Over the next two days I'll be posting a sort of running blog on the OSCon. I'll be posting notes on a per-session basis, then flesh them out with my own thoughts as I find time.

• OS is more than apps
• new paradigm, get usrs to contribute data
• data = value
• data lock in (!)
• challenge
  • "Napsterize address book and calendar"
  • Rethink of email and IM as social software

Talk by Robert Lefkowitz

He talked about how some of the words in the OSS world are used and interpreted (e.g. "patent" apparently used to mean "open"), to essentially illustrate how the "source" in "Open Source" isn't necessarily about "source code". His main point, I think was that "source" != "code". He argues that the specs, the designs, of software were really the source. An analogy he used was food. If you wanted open source food, the recipe is the source, whereas the "source code" would be the actual ingredients.

Talk by Megan Conklin.

This talk was about how the OSS developers can be studied using graph theories, and whether such a "network" could be seen as being a true scale-free network where monopolies occur. Basically, if you think of a project as a node, and developers as links, you can create a sort of "OS developer graph". She gathered data from SourceForge, and determined that such network did show characteristics of a scale-free network.

But her argument was that OSS projects do not achieve monopolies because there are certain safeguards against it. She named two, one was in the organization, where projects generally do not have more than 10-12 developers. The second "barrier" preventing monopolies, she argues, is the "psychology" of developers (noncomformist, personal itch, etc etc).

I more or less buy her argument, but I'm curious to know what the implications are. Okay, so no project monopolizes the developer "market". So what? What does it mean for the softwrae industry? How might corporations learn from that, or adopt a similar model? How does that correlate with user market share?

Talk by Dan Gilmore

I was rather excited to be able to see a "big name" blogger/journalist in flesh, but the talk was somewhat disappointing. He talked about grassroots journalism and what it means, but nothing particulalry exciting beyond: it's there, use it. Or if he said anything more exciting, I must've missed it.

I'm typing up all these entries during lunch break. I was delighted to see that they gave out food, but was somewhat reluctant to grab one of the dull brown lunchbags, stacked on tables in the exhibit hall, which supposedly was sponsored by Microsoft. I mean, this is one of the biggest gatherins of OSS developers... I just hope their food isn't nearly as virus prone as their software offerings.

And, there's excellent wireless coverage... very nice. Oh, and it's amazing how many people have Macs. I'd never been to a developer conference, but I didn't think half or more of the people would have PowerBooks and iBooks. There are like 40 people on the network with iChat/Rendezvous turned on. And that doesn't include this cute PowerBook toting chick sitting in front of me (either that or her Rendezvous name doesn't match the name on her nametag...)

Talk by Miguel de Icaza

Okay, so I feel like a teeny bopper chasing after pop stars, but as you may know, Miguel is another one of the Open Source "pop stars". I, of course, also have a genuine interest in Mono, although he really didn't go into too much technical details.

Only note I have is: sounds cool, must do something with it soon.

Talk by Brad Fitzpatrick Okay, this was a truly awesom talk. Brad, one of (or the) creators of LiveJournal talked about how their backend infrastructure grew from a hosted server to 90+ servers. The presentation material is available online, but here are some of my notes:

-Brad Fitzpatrick
-from 1 to 90+ servers
-mod_backhand
-after a while, replicating 1 db on multiple machines
	-> users split up among multiple clusters
-custom UID allocation (no auto_increment)
-"master-master" (switch master periodically so only one of two are used)
	-> consistency
-MyISAM vs InnoDB
	-MyISAM fast, blocks
	-InnoDB better concurrency 
-Proxies 
	-IP level:  BIG-IP, LVS, Wackamole
	-HTTP proxy: mod_proxy, mod_rewrite, pound, mod_backhand
-Perbal
-MogileFS
	-files in classes, have states, replicas on hosts (no raid), track w/ DB, cheap disks
-cache
	-DB
		-MyISAM : concurrency issues
		-InnoDB: still not as good as memory
		-parse SQL
	-where?
		-mod_perl : no shared memory space
		-shared memory: limited to one machine
		-MySQL query cache: slow
		-??
-memcached
	-http://www.danga.com/memcached/
	-distributed, no master node
	-simple protocol, xml free
	-hash key % num buckets = bucket #
	-slabs, 
-MySQL
	-connections not necessarily good

Unfortunately there was another session I wanted to go to, and didn't get to talk to him about some of the ideas I had when I was working on IlohaBlog. Of course, that was only designed to support at most 100,000, so it's an entirely different scale to LiveJournal's multi-million user-base. But since they did have a time when they had tens of thousands of users, it would've been interesting to see what he thought about the design I had in mind.

A talk by Marcel Levy

He talked about Cache/Lite, and a mechanism where images are dynamically resized then cached. He talked about an area of optimization that I won't be talking about, so that's a good thing.

There's a reception down in the Exhibit hall (i.e. free booze) and an O'Reilly book signing thing, but I'm off to dinner with Alii and her parents. In general, today was pretty fucking awesome. I got to see some famous people up close (Tim O'Reilly, Larry Wall, etc) and enjoyed a lot of the sessions. Actually going to the sessions eased some of my anxiety about my session tomorrow, and a quick chat with George Schlossengade confirmed my suspicions that his talk would concentrate on an area completely different to mine.

I haven't talked to nearly as many people as I should've, but maybe I'll have more opportunities to network tomorrow. There are some parties tonight (more along the themes of "free as in beer") but I'll probably be missing most of that (I got up at 6am, and need to refine my presentation for tomorrow).

I still need to update my post about Tim O'Reilly's keynote speech this morning, but otherwise, this is wrap for the day. More tomorrow.