Computer Chess: How It Thinks!
Вставка
- Опубліковано 6 вер 2024
- Mistake at 2:14. It's 10 to the power 26 nanoseconds.
Chess requires intelligence and thought process, so how can a computer possibly do it?
A computer that is playing chess is not "thinking." Instead, it is calculating through a set of formulas that makes it to take good moves.
Further Reads:
1) electronics.how....
2) www.mayothi.com....
3) chessprogrammi...
Visit our website for interesting science news and blogs: atomstalk.com/
Follow Us:
Facebook: / protonstalk
Twitter: / protonstalk
Instagram: / protonstalk
Linkedin: / 14626400
Additional Image Credits:
www.wikipedia.com
www.pixabay.com
Music Credit:
www.melodyloops...
So the Computer expects that I will play what's best for me? The computer doesn't know me at all. 😆
Computer is just a machine it does not expect anything it just runs programs.
@@hydrolito you clearly did not understand joke
😂😂😂😂😂
@@hydrolito use brain son
@@hydrolito he means that he blunder a lot in chess.
It is only partially correct that computers can beat the best grandmasters because of better hardware. In reality, there has been a substantial development in the algorithms and huge improvements in how computers evaluate positions. More than 50% of an increase in the strength since early 2000s is due to these developments. Currently the best computer chess programs like Stockfish can beat world champion even if running on your phone. For that reason, chess players are not allowed to bring mobile phones to tournaments.
A phone is super powerful, it’s due to their hardware. Try running a chess engine on an intel 8080 with modern algos etc and it get shit on by Magnus
I have a playlist where I breakdown developing a chess engine, including the AI if anyone is interested
Yes please
Head to my channel and you should see it
I am interested.
trump is an alien
so are You- do You realize?
Looking at the source code for Komodo and Hiarcs, some of this doesn't seem accurate for all engines. Komodo doesn't put as much emphasis on piece value as engines such as Houdini, but instead focuses on positional advantage. This description seems closer to Rybka's earlier versions. The depth tree applies to all of them, but not exactly how described here.
"We'd be sitting around waiting for the damn thing to make its move 'til the universe ended" lol
Lol
The computer would have to be bigger than the entire universe...
lol have you played with lvl 10 chess bot still took a lot time to make a move .
But be ready for it, coz the move will be a check mate instantly😂😂
@@martinplayer23 😭 checkmate from the first move
2:16 there have only been 10 nanoseconds since the big bang? :o wow, i knew chess is an old game, but i never realised that it was that old!
yeah wtf is it talking about here?
as the universe i can confirm i'm less than 10 nanoseconds old
10^120 is a very outdated estimation of the number of possible chess games. It is actually even larger. Much larger.
Numberphile has a video about this titled "How many chess games are possible?"
Actually he was referring to board configurations, not possible games, although the number may have come from there.
In short its virtually endless at that point.
I worked out an upper bound once, it was 10^26000.
It was based on around 12000 plies, and something like a branching factor of 40.
The value is way over the top, but its at least an upper bound.
You are confusing the number of chess games with the number of positions.
@Colin Java
>I worked out an upper bound once, it was 10^26000
That's probably the worst calculation I've ever seen, a generous upper bound for the number of possible moves is 2^155.
It's actually quite impressive how wrong you are: you're about 10^25954 off a reasonable answer which is so far wrong that it's not possible to give any kind of visualisation of how wrong you are. Basically you didn't work anything out and you're an idiot.
My computer keeps killing his own queen
Maybe she nags a lot
Are you playing on level 10 the hardest level
The computer be giving you a handicap.
@@masa-qi8cx ouch
If the computer is level 1 on chess, then it is the opposite of what it is said in the video like instead of best move for the computer and worst move for you the computer will simply invert it so you can beat it.
If you're playing with a strong one, then it is a Sac or sacrifice which a human Grandmaster could also do so beware
So this is why chess computer is immuned against trap
Right
Except that chess computers didn't understand sacrifices in chess at first. But Deep Blue in 1997 made that rule obsolete.. Yes, 1997 is the year humankind started its march toward extinction..(just predicting, don't shoot me wtih a laser).
