Hutter Prize expanded by a factor of 10 (prize.hutter1.net)

It would have been better to go from CPU to GPU for compression/decompression, as it can get a better result for the same price, but on the other hand it's understandable that the prize wants to be compatible with earlier solutions.

That's amazing. I remember back in 6th grade, thinking I could take a swing at this... 9 years later, not even a dent made. Maybe one day.

I guess part of it is that if you ask the average person, even if they have ML experience, they might not even know where to start.