Agreed. After analyzing Les' hands yesterday, I think it's clear he needs to adjust today and start folding more sets. I mean, what's with this station? Guy never saw a set he didn't like.

Les is "megabanny" on Stars and therefore quite indisputably in the top 10

rekt.

I was under the impression that the neopokerbot available online played in different settings? I remember when I played it, it had multiple open sizes if you restart the game.

no. you're failing to understand that it's the difference which is important. doug gets same cards vs bot that bot gets vs les. doug wins 115k from bot while bot can only win 70k from les, it means that the humans are playing better because they are winning more/losing less with the same hands as the bot

it would likely be impossible for les to win over that particular sample of hands

Doug also ran ridiculously hot with his choice of 3 betting/raising limp hands (also iirc a hand where he x/c T9 on AK7 and rivered nuts). He was always getting there in bigger pots where the AI wouldn't necessarily play for the same size of the pot (or maybe even would have folded before the river).

Anyway it's early days but bots don't seem as scary as first thought.

Whereas that is true, I frankly expect the opposite to happen. I expect team hooman to come up with a better gameplan every day. All four will mindonk flops today with ideas what to do on later streets. The programmers will not be able to adjust the program fast enough is what I expect and the adjustments wont be very good I think. The lead-programmer apparently has no idea how poker and game theory work and this is pure game theory now. So I think the exploits will become more extreme and the bot will be havocked in a few days. I feel a bit sorry for the head programmer, he seemed to be rather proud of his machine, but seems like it will take a few more years for them to come up with something good.

This doesn't follow. A particular set of outcomes can make a worse strategy do better, even reciprocally, than a better strategy. As a basic thought experiment, consider that two players play hands in exactly the same way except one person always folds draws facing an all-in even when getting the right price. If the draws happen to miss, does that mean the person folding is playing better?

floating flops with low equity and getting there from time to time is pretty standard, what you didn't mention tho is that when doug rivers the nuts in that hand he overbets allin and gets paid- effective betsizing makes a world of difference and it's pretty clear that this bot is just clicking random buttons with its sizing in general.

this will continue to be a bloodbath

lol dude, these guys are 4 of the best hunl players in the world, they're not just playing like dolts and getting lucky ffs

True ... I also feel almost sorry for the guy, and no matter how bad this bot may lose eventually, and even though Tuomas didn't even seem to know what a donk bet is: I still have huge respect for him and his team.

You may be right, humans might adapt better than the bot in the days to come, but either way, I'm happy to follow the fight till the end ... sure enough, in a few years, one bot will destroy humans so badly, that the porgrammer team will feel sorry for the humans

pretty sure i heard the "programmer" say many times that they don't program any strategy into the bot. they created an algorithm that comes up with it's own strategy, and will adjust it's own strategy over night during the challenge.

really interested to see how the bot adjusts over the coming sessions

You're missing the point. That was just an example. Especially over smaller samples, even if both are playing the same hands and boards against the same strategy (let's call it strategy X), the fact that strategy A gets better results than strategy B does not necessarily imply that strategy A is superior to strategy B against strategy X. With that said, the stdevbb/100 will be much lower.

I agree and I highlighted the difference. Lets hope then that the secondary table guys will swap places so Les dont need to play against WCGs cards hahaha.

If you don't understand that luck still exists in duplicate poker then lol you.

No one is disputing that these guys are good, and that it looks like the bots have some leaks so far.

what i'm telling you, and the point you're missing, is that their strategies are not a whole lot different and both les and wcg are world class at making adjustments, so your "thought experiment" is pretty ridiculous

pretty sure i understand more about poker than you do, but w/e

look, this one random guy said it was a problem that les was losing so much to the bot. i explained why it's not a problem and that really isn't up for debate, so i don't know why you're trying to argue

Of course it does but it will be very interesting to measure how standard deviation changes in duplicate poker comparing to normal one. It's a pity HHs won't be released though so we won't be able to see it unless someone datamines the Twitch stream

Ok so that "still a problem" sentence wasnt that well thought but I guess you wanna see all humans beating the bot.

I'm talking about wcg/dong and the bot. Obviously the comparison is completely inappropriate with the other human team. It's not a matter of how good the player is (that's pretty much irrelevant), it's a matter of variance.

wtf are you even talking about

wcg is teamed with les, not donger

So WCG/Les are the ones playing the same cards (idk, I haven't been watching)? If so then just replace dong with les

k bro i'm gonna bow out now, if you want to tard up this thread any further you'll have to do it alone or with somebody else

looking forward to more streaming later today, anybody have any idea what time it may start again?

ah ok. that make sense. thought the programmers are changing strategys and pattern manually. cause if it is so, the project would be kinda useless

I only watched wcg stream yesterday- and IMO he ran well in terms of the duplicate poker scenario (won more than the bot having been lucky with runouts for his 3bet/raise limp range).

I agree that it's a shame hhs won't be provided.