@@ZoolGatekeeper no that move was added by deep blue team
@@zaidalyaseen633 Yes, I have to admit there are suspicions concerning that particular move, cause it's the move that won the whole tournament.The move itself was not a move a computer would normally make. And having myself been playing against computers of that era, they had problems with sacrifices So, I agree with you.
@@ZoolGatekeeper yeah but i played the same game against chess.com computer and it sacrifices it with out no second thought
"a computer that is playing chess is not thinking instead it is calculating through a set of formulas"
I remember in one of Asimov's books, as early as 1956, where the protagonist would get told that "computers don't think; they calculate". Super interesting.
This leads to the important question that drives AI research: what's the difference between thinking and calculating?
Or, is there a difference?
Well, machine learning kinda changes that
@@JohnnyAdroit calculating means you already know all moves.. but has to decide which to choose from according to the level of the computer
Thinking means, you dont know any moves but you have to decide which to move according to your brain capacity
Literally same
@Electro_blob 2 A single neuron in your brain has no internal experience either. All it does is zap other neurons when enough neurons zap it. To me, there is no fundamental barrier to machine intelligence. One can imagine a much more powerful computer in the future that could simulate a human brain at a neuronal or even atomic level. If the simulation was accurate, would that simulated brain be conscious? What's the difference between a simulated thought and a real thought?
I don’t think the description is very accurate. Engines these days do not just use “brute force” calculation. They technically can also take psychology into account if a human inputs programming based on what he thinks will manifest as “psychological play”
possible positions are about 10^50 but not in games. 10^120 is the estimatet number of possible games.
So how does it work when you wanna make the computer not as good. Like reduce its difficulty? Do you just add in a function that has a certain percent probability of not choosing the best option?
I have the same doubt
TTS engine used in this video is probably the best I could find. With this voice, I've used many filters and tweaks to make it even better. Now, I don't see any problem in this audio. It's quite clear and understandable. I don't understand lots of people complaining about the audio. If you don't like narrations like this, please don't watch this video. Next time, I'll try to hire Morgan Freeman for narration.
Thank you and have a nice day!
I think lots of people are just fed up with technology taking over the world and are not big fans of robotic voices talking to them. There are so many videos out there where robotic narrators talk throughout videos and it gets so annoying and just disturbing. The thing is that by saying "if you don't like narrations like this, please don't watch this video", doesn't really make sense completely. Too many people these days say similar things for different situations in life, oh if you don't like it don't watch it, or if you don't like this then don't do that, or the "I don't care" attitude. If the whole world had the I don't care attitude and if you don't like it go away, then nobody would give a crap about anything and nothing would ever improve. You gotta be able to take the hard criticism and satisfy your audience. Anyways, sorry, haha, I like debating, don't have anything against your video, this is just my two cents as to why some folks find these robotic videos annoying. Peace and thanks for the video! Great video!
how about just do it yourself? aah?! :)
Dude he makes a perfect point. Be it a bit long.
well i liked the video and i have used tts too. which one did you use??
ScienceHook I didn't even notice it was a tts voice lol
Me: Yeahhh I've won the game
Computer: 10 inaccuracies, 4 mistakes and 7 blunders
I play on lichess and every so often use the stockfish program to analyze games. One thing I've found out about chess engines, such as my game yesterday, is that certain positions and a certain piece count still confound chess engines. To the side of the annotations is a meter that shows who is theoretically winning, white or black, and in this game white was winning with the meter at 75% white... so why at one point was the engine moving pieces on both sides in repetition? This actually happened twice during this game in the endgame, and both times I made a move to further the game along... both times it lead to a correct move.
No doubt chess engines are brute force, but today they're also very deep and make amazing moves and combinations that are nearly impossible to find, especially over the board, for a human... but they aren't perfect yet. ;}
At that point the game would be meaningless I think ;}
Have you considered the possibility of it being a software glitch on the end of Lichess? I have also had this problem occur to me when a position is already repeated once in the move list. I haven't tested the problem in another environment such as SCID yet to see if it's Lichess or Stockfish itself, but I am quite sure that it's unintentional behavior and not a limitation on the capabilities of a modern-day chess program.
