After Go, what games should be next for DeepMind?

So chess and Go are both games of perfect in­for­ma­tion. How im­por­tant is it for the next game that Deep­Mind is trained on to be a game of perfect in­for­ma­tion?

How would the AI perform on gen­er­al­ized ver­sions of both chess and Go? What about games like poker and Magic the Gather­ing?

How re­al­is­tic do you think it’s pos­si­ble to train Deep­Mind on games of perfect in­for­ma­tion (full-map-re­veal) against top-ranked play­ers on games like Star­craft, AOE2, Civ, Sins of a So­lar Em­pire, Com­mand and Con­quer, and To­tal War, for ex­am­ple? (in all pos­si­ble map set­tings, in­clud­ing ones peo­ple don’t fre­quently play at—e.g. start at “high re­source” lev­els). How im­por­tant is it for the AI to have a di­verse set/​library of user-cre­ated re­plays to test it­self against, for ex­am­ple?

I’m also think­ing… Shitty AI has always held back both RTS and TBS games.. Is it pos­si­ble that we’re only a few years away from non-shitty AI in all RTS and TBS games? Or is the AI in many of these games too hard-coded in to ac­tu­ally mat­ter? (e.g. I know some peo­ple who de­velop AI for AOE2, and there are is­sues with AI be­hav­ior in the game be­ing hard-coded in—e.g. villagers delet­ing the build­ing they’re build­ing if you sim­ply at­tack them).