Elias presents ... a worm!    Thoughts on family, philosophy,
and technology


Wednesday, June 28, 2006

Software is not war

Ted Neward essays on the object-relational impedance software problem, needlessly drawing out his now famous one-liner that it is "the Vietnam of computer science." I don't like the analogy for two reasons:

1) Bad taste. In the Vietnam war, over a million people died, so seriously comparing it to the tensions in software system design is sophomoric, inappropriate, and silly.

2) Vietnam was a catastrophe; software talking to RDBMS's is by and large a success story. There are thousands of working software systems which have objects interfacing to relational databases. At most we are talking about the trade-offs, tensions, and shattered expectations experienced in this domain. Though there is no silver bullet which makes the impedance problem go away, it is not intractable, we have several general solutions.

Tuesday, June 20, 2006

Soccer robots

This contest hopes to incubate robotic soccer players that can beat professional human teams by 2050. The videos this team supplies give a flavor of the state of the art. Slow to download, and kind of boring unless you are kind of geeky. But it is impressive to see a penalty shot showdown: one robot aims to one side of the motionless goalie, swings its leg, and kicks a goal, meanwhile the goalie takes a belated "dive" (aka, he falls over sideways) for the block.

The most interesting thing to me is the timetable they have set up: 2050 for humanlike players. I'm guessing the general software problems here are already understood, so this number is mainly needed to allow for 44 years of the vast hardware and interface improvements required. Even as hard as the AI problem of artificial professional-grade soccer may be, it is an extremely delimited problem.

Now contrast this to those "Singularitarians" who imagine we'll have general purpose, human-like AI within 30 years. Let me just say: if all you've really got is a graph with a curve on it, such a "prediction" cannot be described as knowledge.

Friday, June 09, 2006

A beautiful mind

I recently learned of this site, full of high definition images of famous art, free for personal use.

Leonardo Da Vinci's Mona Lisa must be viewed in detail to appreciate. I had always thought the painting wasn't quite as beautiful as claimed, until I saw this. I still think the woman would be considered homely if met in real life. But Da Vinci did not paint a face, he inexplicably captured something far beyond that: a mind in an active moment of health and happiness. It is incredibly beautiful.

Tuesday, June 06, 2006

Google vs. Microsoft: Not just yet

As usual, Nicholas Carr is the smartest guy in the virtual room, explaining that Google's new spreadsheet app does not and is not meant to compete with Microsoft Excel. Rather, it is just one more service meant to drive traffic into the Google advertising universe.

Microsoft's Scoble sees it as competition, but he is also able to see that a bigger battle is just around the corner, in what I would like to call the post-HTML web. I predict the real battle is not going to be Microsoft's .NET/WPF/XAML versus Sun's Java applets versus something new from Google, as Scoble suspects. The last thing Google wants to do is engage Microsoft in an API war: Microsoft will always win those so long as they own the desktop. Instead, the Microsoft API is destined to become the de facto standard with the ascension of Vista next year, and then we will witness the first battlefield on which Google and Microsoft meet face-to-face: Microsoft Live versus the Googleplex. As I've said before, my money (rhetorical!) says Googlenerds have been secretly developing WPF apps for some time.

Update: A quick and useful run-down of what the hell is WPF anyway -- although the author didn't seem to know about WPF/e, which will be a watered-down version of WPF for running in non-IE browsers. Extra points for paying due respect to Charles Petzold, the star author of Windows programming books, before such training got confused with being "cool."