Quote:
Originally Posted by NINzent
I played a lot with Stongpanyid/bloodywolf who was using RTA and he often acted under 2 seconds
Just because Fedor had a setup that could've been improved doesn't mean a low timebank would stop cheater, it would actually give the ones with better software an even bigger edge on humans.
Super low timebanks are no solution
Yes he can be able to act under 2 seconds for some select spots he might have pre-memorized or easily pulled up but a more complicated tree will require a longer solution/a longer time to look up the solution. We are talking about centaur cheaters here; they are still humans who know what they're doing roughly GTO poker wise and can perform simple actions in a short time (he wouldn't need to consult the dream machine to know that bottom set is a checkraise on the flop).
Anyways yes, read my post, I said short timebanks prevent
this type of cheating (manually looking up an offline database of indexed solves). If there exists dream machines that can calculate every spot - within a reasonable nash distance and with a reasonable tree configuration - within 12 seconds or less (to give a few seconds leeway for the human to actually click the correct action) then no, short timebanks isn't a solution.
If this level of dream machine truly is a reality, then the only thing preventing it from hitting every single online game available is objective cost and opportunity cost; if the dream machine costs too much it wouldn't be worth it to use in lower stakes - if the risk of being caught/funds confiscated/potential criminal charges are high, then it wouldn't be worth the initial cost/keeping a large roll on a dream machine account.
If the objective cost of a practical dream machine drops to the price range of 5k-10k USD in total, with little to no risk for the cheater, then I think at that point all public online games 500NL or higher have reached the point of no return.