Episode 57: Jonathan Bredin and PokerSnowie Follow-Up

14 thoughts on “Episode 57: Jonathan Bredin and PokerSnowie Follow-Up”

1. Flop is 88T not TT8 as shown above

2. This GTO conversation is really interesting, but makes me think how much maths i no longer know. It seems to me that it would really useful to get some sense of what the space of possible strategies looks like, because it’s not immediately clear to me how you go about comparing one complete strategy from another, and a lot of the sort of things you would like to say about potential multiple GTO strategies involves ideas about ‘adding’ strategies, or sub-parts of strategies, or distances and differences of one strategy from another. But it seems as though there’s possibly a level of entanglement within a single complete strategy which makes that sort of thing likely tricky.

   Is it fair to say that at low stack sizes there can be a first mover advantage (in HE), but (as far as we know) the deeper stacks are the ones where position becomes king? It strikes me that the important element is the ability to make the last raise (of some sort of ‘sensible’ SPR), because of the power of being able to polarise your betting range.

3. Here are a few corrections and comments about Tom’s letter (concerning GTO) that you read and discussed on air.

   1. Tom states that poker must have at least one Mixed Strategy Nash Equilibrium because it’s a zero sum game. This is actually wrong whenever there are at least three players playing: the theorem that says that a Nash Equilibrium has to exist holds only for two-player games, and IIRC it’s pretty easy to show three-player games where there is no Nash Equilibrium.

   It’s still probably fine to talk about Nash Equilibrium in poker because if you’re playing optimally vs each player separately then you should be relatively fine. But technically speaking, there doesn’t have to be a Nash Equilibrium.

   2. If there is more than one Nash Equilibriun, then there are infinitely many of them. You can see this because if there are two Nash Equilibria, denote them by NE1 and NE2, then the strategy that flips a coin at the beginning of the hand, and if the coin lands on head we play NE1 for this hand, otherwise we play NE2 for the hand, this strategy is also a Nash Equilibrium. (That’s easy to prove, just from the basic definition of Nash Equilibrium.) It would be an Equilibrium also if we make the coin biased, such that in lands on head with some probability 0<p<1. So, I've just shown a way to construct an infinite number of Equilibria. It's easy to see that these Equilbria are all different than one another if NE1 and NE2 are different.

   3. Having said that, I share Nate's intuition that if there are multiple Nash Equilibria (as we see, an infinite number), they won't be significantly different than each other. I can't think of any way to prove this, tough.

   4. A nitpick: Nate talked about Nash Equilibria being a measure-zero set, and claimed that this means there's "few of them". That's not a good statement, IMO. A measure-zero set can still be *huge*, and complex, and have interesting structure. For example, in a Euclidean three-dimensional space (think of the room you're in), the face of a ball (AKA a sphere) is a measure-zero set but it's obviously a very interesting set with lots of interesting structure. Measure-zero sets even become more and more interesting the higher the dimension is. Since the mixed strategy space of a game like poker has huge dimensionality, the fact that the set of Nash Equilibria is measure-zero actually doesn't tell us almost anything about the set of Nash Equilibria, and certainly doesn't tell us it's "simple" or "small".

   • I guess the real conclusion is twofold: GTO thinking leads to some counter intuitive conclusions, and the guys deserve some credit for venturing into an area where angels fear to tread 😉

     • Thanks! That’s the best we can hope for, I think.

   • Very interesting, thanks Mob. I was hoping there would be someone out there who would know more about this than we, and more importantly that that person would be willing to take the time to write up something like this.

     • I have a Ph.D. in math and computer science, and worked a bit on game theory and on learning theory, so the whole PokerSnowie thing is right up my alley. So far I haven’t noticed any glaring errors in your discussion of game theory, but I’ll of course be happy to answer any questions that come up about these topics.

       I did have some comments on the PokerSnowie interview that I never wrote down, but I’ll try to listen to it again and write some notes and post them in the comments of that episode. In general, I’m in the skeptic camp. I’m not saying PokerSnowie plays badly (I haven’t run it at all), but from the things said in the interview I got the sense the techniques used to train Snowie sound pretty far from what one would need to actually get close to GTO. (But there weren’t enough details given in that interview, so this is just a hunch.)

