I wanted to respond to two types of comments that have appeared more than once. I read all the comments and really appreciate when you all dig deep into these topics. First, we're missing some matchboxes because we don't actually need them! Some matchboxes work for two scenarios -- once for the board position they display, and also for the board position that is a mirror image of it. The computer learns both board positions at the same time, but yes, at first glance it appears as though I just left some out. Martin Gardner didn't think they were necessary, either. Second, Hexapawn is a much simpler version of chess, so terms like "checkmate" and "stalemate" aren't exactly the same. They're simpler, too. In chess, checkmate is achieved when there is no way for your opponent to move without the king being captured. A stalemate occurs when a player has no legal move. A stalemate results in a draw. So, when that occurs in Hexapawn, it has the trappings of a stalemate but has the result and the spirit of a checkmate -- the win is awarded to the player who moves in a way that creates a stalemate for their opponent. Because the situation results in a win instead of a draw, I thought it was more appropriate to compare it to checkmate, though it may have been clearer to avoid the language of "checkmate" entirely.
TheEverydayWizard it wouldn’t be hard to extrapolate this idea into what machine learning is used for today. Imagine if Kevin himself is a machine that can play thousands of games a second, a much more complicated “game” could be solved by the machine very quickly. Although usually it’s not a game, it’s something like recognizing whether a picture contains nudity or not.
TheEverydayWizard you’re right, this oversimplifies how machine learning works, but it’s a a pretty good introduction. You would need some pretty advanced programming understanding to comprehend how it is being used in more advanced systems. But the basic idea is essentially the same: a program “evolves” by keeping what parts of it get rewarded, and removing which parts of it are punished.
I thought about this as well. Technically I don't see why it wouldn't work the same way as Hexapawn, but in reality it would be impossible. Hexapawn has 3 pieces per player and is guaranteed to end after at most 7 moves, so you only need 24 boxes for the 24 possible scenarios. In Chess, on the other hand, you have over 100 million possibilities after just 6 moves. The "matchboxes" for this computer for all scenarios would need an impossible amount of storage.
thats just infeasible, its nearly impossible for you to get to the same exact position again, so the knowledge is useless. there are so many possible games that its impossible to just play them all. this is a rudimentary technique that only works on simple games that you can easily go through all the combinations.
@@grimmsoul3096 not even close. it’d be done by now if thats the case. the computer would have to play atleast each board position with each possible move. if it could be solved google’s alphazero would have done it. we are a far way off. maybe possible if someone ran one of the big super computers for a year but id still be really hesitant to say so.
I think this is actially my favourite video you ever made. This computer and how it can learn through positive and negative reinforcement is absolutely fascinating! Kind of mind blowing.
Me: **accidentally burns hand on stove** OW Random nearby person: omg are you okay?? Me: Yeah, yeah, I’m fine. Let me just... **Reaches into ear** **Pulls out green bead from brain**
As a teacher, this genuinely blew my mind. I have learnt about reinforcement and punishment in training, but never like this. Thank you, Kevin, for introducing this. I am definitely making this game to show my students, but not with Shrek.
Honestly. I write AI software for a DOD contractor and I can whole hardheartedly say, Keven nailed it. In this video, he explained AI, machine learning, better than I have ever seen it explained.
Dashie the Wolf I don’t see why people think flying cars are that important, advanced, and genius of an invention. In the 1900s they invented antibiotics.
The first computer game I ever played was Hexapawn, it was one of the applications that came with some IBM business computer in the late 70's. At that time, that a game could learn and improve was amazing. Especially to someone that didn't understand anything about game theory. I've done way more complex things since then of course, but I still have a fondness for Hexapawn.
The game cannot end in draw cause the player causing the state where no move can be made is the winner. So if played perfectly the game should always be won by the same side.
@@mentoslat6140 No the game can't draw if a computer would play against a computer someone had to start, but the algorithm isn't made for starting thats why the player always starts. A solution would be a random start in this case the result would be random.
The fact that you've made studying for my Game Theory exam a totally fun ride deserves praise! Awesome job, not only on this video but on many others across... the board
@@flippy-6725 Look it up on Google. They are... Quaint? In the "better off if they all got sent to a desolate island in the middle of the Pacific" sense.