That's probably because the game had reached something called a "fortress" position or at least some heavily defensive position. Computers still struggle with defensive positions that require a more positional approach rather than a tactical one. That's where humans come into play for the correct evaulation and approach. For instance, in a fortress position the computer will tell you that you're losing but you can throw that evaluation out of the window. It's a draw, the computer simply can't properly evaluate the position.
Even when I learned about MinMax and AlphaBeta back in 98 during my BA degree study, it was already ooooooooooooooold.
So theoretically, playing against a computer makes you better as compared to playing against humans
Yes and no , chess engine has been beaten by humans
Could quantum computers revolutionise chess?
since quantum computer has 3 bits in stead of 2 there are 9 combinations for each bit. For example: 000,001,010 and so on. For a 3 bit quantum computers, it’s still not possible to predict all moves. We don’t need to enhance our computers for this. We need to find a new and revolutionary algorithm to solve this.😊
Plz explain is it possible to beat such intelligent computers
Another important factor of chess engine is PERSONALITY: Should it be aggressive or defensive? Should it be careful with the opponent or indifferent? More importantly, should this personality be CONSTANT during the whole game, or changeable? Actually that is the reason behind the presence of some people with the chess software in some games against grandmasters: they change the PARAMETRS of the chess engine according to their estimation, e.g if it is open or closed position, if the opponent plays defensively or aggressively, if there is time pressure or not etc.
How personality is important? Let's suppose you end up with a pisition where there are two possible lines to move: One would give you advantage over your opponent with 3 points at least even if the opponent plays the best moves; the second one would give you either an advantage by 7 points if your opponent does not play the right moves, or you would lose by 1 point if your opponent plays the right series of moves.
The chess engine would definitely choose the first line, because it would (I am trying to resist the temptation to put he or she instead of it) automatically assume that the opponent would play the best moves if it chooses the second line.
On the other hand, some human players might prefer the second line: they could estimate that the opponent would not be able to find the winning series of moves, so they would take the risk and plays for a big winning opportunity (7 points is better than just 3 points).
I call the personalization of a chess engine a machine humanization. It will lead to more varied styles and more entertaining games.
Thats just a high risk high reward tactic that could be additionally supplemented with a capability analysis of the engine of its opponent to debileratly use it against him with this tactic. I dont see any "human" personality in that.
The engine calculates as if it was playing against itself, evaluating always the best move for both you and your opponent. If your opponent doesn't know about scholar's mate, you have a high risk high reward of going for it, but if he knows it, you get a worse position. An engine won't go for it, cuz it knows the best move is not to get mated and get a better position. If you will make the computer play moves that are worse if the best move gets played, it becomes worse
Personality = algorithm
@@fracapolligummala3548it's the same thing
How to be the best...
Play chess with someone online then copy his moves vs computer then make the moves of the computer vs your online oponent.
This is my idea
10^120 is for all positions, not only legal and possible positions in a game... so even without evaluation functions it wouldnt have to calculate 10^120...
Even if only 0.000000001% of those positions were legal and possible, that would still be 10^110 positions.
LKC9215 legal and possible solutions are estimatif ah 10^70-80
OK. Still not going to be easily done
LKC9215 Indeed at the end it changes nothing. I just wanted to precise this point because the video is not 100% correct but w/e
Constructive Criticism:
You should get a voice actor to voice the for the videos. The text-to-speech thing isn't really working for me.
Agreed
Agreed as well.
But 10 nanoseconds is 10^-8 (0.00000001) seconds. It has nothing remotely to do with 10^120, which is 1 with 120 zeros.
Also the phrase "10 to the power of nanoseconds" they used in their correction is terribly wrong (or not even wrong). Nobody ever uses such a construction.
