Can player casino interaction be simulated on a computer?

[RS]

Interesting thread....


I'm kind of stuck in the middle on my opinion. For one, you can't quite plug in how you're going to act nor react in a given situation, given that human interaction varies quite a bit. When I do something, it can be interpreted in a bunch of different ways. How many times have you done something where you thought it could not be taken offensively in any way possible, yet the person felt majorly offended.

Some things can't be simulated unless you have a plethora of data on the subject, but even then, it's not too reliable. For example, take the difference between a republican and democrat. Sure, more republicans are going to have "republican-like" qualities, while more democrats are going to have more "democrat-like" qualities....but that doesn't mean the republican is Christian, nor that the democrat supports abortion. Sure, you can say a republican might be slightly more likely to be Christian and a democrat is slightly more likely to support abortion....but you can't have a strong correlation in your prediction unless you know a bunch about that person (ie: plethora of data), like their upbringing, background, as well as current mood or state of mind.

The other thing about sims is, at least how we typically refer to them in the blackjack world -- they are an expected result/value of the long run. If hitting has 0.1 more +EV than staying, the sim has the player hit, because it's aiming for the long run. 10% of the time, that play is going to be more profitable, but in the short run, it doesn't make a difference. (Err...think hitting vs staying on 16v10. Hitting wins 5 extra hands for every 1,000 hands, or something like that.)

Unlike software-optimal blackjack, human interaction doesn't quite have the "Going for the long run" aspect, because as soon as you messed up, you lose. If you hit a 12v6, you've lost a little bit of EV, but it has nearly no effect on the long run (unless you always hit 12v6), because you get to keep playing. But if you're interacting with a pit critter, as soon as you "mess up", you're flagged and you've lost. You haven't lost a little bit of EV with nearly no effect on the long run, but you've lost it all. You can't play anymore.

So, while it might be slightly more "EV" to deal with a PB in a certain fashion, there is much more going into it than just the numbers, like their reaction and how fast you can react, given all other factors.

If a sim were to be created to deal with PBs, human interaction, and all that, it would have to be tailored specifically for your current conditions, to the T. Game speed, types of other players and their reactions, what you look like (young, white, male or old, asian, female), how the PB perceives you, how the PB is feeling right now, etc.

Did the PB have a fight with his wife this morning?
-- So, maybe he's grumpy.
-- -- Does this mean he is more likely to protect his game and stick it to a counter?
-- Or, is he thinking of his fight with his wife?
-- -- He doesn't even care about his job right now and isn't worried about someone counting, he might even feel more generous and let things slide because it's less work for him.
-- Maybe he's trying to get his mind off of the fight, but is still grumpy, so he tries to chat with some players.
-- -- How is the player going to perceive this? If a grumpy PB comes over and starts chatting it up with you.....what do you think is going on? Chances are, you might be thinking about heading for the door soon if a PB seems irritated but is trying to be friendly. (ie: "He's irritated [because I'm counting] but he's coming over here to chat [and he's obviously faking "being happy"] just so he can watch my play.")

From viewing the PB and interacting with him, the player only gets partial information. Without having that full information, he can likely react in the wrong way. Maybe the player says something (about his wife or girlfriend) with absolutely no intention of offending the PB, but he gets pissed off when he hears it? [Look at the last sentence in first paragraph.] If the player just pissed off the PB, and the PB has a sense the player may be counting, he's going to be more likely to stick it to the player.


On the other hand, it's definitely possible since all sorts of things are and can be simmed. You can create objects in software to act how you want them to act (err...how you tell them to act). But, unless you have a huge amount of properties for each of the objects (even as trivial as say, PB being near first base but the player is at third base -- easier to look at the PB while talking and counting. But if the PB is on the third base side and so is the player, the player is going to have to swivel his head back and forth to talk and count at the same time, assuming the player is looking at the PB while talking).

To me, it seems like being able to simulate this is more of something that CAN be done, in theory. But, to actually do it and simulate it properly, it would be far more work and effort than the reward you're getting from viewing the results of the sim.