Yep, because in this game who goes in second, if he, or it, knows all the right answers, always wins... ~You can basically do the same with the # game... but then it needs to play first, and if you know the moves, you can at least make it tie every time...
With tic tac toe, if you go first you will always win or tie unless you make a mistake. The game will result in 3 outcomes. You win early because they make a mistake, you win because there are 2 lines to block with one move which they obviously can't block both to, and a tie due to them also knowing the game and blocking one of those potential win paths early on in the game, limiting you to the first scenario, or a tie..
Really like this way of explaining how computers and genetic evolution work. Now you'll be happy to hear I have an extra "watch one of these videos" bead in my brain box!
But tic tac toe is way more simplistic. Anyone with half a brain cell could easily become good enough so that every game ends in a tie- even against the smartest computer.
Thank you so so much! It's almost like YT knew that I made hundreds of Google searches about genetic algorithms and symbolic learning, and is now suggesting me this amazing explanation! You made this seemingly endlessly complicated subject seem simple, for which I am very thankful!
Rocket Grunt Mason , It took me a while to understand this comment as it repeated myltiple times in the comments section.. Wow... since when are people are this touchy? Or are we still in the trigger-troll era?
@@AAA-bo1uo Iv read your commet six times now and Iv come to the conclusion that you were born stomache first, use your parent's credit card and buy a few chromosome DLC's cuz it looks like you need em
Kevin, fantastic video. As a software engineer I can really appreciate when someone teaches a concept like machine learning in the tangible world. I’d love to make this for my kids and nephews, any idea where I can get the images for the matchbooks?
Inspired by this video, I made a program that learns how to play Tic Tac Toe. After a few hours of trial and error, I was able to do it. At first, it has no idea what to do, but after lots of trials, it gets to a point where it will never lose, only win or tie. Not only will it learn not to make a losing move, it will learn not to play into a potential impossible situation, going back as far as 2 moves. I also made it so that when it plays a winning move, it learns to only play that move.
Nostalgia! Back in high school (mid-1970s), I wrote a Hexapawn program. I modified it so that the board size could be made anywhere from 3x3 to 8x8 and it would remember all the results for each size board.
I've been wanting to program a genetic algorithm in Java, and this video helped me understand the concepts of it. I think next time It'll turn out better. Thanks for the amazing video! :)
Machine learning at its simplest form! Man I think there were more smart people in the past than the present lol. But no doubt that technology is improving everyday.
Yeah, but this version is smaller and takes significantly less time to train. MENACE (the tic tac toe version) is cool, but this version could actually be used as a classroom demonstration of machine learning.
I really love your videos. I have always wanted to make a learning computer ( AI ) And now i can create my own with only matchboxes and beads. I just want to say thank you so much for all the incredibly smart videos and keep up the good work! Also btw. i just subscribed! Yay!
Standupmaths made a copy of MENACE and had it play tic-tac-toe with many people during the International Maths Festival (I think). He did both punishment and rewarding- if the next turn lost, remove that bead. If that move won, add another bead. The experiment was done, reset, and done again the next day, and the results were different.
I wanted to respond to two types of comments that have appeared more than once. I read all the comments and really appreciate when you all dig deep into these topics.
First, we're missing some matchboxes because we don't actually need them! Some matchboxes work for two scenarios -- once for the board position they display, and also for the board position that is a mirror image of it. The computer learns both board positions at the same time, but yes, at first glance it appears as though I just left some out. Martin Gardner didn't think they were necessary, either.
Second, Hexapawn is a much simpler version of chess, so terms like "checkmate" and "stalemate" aren't exactly the same. They're simpler, too. In chess, checkmate is achieved when there is no way for your opponent to move without the king being captured. A stalemate occurs when a player has no legal move. A stalemate results in a draw.
So, when that occurs in Hexapawn, it has the trappings of a stalemate but has the result and the spirit of a checkmate -- the win is awarded to the player who moves in a way that creates a stalemate for their opponent. Because the situation results in a win instead of a draw, I thought it was more appropriate to compare it to checkmate, though it may have been clearer to avoid the language of "checkmate" entirely.
Doodoo
Now do this with chess and let us know how long it takes for the computer to reach perfection xD.
You should not have used the word checkmate. Other then that, great video!