Yeah I presumed that too, but it's still completely wrong. Is it so hard to calculate? 13B years times 3M seconds (per year) divided by 0.00000001 is still around 10^17 . More than 100 orders of magnitude away from 10^120. Jesus guys do you even multiply by 10.
this is a lot of nonsense. in no case are there 10 E 120 moves on the board, those that are possible are in reality a handful up to some hundreds per move, with only a dozen or so, maybe two, but not more, on average likely in the last case. The computer DOES think. It DOES define what the pieces are, what they can do, (that makes it as said a limited number, in fact peanuts, per player per turn) and its calculative "trees" are keeping a score PER initial possible move per piece for each next move in the row calculated down the line, each time adding a little tree that grows to the peanut bush of calculations Pieces stand for values based on their identity af what they can achieve. The apparatus of course rates extra points it adds either subtracts points for opponent or own check, loss of pieces, pin, treat, mate for the opponent (you see it starting to add up, it is rather complicated) and all this is done per position. The computer is aware where the pieces are. As said, it is the worst, minimalised nonsense, it explains nothing and tries to be snug whilst doing so. I am almost thinking it would be dangerous if you would believe any of this pretentious crap. There is zero knowledge in it, nothing true & indeed, it explains nothing.
This should be a fun project for AI grad students.
This is how mankind will end.
Max for computer survival min for human survival.
2:39 that part had me dying for some reason lol
Because he’s mocking something that threatens your ego.
@@SuperYtc1 says the planet based off my asshole.
oh.....i'm searching for the ways to memorize these algorithms ,But your video make me not to think that
UA-cam algorithm works magic. I was thinking about this earlier and it pops up right now
just remember there's only been ten nanoseconds since the big bang
Awesome video, this is how I found your channel. I'm glad you decided to narrate with your voice since this was made.
this is just a very basic chess engine, a very strong chess engine like the best in the world have opening sequences and pattern recognition built into it, it also will not compute every possible move, it will pick and choose moves to evaulate to save cpu
i see... i was dealing with a god👴🏽
😂😂😂😂
Gay
I don’t get how the computer makes a move and it can either be 9 up or 3 up
is the score in each bubble at the bottom of the tree the score for that move in particular or the total up to that point?
Theorically, if there are super computers with infinite ability of processing, all chess engines would agree to draw from the first move. Think about it.
Not necessarily. You assume it's always a draw. How do you know?
koodikotka Chess engines play because they do not know the final future possibilities of their moves. Now, do you agree that there is no winning chess game if both players play the right moves? I as human being could guarantee you this, even if I do not have the ability to analyze every single possible chess game. Why? It is simple logic: both players have the same pieces, and they both have the same rules to apply. You might ask: why do I play chess if I know that there is no DECISIVE winning game? The answer is this: As a human being, I suppose that my opponent (whether human or engine) is not perfect and I would use his/her/its mistakes to my favor. Of course, the human opponent would have the same idea so we both actually become "blunder catchers".
Chess engines do not think in the same way.
Chess engines, in their default parameters, think that the opponent is completely perfect and he/she/it would play the right move; but, and here is the catch, they do not calculate positions to the end of the game from the move 1, so they do not know if their moves would lead to a draw, win, or lost. If a super chess engine with super computer calculates all possible moves, and thinks that the opponent would play the right moves, it will find eventually that ALL GAMES WOULD END IN DRAW, and then, that super engine would ask for a draw from the first move (unless programmed to think like humans).
To recap: Humans play chess because they are trying to catch blunders, or mistakes; while chess engines play because of the uncertainty of the game's end.
No, I do not agree that there is no winning chess game if both players play the right moves. That's the point, it is not proven it's either win or draw if both players play the 'right' moves. At the moment, when we don't know any better, a 'perfect' game can be any of these: draw, win for black or win for white. Hence, you cannot say that super computers with infinite ability of processing agrees to a draw.
There are far less than 10^70 positions! 10^120 is an estimate for chess games. There is a huge difference because you can get the same position in multiple ways.