A silly example might be something like simulating what would happen if ants had the same brain power as humans. Sure, you COULD do it, but is it worth the effort? And what are you going to get out of it? If you do the human interaction sim, what is it going to tell you? Is it going to tell you how you should act/react to a PB?

They have simulated the Big Bang. Is that far enough away.

Ah, yes. The theory of the big bang.

[Norm]

If hitting has 0.1 more +EV than staying, the sim has the player hit, because it's aiming for the long run.

I wish people would stop saying things like this. Simulations need be nowhere near this simplistic.

[RS]

It was a simple example.

[mofungoo]

I seriously doubt that Blackjack Survey Voice will correctly identify any counter that does not sit at a table for hours.

This program can determine if a player is moving their money with the count in as little as 20 minutes. Whenever a review is triggered (because of winning activity) the operator plays the portion of the recording where the wins occurred and profiles that shoe. In most places the report is forwarded to the Shift Manager and/or the Casino Manager if the player is proven to be an AP. They will then decide what action to take regarding that player.

It takes time to complete such a review and it ties a surveillance operator up for the duration of the review. I believe that an AP would have to be hitting them pretty hard for them to go through this process, but you never know for sure.

[AussiePlayer]

This entire thread is completely and utterly irrelevant! While you're wasting your time interacting correctly, in an 'unsimulatable way,' with the pit boss; the Casino, or, Shift Manager is standing behind you and is about to tap you on the shoulder, because he's the one who actually makes the decisions. Interacting correctly with a pit boss is like trying not to offend a lamp post; who gives a sh!t!

And does this all mean you've completely changed your tune? You're the guy who goes on endlessly (and I do mean endlessly) about fractions of a percentage gains that are only present in simulations, and you now don't believe in simulations?

[Three]

If you don't have as large an advantage as you are capable of then giving up a tiny bit to get a bigger top bet and a larger spread can be high risk. With a strong playing advantage you can give up some EV to get more back through higher maxes and bigger spreads as well as longevity. Without a strong advantage the associated volatility can be tough for many to handle.

[Norm]

But, how much advantage are you giving up? That is the point of the chapter on Cover in Modern Blackjack. Most books that talk about cover give examples aimed at longevity. My chapter on cover is quite the opposite. It focuses on the costs of cover. Because, those costs are far higher than people generally realize. Indeed, some unwittingly give up their entire slim advantage with cover play. And, I was able to cover cover, because of cover sims.

[Three]

The higher your advantage the more you can afford. Properly thought out cover is fairly cheap. I have wrote on this before. People still put up quotes of what I wrote on Ken's sight. You are right though most people really don't understand how cheap or expensive cover plays are. Cover betting is even trickier. Any use of cover should be used sparingly. My hourly went up when I started using cover. But then I have a pretty fair grasp of the costs and how to reap benefits. As a red chipper I didn't feel I could afford cover or that it was necessary.

[Three]

I have been thinking about this thread and what Norm has said whic I respect as he is one of the foremost experts on a tleast blackjack sims if not in a more general sense. To me a sim has meaningful results if you can use them to improve your game. You can sim anything. It doesn't mean the results will be meaningful and useful;. The types of pit/player interactions I am referring to, the reads and a feel for what is going on in the pit and perhaps the surveillance room can be simmed but it will give you nothing to improve these things in your play. This is what I mean when I say they can't be simmed. The results of the sim are of no use to your pit interactions and reads and the like. So I guess a better way of putting it is it can be simmed but it is pointless to do so as the results are useless as a learning tool.

[Norm]

The point of such sims is to determine the cost of non-optimal play resulting from heat, and to find less costly measures. Fact is, many players give up huge amounts, if not all, of their advantage in an attempt to gain longevity.

[Three]

That is quite useful. I have agreed with you about many being clueless on the cost of cover plays. All you have to do is read some of the comments on this board. If you understand the cost of cover you can see how deluded they are. Even when experts point it out they still cling to the idea that they are using cheap cover when it is quite expensive. I totally agree with you on this point.