I'm just on the train back following the London part of the Head Conference, and am going to tune into some of the talks over the weekend (you know, instead of working). What's cool about the Head Conference? Well, I can point you to a blog written by someone in Indianapolis to give you a roundup of today's events. 
For me, I quite enjoyed Tim O'Reilly's talk, especially his discussion of how he's more interested in the non-commercial origin of what later becomes commercial technology. I have to agree: "we take user content, stick it up on the web and put adverts all over it" is not very interesting. Building a free operating system is. Building a semantic web is. I guess my almost chronic lack of interest in how one can turn a quick buck from hanging around on the Internet and building cool stuff is one of the reasons I'm shooting for a career in academia. I think Tim plays an interesting, chameleon-like role - he's a hacker amongst hackers, an entrepreneur amongst entrepreneurs, a VC/biz-analyst around them and so on. Still, he uses Vim, which pretty much settles any disputes. (Oh yes, the cult initiation is complete. Esc-7dB-i. No it's not! There are no thetans to be seen here.)
I have to say I wasn't very impressed with Simon Wardley's discussion of cloud computing. Everyone seems to have enjoyed it a lot. I laughed at a few of the lolcats and memetic mental trips, but I have to profoundly disagree with the thesis. The more I hear about cloud computing, the less I'm impressed. Aral's experiences with Google App Engine are case enough: if my provider goes down, I don't have to resort to nepotistic pleas to get it back up again. My site is pretty portable: it's written in PHP, I can roll it up into a tarball and move it. The Java web stack has WAR files. You can find Rails hosts for anywhere between $10 a month and $10,000. It's never really been a choice between the 'cloud' and running your own servers: hosting companies exist. Quite why writing an application that runs on the Google APIs, calling out to Google's proprietary database and so on is supposed to be more portable than generic Python or Ruby or Java or PHP or whatever eludes me. And with consumer cloud services like Google Docs, I tend to try them out and then promptly come back to using software running on my own boxes. Software as a Service? No thanks. Software as air please. Software too cheap to meter. Software that can't be tied down - it just floats around, free like the wind. Software as a Service imagines taking a big raging fire and slicing it up into it's molecules and selling each of them through micropayments.
Jeremy was on top form - asking a real difficult question. How do we build a long-term web? I hate to sound like a broken record, but I think RDF is a possible piece in the solution to this: the ability to annotate resources means that we could piece together resources in a distributed way. Think about it this way: all we have of the original New Testament gospels are versions. Scholars differ as to the original: most generally think that Mark was the original Gospel, while some believe it was Luke. The others emerge from it. There are elaborate theories about the writers of the Gospels: look up Marcionism, Gnosticism and so on. Annotations have this kind of role too - imagine if in a hundred years you were trying to figure out what was thought about a newspaper article that you know the URI of but nothing else. The URI might be tremendously useless - just the paper and an ID number. But you scout out across the web and come across lots of machine-readable annotations - think 'rich tags' and metadata. You piece this together, run each piece through inference processes to detect the inconsistencies, and use some probabilistic reasoning (something like OpenNARS), and so on to give you some rough ideas of where to look. Eventually, you find out enough about the URI to figure out whether or not it's worth pursuing, and if it is, you could go through whatever convoluted process there is to get hold of it. (This is, of course, the last resort: in an ideal world, you'd simply issue an HTTP GET request and you'd get the resource.)
I digress. Chris Heilmann and Simon Willison gave good talks, and there were two panels: one on accessibility testing and the other on the state of Flash. Also: the venue - the Magic Circle - was awesome. I was hoping that Aral was going to perform some kind of Penn and Teller-style blood-and-gore magic act at the end. Even though that didn't happen, Aral did an excellent job of MCing and organising what must have been an administrative nightmare of an event. A round of applause is due!
