Yfw all board games will be solved by advanced future computers within your lifetime

>Win
Winning a game and solving a game are two different things. Winning just requires being better than humans, solving means deciding the absolute optimal play for every possible state of the game.

Yes, you're correct in saying that board games more complex than checkers are unlikely to be solved in the mathematical sense. However, we will probably reach the point where computers can outplay humans at effectively any well-defined game. OP is technically wrong, but the spirit of the post is more or less right.

I want to see someone create a piece of software that can beat the world's best human players in a game of Hungry Hungry Hippos.

>please trade with me [FAMILIAR NAME MISSING], this trade is calculated to be the most optimal and beneficial to both of us. You would have to be a dumb [SLUR MISSING] to not take this deal

Looks like you're going to have to farm your own resources there, you overgrown toaster.

How, when they can't even solve Rock Paper Scissors?

I'm a computer. Quit with all the downloadin'.

Evolutionary path based AIs can sometimes come up with unconventional strategies that humans missed, but overall we are pretty good at identifying optimal strategies; the problem usually lies in execution (both in the mechanical sense, with APM, and in the "choose an option randomly/calculate the probability on the spot" sense). that's where computers really have the advantage.

Both of these things will remain true, unless boardgames so complex get invented that optimal strategies become impossible for humans to find.

Chess is expected to be fully solved within 50 years, just saying

>Even something as simple as Chess has more possible board states than there are atoms in the universe

Are these gamebots going to break into my house to kick my ass at Settlers of Catan and Scrabble like a bunch of fun wrecking terminators?

The upper bound for chess states is somewhere around 7.7 * 10^45

The number of atoms in the universe is estimated between 4*10^79 and 4*10^81.

So yeah, bullshit.

Honestly, that's a pretty good illustration of why most games will never be solved. In any three plus player game that includes politics, the other players can just agree to kingmake each other and fuck you over.

That would be extremely easy if there wasn't inherent rng from the cheap plastic hippoes malfunctioning. All it would have to do is recognize the balls (easy) and calculate their movement and the speed of it's mechanical hippo pushing limb (even easier)

Which is also why those games suck competitively and so aren't worth solving anyway.

Your mom sucks competitively.

Yeah, she's really casual, so she kinda doesn't care about the solutions anyway, I guess.

>if human players decide to be cunts and throw the game for everyone its somehow a good thing
Political games are inherently casual with the game itself acting as a glorified vessel for socialization. They can still be 'solved', though, if you set the game up in a way that no player has an agenda to fuck the AI over:
>equal number of AI players and human players
>all humans are strangers to one another
>ideally, the game is conducted through a mediator so nobody knows exactly which players are AI

Assuming by "solved" you mean "humans will never beat them".

youtube.com/watch?v=jAu1ZsTCA64

As a high level league player I find this far more disturbing.

Yeah, computer vision may actually be good enough for this to be a fun project.

What IS the optimal strategy for Hungry Hungry Hippos, anyway? Press the hippo-moving button as fast as you can?

I think I just want a robot to play Fireball Island with.

You don't "solve" imperfect information games. As soon as bluff is a factor, it becomes impossible to solve.

You can calculate the odds and play accordingly. Doesn't mean the computer will always win, but it'll have a more consistent/better win rate over time than humans.

>socialization is casual
Veeky Forums in a nutshell.

Enjoy your group solitaire.

2p or 2 team games aren't solitaire, but don't have kingmaking bullshit by default.

There probably exist some more-than-two-player/team games that also lessen/eliminate kingmaking, though I can't recall any off hand.

Humans are pretty good at that, too. But if it doesn't guarantee a win, as it does in perfect information games when we say that it is solved, I don't really consider it solved.

Compsci fag here, you're an idiot.

>winning = solving
How about you shut up if you're just gonna spew out some bullshit outta dainty little mouth, eh slut?

While I don't really cares if the game is 'solved' (backgammon was solved long time ago and I still play it all the time), but it does sour the online play. Pretty much no reason to play chess online because you're bound to run into cunts who have programs running in the background. Heck, even my online Dominion games feel shitty because simulators are so widespread in the community.

>Humans are pretty good at that, too.

Nope, we are adequate at best. Calculating odds for the more complex games is hard, but even worse, hidden information games require random moves, and our RNG is terrible.

Case in point: RPS. The "optimal strategy" is throwing 33.3/33.3/33.3% r/p/s, but the human brain is incapable of not falling into patterns, and using that you (or a computer) can get a better win-rate than 50%.

