WASTE is P2P IM

Related link: http://www.gonze.com/index.cgi/2003/06/01#wastenotes

• Web home for WASTE: I've gotten a bunch of queries about WASTE, generally with a focus on user
  problems like how to get a connection or FAQ issues like how to build on OS X. To help people to help each other, as well as to foster discussion of technical issues related to WASTE, I have created a mailing list and web home at Yahoo! Groups.
• Spin control successful: the New York Times adopted the argument that WASTE is IM. Hopefully this means that we won't have yet another wave of paranoia about P2P.
• Where's the OS X version? Is it really possible that it's been almost a week without an OS X port running?

The early reaction to WASTE is that it's yet another filesharing tool, and that's wrong in a way that matters.

The reason that WASTE is not a filesharing tool is that it doesn't support searching in any meaningful way. Under the hood, Kazaa et al are nothing but search engines, while WASTE clusters are too small for search. A real cluster would be 5-10 people, not even the 50 given by developers as the maximum size, and at that scale searching is pointless.

WASTE is a tool for chat and IRC, with no more or less suport for filesharing than AIM. Indeed, this is a very good reason for AOL to come down hard on the project. AOL's strategic leverage is that it has the largest base of instant messaging users, and hence is the easiest way to reach somebody over instant messaging. WASTE not only ignores the AOL Instant Messenger namespace, it sets up a new namespace, and the decentralized nature of that namespace means that
no provider -- not Jabber, not MSN/Passport, not even AOL/Nullsoft -- can get a strategic edge. If the WASTE namespace were to to take off, the new mega-namespace established by the Microsoft/AOL truce would be obsolete.

WASTE is scorched earth for AIM.

As with a lot of P2P, WASTE represents scorched earth for companies that want to own a namespace. If you want to use centralized IM, you have to pass ownership of your IM identity to a service provider like AIM or Passport. But if you don't want to talk to everybody in the world, just the few that you actually know, and the overhead to whitelist your friends is low relative to the length of time that you'll stay in contact with them, there's no reason for identity service providers. In WASTE, nobody owns your identity but you.

WASTE also bypasses much of the need for presence providers to interconnect people behind firewalls. If any member of a WASTE cluster is not behind a firewall, messages are passed through that node. In my WASTE cluster, for example, there has always been at least one node not behind a firewall. There has always been at least one broadband node, and there's no incentive to be a free rider, since clusters are formed around meatspace relationships. Most people are happy to help out their friends, as long as it isn't too much of an inconvenience.

WASTE is nothing new.

Gnutella was fundamentally new in that it gave slacker developers a way into autonomous networks, which were ivory tower stuff at the time. WASTE introduces no new ideas -- everything in it has been shipped already in Groove, and WASTE is missing a lot of things that Groove has.

The only thing new about WASTE is that it works. Groove is too bloated for normal use. If you need Groove's quality of service, security, and replication, there is no competition. If you don't need all that you're better off with WASTE. Fair enough: Groove is for enterprise workgroups, WASTE is for friendnets. WASTE is worse. Which is better.

TODO

There are two big things that could be improved.

1. It's not worse enough. There's still a bunch of junk that can be factored out -- we'll know it can get no smaller when there's a Javascript version. The Unix console code doesn't have a bunch of functionality, but it works, and console versions imply good things about simplicity. Still, there are console versions of SOAP, so that by itself doesn't prove anything.
2. Poor support for transitive relationships. You automatically get a connection to a friend of a friend, and after that, friends-of-friends are treated the same as friends themselves. That's not good enough. There needs to be a way to keep introducers in the middle, so that names and bytes continue to be vouched for and guaranteed by the introducer.