The "Hutter Prize" has been around for a long while, but the prize has just been dramatically increased to half a million euros (and changed the test-corpus and machine spec too).
Many years ago I messed around trying to devise compression algorithms, and was struck by how straightforward it was to take pretty much any mainstream or invented scheme and squeeze the 100MB Hutter test corpus down to about 20MB. You can get there even when you use LZ77 (behind gzip, zip etc) as your base.
Going below that 20MB was extremely challenging.
I warmly recommend everyone waste free brain cycles thinking about compression algorithms. It ought be right up there with the 'wrote a ray tracer' rite-of-passage for 'true programmers' :)
Many years ago I messed around trying to devise compression algorithms, and was struck by how straightforward it was to take pretty much any mainstream or invented scheme and squeeze the 100MB Hutter test corpus down to about 20MB. You can get there even when you use LZ77 (behind gzip, zip etc) as your base.
Going below that 20MB was extremely challenging.
I warmly recommend everyone waste free brain cycles thinking about compression algorithms. It ought be right up there with the 'wrote a ray tracer' rite-of-passage for 'true programmers' :)