Help computer

>tfw transhumanism takes off and human/AI hybrids invent increasingly complex games until eventually all gaming is just a perfectly accurate simulation of the universe to play god in, and the nested universes we are in grows 1 deeper.

We cannot but expect for computers to become even better at protein simulation to the point of live tissue models

>playing games to win

Games based on random chance can't be "solved". How is a computer going to be better at Clue than a human?

How about (and I know this is a radical idea here) you play for fun?

Why do that when I can fuck the computer instead?

Does....does that mean Thulsa Doom gets robot bitches?

How about I only have fun when I'm also playing to win, instead of just randomly pushing pieces around like some kind of retarded preteen and his imaginary tiger friend?

Not steel but flesh. Bioengineered bitches for Doom

en.wikipedia.org/wiki/Shannon_number

>The Shannon number, named after Claude Shannon, is a conservative lower bound (not an estimate) of the game-tree complexity of chess of 10^20,

Whoops, edited out the 1 to put the exponential, which didn't copy paste. It's 10 to the 120th, not 20th.

Truly this is the best of worlds.

>As a comparison to the Shannon number, if chess is analyzed for the number of "sensible" games that can be played (not counting ridiculous or obvious game-losing moves such as moving a queen to be immediately captured by a pawn), then the result is closer to around 10^40 games.

I am kinda hoping at some point they will find the right amount of point advantage that they should give white on Go. Having the first move is so powerful that over the years they have been increasing the amount of points you get by simply going second and even then most players still want to go first because it is never enough. Finding the sweet spot seem like something on an AI would be able to do.

Does that number of "sensible" games also exclude silly things like both players shuffling as many pieces across the board as possible without actually making any captures?
Me moving both my knights and bishops right in front of the other guys pawn line isn't a losing move if he's not taking any of them.

I have no idea, I was just quoting the same Wikipedia page. You can probably find the paper behind those numbers which might detail their criteria.

there are already game assisters (basically bots that take over for you) that have near instant dodges or counterabilities so its not in riots interest to make something that just outperforms humans in the phisical sense

>solving monopoly
does this mean i will never have to play monopoly ever again

Its easy to solve that, though. Auction the right to go first.

Due to the nature of the game, bots would probably get to the point where White wins 100% of the time, due to always being one step ahead on the optimal strategy.

>.t

t. idiot

>what I set out to do for 8 years
That asian sounded thoroughly demoralized.

Sorry brother, chess is already a game where AIs regularly beat human players, and Go, considered untouchable for decades, has fallen.

You don't understand the world you live in, dinosaur. Or to quote,
>.t idiot

>Doesn't know what solving a game means.
Try again. I'll give you a hand. Tic-Tac-Toe is solved. Connect 4 is solved. Chess is not. Can you figure out the difference?

He never mentioned solved, but I think took the idea that since AI would be the only one doing the solving it would also surpass us in every game.
On a mildly related note, it seems like they're slowly increasing the spatial component of games AI are applied to. Once these AI are given means to navigate the physical world, their domination will be swift.

I said nothing about them being solved, only that AIs consistently outplay humans in games that were thought untouchable precisely because of 'solvability' concerns. It is expected that these games will in fact be 'solved' on the near horizon, and a betting man would be wise to say that AI will not only regularly outplay humans in most games in our lifetime, but also that they will arrive at a practically solved state in that span.

Well, wouldn't you be? This is his livelihood, and he just got thoroughly stomped by an AI that's 6 months old.

>OP talks about how all board games will be solved
>Post about how no, that's not actually true
>user in particular writes about how that is foolish because AIs can beat humans at games like chess and Go

Either wrote the post in an attempt to contradict and is simply wrong, or he brings up a fact as irrelevant as saying that footracing is the way of the dinosaur, because even horseback races are faster, let alone something like automobile racing.

>I find this far more disturbing.
How? Do you know how much fucking supplementary information the AI has access to which humans couldn't possibly measure? Once it defeats humans with mech hands on a mouse and keyboard with the humans having functional access to the exact same information, I'll agree. Obviously thats inevitable simply because they won't have our biological limitations, but there is some overhyping going on.

are you literally stupid? IBM won a chess game against the best player in the world by bruteforcing it on a supercomputer many years ago

I'm not , but seems like he's not being entirely honest. AI already soundly defeats the top Chess and Go players, those AI have progressed beyond brute force because it is unwieldy for transportation, and most of the states included in his claim are outside of the 'sensible games' category.

>Again, not knowing what the fuck "solving" a game means.
Glass houses and thrown rocks

.>AI already soundly defeats the top Chess and Go
Irrelevant. The best player in the world is not the same as solving a game.

