@colin_fraser Every time I see stuff like this I can't help but feel we're just fundamentally on the wrong track here and will look back at this all as an amusing detour on the road to AGI.
@littmath In one of my Differential Topology lectures the professor had extremely bad handwriting, and students were whispering to each other about what a particular "backwards epsilon" symbol meant. Eventually someone got the courage to raise their hand and ask. It was a 3.
@colin_fraser This is very impressive, in the sense that I'm surprised it works at all. The cat is also definitely not jogging or darting or doing most of what's in the prompt.
@wizards_magic@MTGSecretLair Is there a way to get a nonfoil version of the mox box secret lair? All the matching Dan Frazier ones have had nonfoil versions.
@colin_fraser@MikeIsaac @anvaway One potential outcome: aggregator sites pop up and people use them to identify "enemies" who block too many of the people you like. Could enable brigading, and be more confrontational than screen shot / dunk dynamics we see now.
@wrmacrae @carlgoodtoseeu1 Then think about what happens over an average 9 games. Any given card (in either deck) is drawn in 5 of 9 games - 4 of the 6 long games, 1 of the 3 short games - so aggro cards win 1 of 5 for a 20% GIH win rate, control get 80%. (But remember the aggro deck wins 33% of games!)
@wrmacrae @carlgoodtoseeu1 Here's a stripped down model that might help: Only one matchup: Aggro vs Control. Games only short or long. Aggro wins 100% of short games, control 100% of long games. 2/3 of games go long. Suppose also that each player draws 2/3 of their deck in long games, 1/3 in short games.