If you trojan horsed poker into some other game, would it generate new theory?
??? Assuming that you could do this: remap poker 100% perfectly into a narratively unique visual logic so that one player is having an entirely different gameplay experience on some other system, and to the extent that they mightn't even realise it’s the same game.
i.e. imagine if you were playing Tetris but really behind the scenes it was perfect chess. It requires a 1-to-1 ratio of consequenceness, and would need to effectively match the timing and logic perfectly. But while it would play exactly like poker, you wouldn't be looking at cards.
You could then connect both systems/skins globally via API in one giant onion network to make a super game. Each ‘skin’ would naturally generate its own style of training content and focus per target audience and per visual logic of its particular ‘world’. Housewife poker. Warhammer poker. Dog poker. etc. Still always fundamentally poker, but built around some other lore than the traditional deck.
Spoiler
And if that can be done then my follow up question is: *couldn't* one of these alternative skins generate entirely new poker theory? Based on some as yet unforeseen unconscious human behaviour? What does that look like? It's got to be easier (and thus quicker) to cognise and process solver data if there’s some profound or personal gameworld story that links everything up, especially if it promotes more gto play. Mnemonic shortcuts and trapdoors seem inevitable. And maybe online poker is more fun anyway the more customisable and visually flexible it is for the end user? i.e. even regs mightn't go back (cbets would be 10x more satisfying i think if they could be customised as kinetic bullets that physically maimed your opponent - for example. and things like MDA or recent stats could visually remain stuck on avatars as scars or gaping flesh wounds or whatever)
Or would it just be turgid animated crap? Is the reason poker is successful precisely because of its simplicity and politically neutral aesthetics? (I would argue against that apparent 'neutrality' but that's another topic). CPU horsepower is also always a concern ofc but you could still build a coherent and fluid 'game' client side I think these days, even on the most primitive of smart phones, and just ping the same core simple gamestate data back and forth in the background.
Plus presumably poker variants have already altered each other’s meta often enough already? If so, then there is some precedent for this silly idea.
21 Replies
Ceres, you are the ultimate jazz improviser of technical and scientific terminology.
To merge poker with some other game, first you would need to find a key similarity in the rules that connects both of them. Only then you could combine them and search for the possible ignition point of some new game theory. Of course you can create the 2nd game from scratch, all by yourself. Then it would be easier to connect its rules with poker. But it still would be difficult, I think.
I'm not sure whether secretly "trojan horsing" poker into other game would be different than just normally merging them two and informing the players, but probably the moment of disclosure would be interesting.
If payoff matrix is the same you can represent game in various ways. This won't change underlying theory.
Wait a minute... I think I might be accidentally channeling the plot of Severance :(
Ok. yes; it probably is a little bit of an overstretch.
Maybe human minds couldn’t, but… what about non-human minds like, idk.…mr AI? Don’t they already train models on video games? Seems like all you need then is something like a Snowie core masterplan to use as your kernel and then mr AI could break down every decision/spot into hundreds of theme wrapped independent mini-games, click ‘simulate that then AI in the form of a retro cutesy robot platformer’, and that’s it! The button does the rest.
We already practice on toy game gto trainers. Snapshots of the tree where single decisions can technically be won or lost and tallied ‘as is’; all totally independent of the traditional, your go/my go gametree itself. These are technically mini games that perfectly simulate the actual game (at least from a player v npc/theory pov). With just a few AI directed layers on top..
Some kid is playing Duck Hunt in Zambia with no clue he’s crushing 300nl regs on Coinstars... until level 13, when he wins 20k capping some mallard's ass.
If payoff matrix is the same you can represent game in various ways. This won't change underlying theory.
Well it mightn't change the theory per se, just demonstrate some property of it we haven't channeled yet. eg. the EV loss of taking too long on less meaningful decisions could be weighted into the toy game so that it leans you towards identifying where to shovel mental horse power in more efficiently (in the same way solvers have taught us counting combos is for the birds)
I think trading is a perfect target for this, thousands of thousands of homebound young adults with something in the bank willing to put in the hours looking at numbers they don't understand.
Poker would be a bad trojan horse as it isn't really useful, much of the value from playing poker comes from being a human and agreeing to the ToS of a poker room. (As opposed to a machine playing the game) it's unautomatable by design.