Vsauce2 you need to wear a Shrek shirt in the next vid
this video seems like a passive-aggressive way of communicating with someone and more importantly teaching them something
Machine learning visualized in 13 minutes for mere mortals without math. Great video.
But there is mathboxes!
WARNING! I have 2 (!!!) dangerously HOT girlfriends and I show them off! Thanks for being jealous, dear dude dantalion
TheEverydayWizard it wouldn’t be hard to extrapolate this idea into what machine learning is used for today. Imagine if Kevin himself is a machine that can play thousands of games a second, a much more complicated “game” could be solved by the machine very quickly. Although usually it’s not a game, it’s something like recognizing whether a picture contains nudity or not.
TheEverydayWizard you’re right, this oversimplifies how machine learning works, but it’s a a pretty good introduction. You would need some pretty advanced programming understanding to comprehend how it is being used in more advanced systems. But the basic idea is essentially the same: a program “evolves” by keeping what parts of it get rewarded, and removing which parts of it are punished.
And Shrek
Forget about robots, learning matchboxes are gonna take over the world.
Lane Girl just remove the “take over the world” bead.
Raid Shadow Legends is gonna take over the world
AI's are just a more complicated version of the learning matchboxes!
That one kid who doesn't understand anything : bUt MaTcHbOxEs CaN't WaLk
It would be bigger than the earth one of them but it could work
imagine challenging a guy to a game of checkers and he rolls up with 304 matchboxes of pure perfection plays.
Yes
Um, actually, it would take way more than 304 matchboxes for checkers 🤓
Tic-tac-toe is 304, checkers is really.. more than that.
@@사다드-j6x ikr thats what im saying bruh
Imagine chess with the 10⁵⁰ matchboxes -_-
I bought the Shrek Chess set used off Amazon and it smelled really bad. Thought you should know that! Bye!
Very nice.
*This is my swamp*
Da fark
Did it smell like onions and swamp musk?
How about the “Go board” and early computers where just huge calculators/cryptography machines.
So the game ends with the winning player yelling "Shrek Mate!"
200th like
64 points for that comment
Shreksapawn
Yeet
Sherkess
Imagine how many match boxes and time it would take to build the perfect chess computer
Im trying octopawn
I thought about this as well. Technically I don't see why it wouldn't work the same way as Hexapawn, but in reality it would be impossible. Hexapawn has 3 pieces per player and is guaranteed to end after at most 7 moves, so you only need 24 boxes for the 24 possible scenarios. In Chess, on the other hand, you have over 100 million possibilities after just 6 moves. The "matchboxes" for this computer for all scenarios would need an impossible amount of storage.
thats just infeasible, its nearly impossible for you to get to the same exact position again, so the knowledge is useless. there are so many possible games that its impossible to just play them all. this is a rudimentary technique that only works on simple games that you can easily go through all the combinations.
probs somewhere close to a year
@@grimmsoul3096 not even close. it’d be done by now if thats the case. the computer would have to play atleast each board position with each possible move. if it could be solved google’s alphazero would have done it. we are a far way off. maybe possible if someone ran one of the big super computers for a year but id still be really hesitant to say so.
Kevin: *wins two rounds*
*Looks at board*
*Throws the third game*
"Phew, that was close"
So wanted him to win that third game, but you know what is the probability of that? What is PROBABILITY
KK-
ohimnotsayingthat
kks
What about ss
SSkkss kkss
Police: Kevin, you're arrested
Kevin: What is *ARRESTING* ?
lol
*looks into camera*
.... or am I?
Let's get into the basic definition of arresting.....
*escapes arrest using knowledge*
LMAO HONESTLY
I think this is actially my favourite video you ever made.
This computer and how it can learn through positive and negative reinforcement is absolutely fascinating! Kind of mind blowing.
Same bro, the best video ever made.. When i was little, i'm wondering how computer learn.. This video is like a light in the dark❤
I really miss when he made math/logic/computer videos for fun.
I understand that crime is important but man...I miss these
Why this game gets smarter:
0% beads
300% Shrek
Hello, Justin Y
2nd reply. Heh.
what would happen if Shaggy played the game?
Uhh...
True.
Kevin is the only one who remembers his password
*pbuttword
Did you just passwear?
