Wednesday, July 25, 2007

Tim O'Reilly's Keynote: Some Livebloggy Notes

Tim O'Reilly:

Owning your personal identity/data:

"There's a race to own these things, and therefore, there needs to be a race to open them."

The Open Genome Project was doing that for genetics; OpenID is racing to keep identity open.

Of course Congress needs version control.

Tim O'Reilly said that Ruby on Rails has only had 14 contributors. I think he means COMMITTERS. Rails had over 100 contributors in 2006, when I went to Canada on Rails, and it's EXTREMELY likely that the number has grown since then, rather than shrunk.

New platforms: Web 2.0, like open source, is all about leveraging network effects.

Open source business model - Stumbleupon sold to eBay for millions, a Firefox plugin, open source, sold their user base (essentially).

Openads - open source ad network builder.

memcached - open source infrastructure.

Hadoop - open source version of Google filesystem. Extremely cool. Yahoo's getting behind it. David Filo called Tim O to talk about it (no detail! arrgh!).

Open source hardware: new frontier.

Nat Torkington is right now announcing that O'Reilly is soliciting donations for charities, just as we did at RailsConf. That was my idea! W00t w00t. (Obviously it was really RailsConf's idea. But I e-mailed Nat and got him into it.)

As thrilling as that is to a dork who actually cares (e.g., me), I have to say, I don't know if they're going to be able to do it. They really pushed it at RailsConf, and the community there was one community with a very cohesive sense of identity. They're not pushing it as hard here, and the community here is much less united. There was a lot of one thing there; there's a little of everything here.

Now Simon Peyton-Jones is talking about task parallelism. SPJ is extremely smart, and quite funny in a very dry, eccentric English way. I had dinner with him and Andy Gill (another big Haskell guy). Pretty much everything went over my head but it was very interesting. He's talking about a method of handling task parallelism (optimistic concurrency) which massively simplifies its hard problems.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.