server, and
was seemingly an odd departure for a company that up to that time had
been largely UNIX-centric. But we saw that the thing we liked best
about the web was threatened by an architecture in which 90% of the
people using the web were on windows-based browsers, and didn't have
any opportunity to publish.

But even though we made WebSite cheap and very easy to use, we didn't
get it quite right. Napster's breakthrough, in my mind, was to make
the "browser" and the "server" into the same program. Efforts from
editthispage.com to wiki are an attempt to retrofit the web as a
writeable medium, but the asymmetries are so deep at this point that
the web has lost some of its punch.

As Clay Shirky has pointed out, dynamic addressing, firewalls and
NATs have also turned a huge class of web users into second class
citizens, having to go through their IT department if they want the
power to distribute content on the web.

What's more, as the web grew in size and complexity, a relatively
small number of search sites became gateways to the content that has
been published. This may be the nature of the beast, but I'm
interested in architectures that distribute the ability to find
content as well as the ability to create it.

In looking at the history of the net, I see a series of swings
between centralization and decentralization. Someone comes up with a
decentralized architecture, and a whole lot of other people start
working to centralize it again. This may be OK, a dance, an
evolutionary cycle, that is completely natural. But I don't think we
can say that the forces of decentralization will always win, as so
many of the early partisans of new decentralized architectures always
do. We always see them as liberating, but then they morph back into
what we were trying to get away from, and need to be reinvented.

Anyone who hasn't done so should read Larry Lessig's book, Code and
Other Laws of Cyberspace, which provides an extremely useful
framework for thinking about these kinds of issues.