Your password is p@$$w0rd
😂
🤣🤣😂😂😆😆😀😀😊😊☺️☺️🙂🙂😌😌😐😐😮😮😯😯😓😓😔😔🙁🙁☹️☹️😟😟😦😦😢😢😭😭😭😭😭😭😭😭😭😭😭😭😭😭😭😭
So if you made two computers play against each other, the second player would eventually always win, right?
WRONG
Try it
@@rickastleyunofficial4455 OR IS IT ?
(VSAUCE INTENSIFIES)
earth would break in two
Yes, i built a computer that plays white and black always wins
win board: KK
*sweats profusely*
Using this logic we can deduce that Kevin cannot win more than 2 times in a row
What’s wrong with being in the
Kool kids Klub?
It's the Krusty Krab.
Win board :KKSKKSKSSKSSSS
and ioop ksksksksksk
@@abstractdomino8423 the Krusty Krab Krew
Me: **accidentally burns hand on stove** OW
Random nearby person: omg are you okay??
Me: Yeah, yeah, I’m fine. Let me just...
**Reaches into ear**
**Pulls out green bead from brain**
under-rated comment
Way underrated. That was was good.
not really
That’s hilarious and so underrated
but in reality he's actually a matchbox
Me: Calmly eating ramen.
Kevin: "WHAT IS LEARNING??"
Joe
“I’ll put a K here every time I win.”
Me, round three: “COME ON COMPUTER YOU BETTER”
Lmao I just realized........
I'm black too
@@spoopy2180 How are you just now realizing you're black?
@@mr.dudebro2609 no on round 3 shell get kkk and we all know what that means to black ppl, plus issa joke
@@mr.dudebro2609 also I said I'm black too, but I said I realized as in I realized the joke
@ButteredPopcorn (it's a joke)
*Ancient Greece building a pc colorised*
“Ay yo dawg what are you using for your cpu”
“Beads on a stick”
“Niiice which model”
“Abacus”
Everyone knows the xrocks 3600bc has better graphics
Not to ruin the joke but abacus was used in ancient china
@@nicholasaldrich8419 you did, you did ruin the joke, go back to your corner.
Abacus master race y'all!
Guys, look, they've invented a new model: The papyrus-tRanSlucent !
For those who wanted to know, the song during the game montage is called Synthetic Life by Julian Emery. 😁👍
Nobody asked
The two most recent comments were unnecessary
@Gurnaj Virk Lukas R said it best "Nobody asked"
@@lukasr1166 Aaaaaand who asked for your input?
Thanks! I was wondering about that! 🥰
As a teacher, this genuinely blew my mind. I have learnt about reinforcement and punishment in training, but never like this. Thank you, Kevin, for introducing this.
I am definitely making this game to show my students, but not with Shrek.
What's wrong with Shrek?
if i was one of ur students i would greatly appreciate having shrek in it
But what about shrek? :(
shrek is too good to play with foolish mortals
Keep Shrek, he will motivate your students
Soldier Guy: Your Father died in action during th--
Kevin: *What is.... DEATH?*
Lol. I noticed he does that sometimes
🤣🤣
Death is the state of not being alive
@@bryceyork4173 What is... ALIVE
Kerat Alive is the state of not being dead
Honestly. I write AI software for a DOD contractor and I can whole hardheartedly say, Keven nailed it. In this video, he explained AI, machine learning, better than I have ever seen it explained.
1900s: we’re gonna have flying cars in the future
2019: Shreksapawn
Dashie the Wolf I don’t see why people think flying cars are that important, advanced, and genius of an invention. In the 1900s they invented antibiotics.
Planes and helicopters are essentially flying cars. SHREKMATE.
100th like uwu
What is your favorite game?
9 year olds:fOrTnItE
Kevin:S H R E K S A P A W N
M I N E C R A F T
Ha Tran u r correct Minecraft
For- wait what? 9 year old playing FORTNITE?!
No. Unacceptable. No.
*OH NO! CHILDREN ARE ENJOYING THINGS? THIS MUST BE STOPPED!*
No nine year old would play a game that isn’t popular.