>and most of the states included in his claim are outside of the 'sensible games' category.
The claim that there are more possible moves in chess than atoms in the universe, even if it is not in fact accurate (and since you have to change his definitions to get there, I hardly think that's intellectual dishonesty), says nothing about his point: that it's actually very difficult to solve a game even as mechanically simple as chess, let alone something far more complicated, of which there are many games. How would you "solve" something like Warhammer 40k, in which there are variable outcomes for identical play?

>Irrelevant. The best player in the world is not the same as solving a game.
I'm not sure if you missed a part of the reply chain, but I explicitly acknowledge that here >He never mentioned solved, but I think took the idea that since AI would be the only one doing the solving it would also surpass us in every game.

Basically what I'm trying to say is I'm not that interested in solving given we don't have adequate tools for understanding network reasoning or probably even the memory to retain the optimal move for every board state which would lead to certain victory. AI feats seem to increase every month and I'm more interested in discussing AI in games. I guess you are more interested in sticking to the solved topic. I guess we can part ways now since this seems to be an impasse.

>The "optimal strategy" is throwing 33.3/33.3/33.3% r/p/s, but the human brain is incapable of not falling into patterns, and using that you (or a computer) can get a better win-rate than 50%.
Actually, you can just memorize a long string of random numbers beforehand, and then decide what to throw by your position in the sequence.

>power grid goes out

No shit an AI wins. The computer is a cheating bastard.

How did you generate the string of random numbers? By hand? If so, then there's a pattern to it. If you got a computer to generate it for you, then technically you're playing the moves a computer came up with.

When AIs become advanced enough to pass a pen and paper turing test, what will be more enjoyable? Playing with a bot DM or DMing for the bots?

Computers can't even figure out how many possible solutions there are to a game of solitaire.

They dont take over for you completly though.
They only dodge skillshots and do some combos with a script.
That does not make it an instant win, because moba games in general are not primarily about mechanical gameplay but decision making and teamplay.

Now imagine an AI who can do everything perfectly. For someone who actually climbs ladder its fucking scary, because its all about improving and you will NEVER be that good.

There was a four month long period were LoL bots would kick the ever living out of a team that was less then gold if it was pass ~35 mins. They pulled some really good combos. Riot was toying with the idea of finally putting in 'hard' bots but they could not balanced them out to were they were a good next step after the 'intermediate' bots.

I think that was back in 2013.

>Veeky Forums starts discussing science of any kind
I get that technology will eventually reach the point where whatever crazy dream you can come up with will be a reality, but it won't be in your lifetime
>B-but muh chess and muh Go!
Its a weak argument as for why this things will somehow happen tomorrow, they won't happen tomorrow, they will happen right after your funeral and your grandchildren will enjoy them.

Honestly progress on the field of high level computing may well be coming to a end soon. Why? Because it moves forward on the back of hardware improvement. Hardware improvement moves do to groups needing it to be better.

The history of it goes as follows...

> military go into the market
> time jump
> clerical workers/record keepers go into the market
> gamer go into the market
> graphic design become common to the point of creating demand
> military starts going with off the shelves part thus no long drive demand for new high end hardware
> big data era starts
> mobile computing start becoming important
> Us army makes a supper computer out of 500 PS3
> big data learns from above that off the shelve parts do work for most of the types of work they do. End effect is that they stop creating demand for new high end hardware.

The only things left pushing hardware forwards is gamers and mobile users. Gamers may well drop out inside the next few years because effective photo realism is getting very damn close for pc's and laptops are only about 3.5 years behind pc's. High end laptops will hit effective photo realism in about 5 or 6 years. After that the only thing pushing hard forwards will be mobile users. However they may well fully do everything the customers want by or shortly after that time.

When that happens there will be far less profit ( or maybe no profit) to creating better hardware.

the real question is will the gaming industry or the porn industry be the first to create and interactive universe for you to play god in?

I just want the matrix as a game. So if u want porn you can just fucking load in the red lady.

>How would you "solve" something like Warhammer 40k, in which there are variable outcomes for identical play?

You solve games with random elements for each outcome. Solving a game doesn't guarantee victory, only that you have a sequence of moves (which is sometimes a branching path) that is always the most optimal.

Meanwhile in real world we can’t even make cruede predictions of conditions for protein crystalization.

Whynotbothmexicangirl.jpeg

>
RPS as played (a game of guessing someone's response via microexpression and hand motion by this robot youtu.be/3nxjjztQKtY RPS as a theoretical exercise with a randomizer you are correct.

>when an AI solves mtg in all formats

>Challenges AI to a game of Motherfucking Fireball Island