       • Thanks very much for contributing here! And, of course, for correcting me.

4. Poker Snowy makes a lot of noise.I see several powerful computer programs which can change or revolutionize the game of poker.
   I mean http://www.affdex.com/ or http://www.emotient.com/
   I believe that both programs will be available on mobile platform next year via cloud.
   I will not be surprised if US agencies have much more powerful technology where they can combine facial coding,voice sensing, gesture tracking and gaze tracking.

5. Andy,

   have any of these technologies proven out if the people were intentionally trying to hide/disguise their feelings? I could not find any references to that question on eithr companies’ site.

   which

6. interesting that Jonathan Bredin felt that Thinking Poker was not heard much in that part of the world. Y’all seemed to find that amusing, but do you have download info available that separates out where your listeners are from?

   Reason I ask, I wonder if podcasts for poker (or other strategic games ) are a cultural phenomena. Would Europe have more ‘learners’ for games than the U.S.? It seems to me that once Western Europe (especially German and Netherland areas) get latched onto a game, they take it more seriously perhaps (per capita/time invested in learning) than the average American. And perhaps Americans take poker more seriously than other parts of the world.

   Not sure how I might quantify or even use that info, but might be cool to know.

   which

7. I’d like to thank you guys for your interview with Jonathan. I was really impressed that you went to such great lengths to interview a non-verbal player, and to share it with us. The issue is particularly important to me as my 5 year old son was born with a number of congenital issues, and is also both autistic and dyspraxic. The advent of adaptive technologies over the last few years, especially via the iPad, has led to a literal breakthrough with thousands of non-verbal people. My son has struggled to communicate with us since birth, but as he is learning to use various adaptive apps it is amazing to see the strides he has made. Only a decade ago, even therapists who work with non-verbal kids every day had no idea how much was really going on in their minds. Seeing Jonathan not only adapt to his disability but thrive gives me hope as I watch my son grow. Knowing that you guys were willing to take the time to interview him, even though the process was cumbersome, give me hope for the rest of us too.

   I’ve played poker with a number of players with varying degrees of disability, from vets who’d lost multiple limbs to guys with significant genetic issues who required an aide to play. Poker allows all of these people to not only participate, but to compete as equals. That’s an amazing thing. I think I may teach my son to play cards this weekend.

   • Thanks, Tony. I emailed your comment to Jonathan as I’m sure he’d like to see it and I don’t know whether he’s following the comments here. It made me smile to read it as well.

     I for one would be very curious to hear how you go about teaching card games to your son, how he progresses with them, etc. You probably heard us discuss this topic on a recent show (well, that was how to introduce poker specifically), and the last time I was at his house my cousin made a passing reference to my helping him teach his 7-year old son poker, or other card games that could be a stepping stone.

8. Frankly, I have no idea how I plan to teach him. I will probably start by just teaching him to match cards of the same rank, as that type of organization is something they do a lot in his therapy. He is really too young (not 5 yet) to learn the game itself though, even if he was neuro-typical (it pains me that I use politically correct weasel words like that, but I do), but I’ve taught my 8 year old bits and pieces over the years and I can use him as a test subject.

   My thinking used to be that you teach kids poker like I learned it…via 5 card draw. They learn the hand rankings and the mechanics of making a pair and going to showdown. The nuances of a community card game can come later. With my 8 year old, though, he got his first exposure to poker back when Pokerstars was still viable in the US. He would watch me, and if I needed to run around the house or check on the baby, he would call out the hands and I would tell him what to do. This led to an amusing moment during a charity tournament I ran. He came in and I showed him my hole cards (7-2os) and asked him out loud what I should do. When he responded “fold,” I asked him why. He announced to the table “because that’s the worst hand in poker.” He clearly picked something up.

   My autistic son may never speak on his own, but the idea that one day he could sit at a poker table and touch a tab on his iPad so that a computer-generated voice announces “raise” when a scare card comes on the river makes me smile.

Comments are closed.