Wish my DOTA 2 team mates were as smart as these match boxes
You have to remove the flawed cells though
Fair point
Forget DOTA 2 think about Among Us
@@usualunusualkid7149 yeah
@@usualunusualkid7149 yeah
The first computer game I ever played was Hexapawn, it was one of the applications that came with some IBM business computer in the late 70's. At that time, that a game could learn and improve was amazing. Especially to someone that didn't understand anything about game theory. I've done way more complex things since then of course, but I still have a fondness for Hexapawn.
I want to see a computer vs computer version.
I think it will eventually keep ending in draws
The game cannot end in draw cause the player causing the state where no move can be made is the winner. So if played perfectly the game should always be won by the same side.
@@mentoslat6140 No the game can't draw if a computer would play against a computer someone had to start, but the algorithm isn't made for starting thats why the player always starts. A solution would be a random start in this case the result would be random.
@@Swompy I realized the game doesn't end in a draw when someone has no moves, so you're right
They will learn until eventually every game ends with player 2 winning - it only takes maximum of 6 moves for him to win, but 7 moves for player 1.
What is a computer?
School textbooks:A computer is a electronic device that manipulate data.
Vsauce2:Shreksapawn!
The fact that you've made studying for my Game Theory exam a totally fun ride deserves praise! Awesome job, not only on this video but on many others across... the board
Shrek is easily the most important part of the game
230th like, first reply, and this is an underated comment XD
😱
I was worried kevin would win when it was just KK
You mean SS is better 🤔🤣
KKK
?!
I was about to say the same thing 😂😭
Dont understand why he didnt just right a K and and S then make tally marks....
@@ditzfough To show an increase in the computers victory over time
10:43 "add another get good grades bead to your matchbox head computer."
thats a brand new sentence if ive ever seen one
Vsauce2: *But what is a SHREK?*
LonelyPoni Shrek is love, and life😏
And a meme
*But* *what* *is* *a* *meme?*
God
Everyone always asks what is Shrek but nobody asks how is Shrek
But I’ll do you one better: why is Shrek?
when you create an ai with matchboxes and beads
And Shrek
@@alexlau5146 most importantly shrek
More like you train a model that is stored in matchboxes via beads.
Sorry, computer scientist and it is helpful to be more precise about what AI is.
@@jlaw131985 This is the easiest explaination of how machine learning works I have ever seen.
But fir--
Me: *presses furiously on the right side of my phone*
But fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HEREBut fir--
Me: RAID SHADOW LEGENDS HERE
Can we all appreciate Kevin for remembering his channels password
Unlike Micheal
Your game gets smart but your head gets dumb
so much to lose so much to win so what's wrong with taking a bad bead?
You'll never lose if you don't go
You'll never win without a bad roll
Hey now
You're a Shreksapawn star
Get your game on
Get braaain!
Hey now
You’re a winner
Get 1 bead more
You’re smarter
All that bad bead is gone
6:50 we all know he just wanted to avoid getting demonititised
Good thing Kevin didn’t win 3 times in a row
Why?
@@tahairfan4630 if u don’t know ya don’t know
@@skidoodle2150 ????
kevin writes 'K' when he wins, so if he wins 3x in a row
@@ravinerenegade so if he wins 3 times in a row what? I don't understand
Good thing you didn't win the first three games, you'd've got demonitized.
He came so close to spelling it, lol
What is KKK?
@@flippy-6725 Look it up on Google.
They are... Quaint? In the "better off if they all got sent to a desolate island in the middle of the Pacific" sense.
jotabeas22 ohh... I knew the "KKK" I just didn't knew they were called like this.
Can we take a moment to think about the freak of the word “you’d’ve”?
This is actually how chess computers learned to play their end-game from the late 1980's onwards
8:55 “I dont have to worry about these matchboxes quitting, and getting upset, and slamming the door in my face and telling me im not their real dad”
.... Its sounded suspiciously specific..
R/oddlyspecific
Why are you doing a r/ if this isn't reddit
@@ItsWhiteout Who are you talking too, me or the person I replied to?
@Carter Lynch The Hot Dog Vsause theme plays*
That bead computer is still more powerful than mine...
WhaT CompUter yoU GoT?
but can it run Crisis?
Kommentor Postker A bead computer only has beads. you think BEADS can run a VIDEO GAME??
@@mariafe7050 yes.
@@mariafe7050 technically even you can run a video game (a very slow one, but it's a video game nonetheless!)