Example: [1. a3, a6 | 2. b3, b6] can also be achieved in different order (3 extra ways) or you can even move the knights/bishop back and forth (>10000 extra ways)
The positions can be reached forever 😂♾️ if there's no repetition
This game 😀 weird
I’m doing a huge school project about this! If anyone here knows something about the programming of this algorithm, please leave a reply :)
So, I am creating this tree structure at this moment. Difference is, mine is about the game ‘Mastermind’. For those who don’t know, mastermind is a game where there are 6 colours and 4 positions. This gives a total of 1296 (6^4) solutions. The problem i’ve run into, how do i put this tree that will lead to the answers, into a computer? I was thinking using if/then statements, but i have no idea in what language etc.
Does anyone know how i can put this tree structure into a computer that can perform these steps i made? I really appriciate all imput!
DankAsFuck you can try python if you want to do this easily. Python is probably the easiest language for these kinds of tasks.
And how do people know how many atoms there are in the universe?
Today: Best Human, so far - 2882 ELO (Magnus Carlsen, WCP) - COMPUTER: above 3300+ ELO
So if we study this instead maybe we'll play like computers in no time.
You have to be a very good calculator
Well, if you know how the computer is thinking, then we can beat it..
@@ZoolGatekeeper ehhhhhh dumb
Yr right, the computer is actually dumb. Homo sapiens is by latin definition, wise.. and built the Computer.
@@ZoolGatekeeper bruh u are so stupid if u think u can ever beat a computer at chess
Hi, thanks for this wonderful audio. Im not a native english speaker and I could understand everything in the video. Congrats!!!
DRAWING AN ENTIRE BOARD GOOD JOB
Also mostly chess bots that are 1300 below like nelson blunders alot
giving off queens but the 1200 and 1300 bots in the engine thing without personalities are just way better on not blundering
No mention of Monte carlo trees
I think I should quiet in front of computer. It always wins😭
Computer intelligence is just calculating ALL possible variatsion of the game...and choosing the most expected countermove? With no hesitation- coldblooded calculation of possibilitys. Thats booring....and interesting at the same time. Keyword calculation speed? Put two supercomputers againt each other...who wins- the one who starts or the one with higher voltage :)
draw
I'm suppose to be doing my hw but this one video keeps grabbing my attention
Your hardware?
Really keeps me thinking 🤔that fundamental chess principles known overtime are violeted and supurssed by the wonderful chess machines,I wish to understand and use this brute force calculation in a simpler fashion
But how it plays openings?
I would have liked to see an explanation of alpha beta pruning, that's the one I usually have trouble with.
There are some videos for it. Please check: www.google.co.in/search?q=alphabeta+pruning&oq=alphabeta+pruning&aqs=chrome..69i57.4463j0j7&sourceid=chrome&ie=UTF-8#tbm=vid&q=alpha+beta+pruning
Good be explanation on brute force approach to chess tree. But nowadays, there are more approach with better results now, e.g. neural networks, deep learning etc.
if you know no game no life, shiro can built the tree top to bottom
if this is true then how did she get puzzled by tet
More atoms than the entire universe? not in the entire universe, but in the entire "observable universe"...big difference.
No you are wrong …… we know how big the universe is from expansion from the big bang.
It doesn't explain why the AI changes the moves every single time. If you play the same moves, the AI doesn't repeat the same mistakes that leads to a loss. It changes the moves. There is this element of "randomness". What causes that "randomness"?
bruh can u teach other parametrs to, how to give number to position, not only materials
What a great video how did you know that stuff???????did you research it?? please tell me as soon as possible
There are books on this, articles on this, etc. There's actually a lot of into on this, and computer chess started in the 1960s I think. Technically, it probably started way back with Alan Turing, but as I recall, he manually applied his own algorithm for chess as he didn't have a computer he could run it on at that time ( this is from memory)
I didn't quite understand what the program is supposed to do
i got triggered by the way black and white squares were mistakenly drawn at 0:08
how do you mean?
no, that's the proper way to set up a chessboard. rules that I've heard so far are white tile in the bottom right corner, white's kingside castle is on the right and black's kingside castle if on the left, from the player's perspective. This board and piece setup follows those rules
edit: changed "bottom left" to "bottom right"
That last statement about the hardware is only partly true. The top alpha-beta chess engines employ iterative deepening, principal variation search, transposition tables, quiescence search, learned evaluation and many forms of unsafe pruning that were refined within the last 20 years. Yes, it is true that a laptop these days is pretty powerful, but stockfish can run on your phone. A lot of elo came from advances in the software.
