09-25-08 - 3

Compression's really fun to work. For one thing the algorithms are fun, the theory is complex and mathematical, but the practice involves a lot of trial and error and hacking. It's a nice intersection of science and engineering. And then you have an absolute metric to test yourself on. In a lot of fields I find people going "oh yeah our method for this and that is awesome". Is it? Really? Did you test it? Against the known best algorithms, on the exact same data that they used, with a non-subjective measure of performance? No, you probably didn't, and in fact your method probably sucks bawls and you're just kidding yourself. Compression lets you check your numbers and there's no way to hide.

No comments:

old rants