Seems like #jpegxl has been written off too early, with it seemingly having better #performance that #avif https://giannirosato.com/blog/post/image.. Apparently #browser #developers have up on it due to " #complexity ", which seems like as great #excuse not to #implement it.
Just a #reminder that #sorting can be super easy (in terms of #algorithm #complexity ): https://arxiv.org/abs/2110.01111 Of course, it's not that #efficient ... But there could be a lot of #opportunity to run it in #parallel or make #domain specific #shortcuts ... As the #hackernews #comments point out, it's been #discovered many times before: https://news.ycombinator.com/item?id=287.. Could be really fun to take a look at this at a later date!
This #proposal #abstract is often looked at as a #failure in #ai - but I think it was actually a great #success http://www-formal.stanford.edu/jmc/histo.. In just two months, some of the biggest names got together and really fleshed out the #complexity of #artificial #intelligence - and made tonnes of advance in further understanding how hard the problem really is. In this way, people could make better estimates of how difficult the #problemspace really is.