This is the reason why computer are mathematicians genius
It's actually not like that any more. This video explains how computers played in the 80's
What about AI levels?
Most programs or apps have options for the AI's ELO before the game starts.
AI Level normally mean the depth of the search tree. The deeper the tree, the stronger the AI level is.
"which is why computers couldnt usually beat humans in the 1960's but now it can thrash grandmasters"
there were computers in the 60's????!?!???!?!!?
“There have only been 10 nanoseconds since the Big Bang.”
oh ok
But how does it get the score
To not get confused start the video by 2:26
btw, there have only been 10 nanoseconds since the big bang
kept thinking my alarm was righing hearing this song
i want to see this algorythm in the quellcode. this must very intresting :D
I find it cute that you used the verb "think".
Akademic explanaiton! Thanks!
I want to see the computer that thought this was passable text to speech work
In other words, the video maker is lazy asf and can’t even proofread themselves
“10 nanoseconds is the typical time required for each neutron to cause a fission event “ -Google ... there have been 4x10^26 nanoseconds since the Big Bang.
Computer is like doctor strange with time stone reviewing every possible outcome
That's pretty cool.
you could also use machine learning
(also, on an unrelated note, can people PLEASE shut up about the TTS? some people don't like using their own voice, don't have a mic or don't have a good mic. it's still a good video.)
Thanks for the knowledge 💞
I wonder if Magnus Carlson would loose against the most powerful chess program. Eg IBM Deep Blue defeated Gary Kasparov ⬛⬜
he would get crushed
Magnus Carlsen could get smashed by a iPhone engine these days.
5:25
"10 nanoseconds since the Big Bang"
Say what?!! 😵
Imagine the same AI from Bot ai of DOTA 2 doing chess.
The Question is this. In theory the computer play the "best move", so in the same possision computer should play the same move. This is not hapening anyway... So with near equality the computer is possible to choose other move
There are 6.022 * 10^23 atoms in one molecule.
No, it's the Avogrado number
Bruh, a molecule is a group of atoms, it's not a huge number of atoms ranging in billions or trillions though.
Thanks. It's amazing, isn't it?
I knew all of this I just like to watch your videos!
How does alphazero think?
alpha beta pruning is the best algorithm for creating a game like chess
@2:40....I expect inserting the word "damn" was an attempt at humor. It failed.
I disagree. I gave a chuckle at that part. It lightened me up enough to somewhat forgive the writer for calling stars "suns".
oh youre a flattard. my apologies for taking you seriously or thinking you had anything even remotely intelligent to add to the conversation.
woah who you getting angry at
the computer has 10 pieces? 😂😂😂
Wow the person who wrote the program is great
Cool video dude
thanks
hi thanks for great video. I was wondering what tools you used to make the video effect (like hand writing effects ). if someone knows please update.
Damn fuck, it is more confusing than playing chess
2:15 for example there have only been 10 naynoseconds since the big bang... are you sure? oh wait nvm. What a save! OMG! Wow!
Turn on the captions or read the description :P
Thanks, that is a very good explanation!
Hi, I run a small retro channel at ua-cam.com/users/perifractic and would love to feature some 3 stills from this video and credit you with a thank you link in the description. It seems to be marked on Google Images as public domain.
Whilst I have no doubt you'd be agreeable, I always like to check in with people in advance. But of course if you'd rather I didn't, please do not hesitate to say so too :)
Thanks so much for your contribution to the retrocomputing community.
Your friend in retro,
Perifractic
wow the same guy from 4chan greentext stories is making a foray into chess. Good for him.