I'm pretty sure that traders are used as a vehicle to perform all kinds of detached computation ala severance.
An especially important usecase of this I think is the correlation between prices of different public companies into what are effectively implicit partnerships. For example maybe google starts showing more videogame ads which causes more nvidia sales. Writing a referal or ad contract for gpu sales might be cumbersome or illegal due to antitrust, so traders are tasked with tightly arbitraging the GOOG NVIDIA pair to correlate with the amount of videogame traffic pushed by youtube.
Or I guess a commodities trader might just find a correlation between commodity A and commodity B, and maybe swap in a currency pair in there and arbitraging on a daily cycle. Unbeknownst to the trader the commodities were steel and oil, the currencies were Arabian and Russian money, and the timezone difference was that of those countries. So effectively the trader is providing liquidity and funds to arms traders.
To cite an actual case, used in one of Taleb's books
https://en.wikipedia.org/wiki/Antifragil...
A trader made money trading a commodity called green lumber, which means that the lumber hasn't yet been dried. And the trader actually thought it was visually green.
So yeah, you could pay traders to play poker, especially broke memecoin traders. They could compute on realtime or presolve, but I doubt they would outperform a machine. Maybe the advantage could be in exploitative as opposed to optimal poker, as if you force a human to do computations they will likely mimic human thought and exploits.
Interesting post, Ceres. In short, the question is:
Haizemberg93 is right: the same mechanics mean the same underlying theory. However, a new framework can shift your perspective.
For example, I used to play on a site called Range vs Range (now sadly defunct). It was like correspondence chess for HUNL: instead of playing one hand at a time, players took turns setting their entire range's strategy, and then a single action was randomly chosen based on their assigned frequencies.
Players quickly discovered that their default strategies were wildly imbalanced, especially in defensive lines like missed c-bets. Being forced to play a range full of garbage was a punishing experience, so most players quickly adopted more balanced approaches.
In short, RvR "embedded" the game of poker in a new framework. The payoffs and game structure were identical, but the experience was very different. The psychological incentives changed, and players adapted new strategies in response. Furthermore, the new perspective immediately revealed imbalanced lines, which were hard to see playing traditional poker.
Well it mightn't change the theory per se, just demonstrate some property of it we haven't channeled yet. eg. the EV loss of taking too long on less meaningful decisions could be weighted into the toy game so that it leans you towards identifying where to shovel mental horse power in more efficiently (in the same way solvers have taught us counting combos is for the birds)
If they turned it into a video game, when a player at your table was habitually tanking you could get bonus points for capping his ass. 😂
I was visualizing your whole idea of turning poker into another game, and realized how frustrating a video game based on poker would likely be. Like you play the entire level perfectly but 40% of the time you just die anyway.
It seems like it would be difficult but maybe something like this could actually be done though. Maybe as AI advances?
I've actually had similar ideas about encoding poker to a different format that would be easier to master. Like when I was a kid I used to be good at intuitively figuring out easier ways to solve math problems than the way the teacher was showing us.
So much of poker is about frequencies. If you could just figure out a shorthand way to mimic the frequencies almost exactly you would destroy most players, even if your combo choices weren't what a solver would use.
Like I've wondered if you could just memorize the frequencies for each hand class on given boards. Then use some sort of random number generator to decide which action to take, instead of worrying about individual combos.
For example if you know you are supposed to bet with a flush draw approximately 50% of the time you could just flip a coin each time you have a flush draw to decide what to do.
If you're supposed to call a bet 1/6 of the time with bluff catchers you could just roll a die every time you're in that situation and if you roll a one you call.
Anyway, interesting thread idea.
Good stuff, exactly where I was thinking.
I think trading is a perfect target for this, thousands of thousands of homebound young adults with something in the bank willing to put in the hours looking at numbers they don't understand.
So many parallels between trading and poker. We know they use bots to grind out daily EV. Is it conceivable that AI prediction models reach a level of predictive accuracy/return so far in advance of humans that it becomes de facto daddy in all things %?