4:28 ah yes the Lord farquaad opening
we all know why kevin let the computer win the third round
( ͡° ͜ʖ ͡°)
that’s exactly what i was thinking lmao
Same
DEMONETISED
Woooosh me, idc, but can someone explain the joke? It doesn't make much sense to me
You are ALWAYS weirdly creepy Kevin. :) and it's ok
This whole time that shrek on the shreksapawn board was staring into my soul.
The game gets smarter, meanwhile, I become dumber and dumber
a classic scenario of my life
An organic organism gets dumber while an object gets smarter. How does that feel?
Hey, How are you Justin Y 2?
Didn't I just see your comment on a Gamexplain video
Justin? Christopher moon? It's all making sense now!
So basically, After a couple of rounds it is impossible for the human to win.
Yep, because in this game who goes in second, if he, or it, knows all the right answers, always wins...
~You can basically do the same with the # game... but then it needs to play first, and if you know the moves, you can at least make it tie every time...
I don't think so. I think so because Tic-tac-toe is nearly the same as Hexapawn; however, I cannot guarantee my answer.
With tic tac toe, if you go first you will always win or tie unless you make a mistake. The game will result in 3 outcomes. You win early because they make a mistake, you win because there are 2 lines to block with one move which they obviously can't block both to, and a tie due to them also knowing the game and blocking one of those potential win paths early on in the game, limiting you to the first scenario, or a tie..
This game is rigged for second player, yesh
With tic tact toe, you can tie,
Hexapawn does not have a tie state..
my favorite thing is how this guy doesn't have some fancy intro with music and an animation, he just goes VSAUCE!!!!!!!
VSCO Girls really like his scoring system.
SKSKSKSKSK
lol
lol
lol
lmao
lol
GREAT video. My whole family (8 of us ranging in age from 2 - 43 years) watch your videos over dinner. Love them!!
You need to make more of these. They're incredibly amusing.
Kevin didnt win for three rounds in a row
Coincidence?
I think not...
IT WAS GONNA SAY KKK!!!
@@MrPoop-if3cs What about SS by the computer?
@Rishab Tirupathi lol
Introducing the new smart iMatchbox! Only $9,999.99!
Love the comment :D
basically sums up apple
Yeees tou are amazing
Lol
The way in which you articulate and illustrate this concept is top notch. Well done.
Thanks for the spare brain cells, really appreciate it!
@@RyanBoggs He took some out of mine I had to many. The Doctors say I should stop Snorting Beads.
1960: Lets make a computer to master tic tac toe
2019: SHREKSAPAWN
I always leave videos more confused than before! Keep up the good work!
Really like this way of explaining how computers and genetic evolution work. Now you'll be happy to hear I have an extra "watch one of these videos" bead in my brain box!
Me: Kevin got the new game!
You: What?
SHREKSAPAWN
Yeah shreksapawn
0:13 OMG I DIDNT KNOW MY BOYFRIEND WAS ON VSAUCE
"No he's mine" man I wish I could say that but imma boi
@@applelmao1973 what's stopping you? 😏
@@manioqqqq ayoooo 🤨📸
lemme take my lil matchbox computer to a worldwide hexapawn championship
Shreksapawn
S H R E K S A P A W N
I was so scared of Kevin winning 3 times in a row.
This is one of the best representation/visualization I've seen for this concept
Title of the video: The game that learns
Vsauce2: WHAT IS A COMPUTER
Why isa computer.
@@ars8223 🥁 ba dum TSSSS
This is an excellent video. Absolutely LOVE how you used the match boxes to show the positions and movements. Very well done!!
5:58 mans was so close to writing kkk
Matt Parker did a similar thing using matchboxes for tic tac toe
Thank you for this comment, I assumed this video was going to be exactly the same thing and would've skipped it.
and to be honest his was better
Matt actually built Menace, this is a simplified version of it.
But tic tac toe is way more simplistic. Anyone with half a brain cell could easily become good enough so that every game ends in a tie- even against the smartest computer.
true, but it's also a game you are much more familiar with and computer's progress can really be seen
At 8:52 I was really hoping he would say OR ARE THEY?
Ninja Dominator or did you?
@@mrbenoit5018 You didn't mean "or were you", OR DID YOU?
