Poker Has Fallen

[LVH2]

I, and some people whom are more smart than me at gambling, are pretty skeptical about this.

The sample size is small. The results are kind of absurd. The computer crushed them like bugs. Real elite NLHE players are thought to be not all that far removed from optimal play.

Until this came out, the general consensus was NLHE was a long way from being solved. The team here did not claim to have solved the game, but the results are way beyond what anybody suspected.

Not that I'm much of an expert, but these were the points raised about this article in my gambling forum of choice.

[Smitty-48]

The military applications are limitless, that's the sort of AI that Smitty LeMay needs to find the mobile ICBM's after dispersal, in order to launch a preemptive counterforce.

[Hanarchy Montanarchy]

My guess is that the computer could win because it wasn't programmed with any meaningful sense of the consequences for losing, so it was willing to bet the farm on small pools.

Humans are naturally loss averse, so this strategy worked here... but I am not sure if that can be universalized in a useful way.

[Hanarchy Montanarchy]

The military applications are limitless, that's the sort of AI that Smitty LeMay needs to find the mobile ICBM's after dispersal, in order to launch a preemptive counterforce.

I think I can actually hear you hyperventilating with excitement all the way from Canada.

But I don't think that is the point. In your best-case-scenario, the decision to launch a preemptive strike is still left to humans right? Letting an AI make that calculation based on opaque rules is troubling.

[Smitty-48]

The military applications are limitless, that's the sort of AI that Smitty LeMay needs to find the mobile ICBM's after dispersal, in order to launch a preemptive counterforce.

I think I can actually hear you hyperventilating with excitement all the way from Canada.

But I don't think that is the point. In your best-case-scenario, the decision to launch a preemptive strike is still left to humans right? Letting an AI make that calculation based on opaque rules is troubling.

The role of the humans is to get themselves right up against the brink in the fog of war on the precipice of losing control by misapprension, miscalcualtion and/or mistake, the job of the AI is to hollisitcally crunch big data while reading adversary behavior, to give ol' Smitty LeMay them mobile ICBMs, up in the saddle for the counterforce, subs ain't the problem, you can follow subs around, the problem is where did Ivan put those mobile ICBM's upon dispersal? Those are the tricky birds to hit.

[LVH2]

My guess is that the computer could win because it wasn't programmed with any meaningful sense of the consequences for losing, so it was willing to bet the farm on small pools.

Humans are naturally loss averse, so this strategy worked here... but I am not sure if that can be universalized in a useful way.

I doubt that. The value of aggression in NLHE is well known and many of the top players are pretty much fearless. That's one reason many of them have built fortunes out of nothing, then lost them back by being overly reckless. Tom Dwan, who was worth 8 figures, recently went broke playing over his head against Chinese business men.

If I, or probably most of us, ever had that kind of money we'd never even dream of putting it at risk like that.

Also, my understanding was that they were not playing for huge stakes here. Maybe even for play money. I mean, were these college profs or whatever, putting up millions of dollars?

[Hanarchy Montanarchy]

The role of the humans is to get themselves right up against the brink in the fog of war on the precipice of losing control by misapprension, miscalcualtion and/or mistake, the job of the AI is to hollisitcally crunch big data while reading adversary behavior, to give ol' Smitty LeMay them mobile ICBMs, up in the saddle for the counterforce, subs ain't the problem, you can follow subs around, the problem is where did Ivan put those mobile ICBM's upon dispersal? Those are the tricky birds to hit.

I don't mind if a robot is aiming the gun... I just don't want it pulling the trigger.

[Smitty-48]

The role of the humans is to get themselves right up against the brink in the fog of war on the precipice of losing control by misapprension, miscalcualtion and/or mistake, the job of the AI is to hollisitcally crunch big data while reading adversary behavior, to give ol' Smitty LeMay them mobile ICBMs, up in the saddle for the counterforce, subs ain't the problem, you can follow subs around, the problem is where did Ivan put those mobile ICBM's upon dispersal? Those are the tricky birds to hit.

I don't mind if a robot is aiming the gun... I just don't want it pulling the trigger.

There's no trigger, as I always say, it's like a gravity well, once you get caught in it, you are going to go over, the question simply becomes, will it be by mutual countervalue exchange, or preemptive counterforce?

[Hanarchy Montanarchy]

My guess is that the computer could win because it wasn't programmed with any meaningful sense of the consequences for losing, so it was willing to bet the farm on small pools.

Humans are naturally loss averse, so this strategy worked here... but I am not sure if that can be universalized in a useful way.

I doubt that. The value of aggression in NLHE is well known and many of the top players are pretty much fearless. That's one reason many of them have built fortunes out of nothing, then lost them back by being overly reckless. Tom Dwan, who was worth 8 figures, recently went broke playing over his head against Chinese business men.

If I, or probably most of us, ever had that kind of money we'd never even dream of putting it at risk like that.

Also, my understanding was that they were not playing for huge stakes here. Maybe even for play money. I mean, were these college profs or whatever, putting up millions of dollars?

I thought they had it playing against experts? Did they mean that the computer got so far ahead of non experts that if you popped a few pros in they wouldn't be able to make a comeback?

That is less impressive.

[Hanarchy Montanarchy]

There's no trigger, as I always say, it's like a gravity well, once you get caught in it, you are going to go over, the question simply becomes, will it be by mutual countervalue exchange, or preemptive counterforce?

That decision... is that President Terminator Watson or Smitty LeMay making it?