I think so. At some point. If it’s not there already. Say 90% of trade could be reconciled reliably to within some margin of error then, in principle, individual poker outcomes could be paired to mini trades with similarly weighted failure/success rates. This being unlikely to happen in real time ofc. But that could even distinguish the character of the imposter game (i.e. you play disconnected ‘spots’ and you only find out how it went the following day).
As you say, they're playing with similar things in crypto finance already. Algorithmically managed money would be useful scaffolding behind the scenes given now there's potential to add some form of instantaneity and genuine, locked in, loss to every decision.
In short, RvR "embedded" the game of poker in a new framework. The payoffs and game structure were identical, but the experience was very different. The psychological incentives changed, and players adapted new strategies in response. Furthermore, the new perspective immediately revealed imbalanced lines, which were hard to see playing traditional poker.
Yeah this is a perfect example. I remember seeing it years ago but I didn’t have the skills to play. Separating out the range as a broader responsibility seems like a very logical advancement in broader perspective/ tactical awareness; bit more like an RPG really. I think this format is more amenable to narrative conversion too: i.e. hero’s ‘army’ against villain’s ‘army’ etc.
Presumably the delay was problematic. The buzz of poker being the instantaneous thrill. But I don’t think that’s unsurmountable either with enough players in a pool. Or the imposter game is slowed down in a lore appropriate way like a world building sim etc.
But connecting all that up to a real-time poker game... er, tricky
I was visualizing your whole idea of turning poker into another game, and realized how frustrating a video game based on poker would likely be. Like you play the entire level perfectly but 40% of the time you just die anyway.
This sounds like my experience playing Souls games. 😉
Ultimately people do love dying so idk. They actively LOVE it. :Crazy:
So much of poker is about frequencies. If you could just figure out a shorthand way to mimic the frequencies almost exactly you would destroy most players, even if your combo choices weren't what a solver would use.
Agree, this is the crux of it. How to ‘sell’ frequencies in a format that *feels* like something else. I’ve also experimented with various (extremely) basic training drills/toy games using the idiot proof kiddie code program Scratch. So I do believe every spot in poker could be broken down into a concise mini-game. But linking them all up into one cohesive playable system?
Maybe it'd need to be less Duck hunt and more Sniper Elite.
Ok. Quick idea. Say you had a neuralink inserted into your brain, reading all your thoughts.
According to our best guess, we make around 35k decisions a day. How many hands of poker is that? It can’t be long until AI could approximate the likely success of many of those outcomes to a high degree of accuracy (they already use it to predict criminality), and all in the background of your day while you’re busy making decisions. Then it could total everything up overnight like a pokertracker for the mind.
Then it could distribute you ‘decision EV’ per choice actually made, scrape that data and match it up against some other human playing actual poker in the same predesignated skill pool with similar or preferably identical EV/equity splits for every decision paired. The final BBs (or ‘profit’😉 would then be distributed equitably per each player’s overall performance. You could still replay hands/decisions and check your work, or how you played the hand either as IRL brain or IRL poker player, revealing new insights into how your thoughts and attitude affect decisions.
The key difference here would be that the IRL poker player is no longer caring about variance, because they know it’ll be evened out. Everything will be squared out in EV. (tangent: surely someone has attempted to create pure EV poker yet?)
You could even take turns. Monday: neralink. Tues: pokerstars. etc
This could result in literally everyone on the planet playing poker against each other all at once. Think of the rake!!
Hey, when i'm in charge we all get rich. Remember that.
I'm all individual.
I'm all got to think for ourself.
One could see a message on his screen that says:
"You are being currently trojan horsed. You are currently in the 158th place of the Sunday Million MTT, 36th in chips. Open Instagram or TikTok to raise your VPIP."
One could see a message on his screen that says:
"You are being currently trojan horsed. You are currently in the 158th place of the Sunday Million MTT, 36th in chips. Open Instagram or TikTok to raise your VPIP."
Obviously, I haven't play any poker for a long time 😃
But what I mean is, it could be interesting to realize that you are playing poker through another software.
How about a piñata? 🐎
That’s basically already a horse so you wouldn’t need much of a disguise. Just batter the thing open, in a whacking style wholly congruent with GTO, to release pokerstars chests and satellite tickets. ¡Blam!
No input on the topic but you make a good case for recreational drug use
He's just on fire.