Thank you so so much! It's almost like YT knew that I made hundreds of Google searches about genetic algorithms and symbolic learning, and is now suggesting me this amazing explanation! You made this seemingly endlessly complicated subject seem simple, for which I am very thankful!
Me: *see kevin win twice in a row and put a K down*
Racists: *On edge of seat*
Lol
Rocket Grunt Mason ,
It took me a while to understand this comment as it repeated myltiple times in the comments section..
Wow... since when are people are this touchy?
Or are we still in the trigger-troll era?
@@AAA-bo1uo Iv read your commet six times now and Iv come to the conclusion that you were born stomache first, use your parent's credit card and buy a few chromosome DLC's cuz it looks like you need em
@Rocket Grunt Mason I've* comment*
@@rocketgruntmason3269 *_-no-_*
Also I don't get it
YES, EMBRACING THE MEMES.
THIS IS WHY I LOVE VSAUCE
*and the learning of course...
This was a fantastic tutorial on "What is Reinforcement Learning?" Next time someone asks me to explain this concept I'm sending them to this video.
*fantastic
@@blakksheep736 ROFL. Thank you my Editor at Large. I have corrected.
This was amazing and very informative! I’m really into AI, so this got my mental gears turning. Yet another bead for the Vsauce 2 channel in my brain!
What is a computer???
Me: one that computes
Ba-dum dish
What's a computer?
*Apple iPad ad plays*
Well, technically that's correct.
Is this supposed to be a joke?
dewolen yes
So glad that Martin Gardner invented hexapawn! He was a true lengend.
No one:
Score board: SKSKSKSKSKSKSKSKSKSKSKSKSKSKSKSKSKSKSKSKSKSKSKSKSKSKSKSKSKSKSKSK
It's obviously KKSKKSKSSKSSSS
@@mariafe7050 please tell me that was a joke...cause...you..i...
and i oop-
maria fe
no it’s SKSKSKSKSKSKSKSKKSKSKSKSKSKSKSK
The score board is obviously a VSCO girl
Next question "What is a woman?" "I dont know" - proceeds to cry in a corner-
-proceeds- -to- -cry- -in- -a- -corner- its okay fam
hearing "bead" and "chess" together reminds me of a game that "speaks for itself"
OH GOD why would you bring that up…
NAH
i cant get shreksapawn star out of my head now
had a big uh oh moment when i thought kevin was gunna get 3 wins in a row. suddenly the video'd get flagged
we got some "SS"s going tho
Kevin, fantastic video. As a software engineer I can really appreciate when someone teaches a concept like machine learning in the tangible world.
I’d love to make this for my kids and nephews, any idea where I can get the images for the matchbooks?
I posted the templates for you to download for free on Twitter!
Inspired by this video, I made a program that learns how to play Tic Tac Toe. After a few hours of trial and error, I was able to do it. At first, it has no idea what to do, but after lots of trials, it gets to a point where it will never lose, only win or tie.
Not only will it learn not to make a losing move, it will learn not to play into a potential impossible situation, going back as far as 2 moves. I also made it so that when it plays a winning move, it learns to only play that move.
love your vids! Also, 2:58. In chess, when opponent has no legal moves it's called stalemate and counts as a draw
Shrek holds the power of the universe he is one and everything for Shrek is love Shrek is life
How does this have zero replies bro
Nostalgia! Back in high school (mid-1970s), I wrote a Hexapawn program. I modified it so that the board size could be made anywhere from 3x3 to 8x8 and it would remember all the results for each size board.
I've been wanting to program a genetic algorithm in Java, and this video helped me understand the concepts of it. I think next time It'll turn out better. Thanks for the amazing video! :)
Kevin: WHAT IS A PARADOX?
Also Kevin: WHAT IS A COMPUTER?
Also also Kevin: WHAT IS LEARNING?
My dad when i want a new pc: 24 matchboxes is computer
Machine learning at its simplest form!
Man I think there were more smart people in the past than the present lol.
But no doubt that technology is improving everyday.
Dumb people didnt have much access to media in the past and honestly that may not be a bad thing😂
Also rip natural selection that leans towards the intelligent (in developed countries at least)
I remember when Matt Parker did this with tick tack toe, and a crowd.
Yeah, but this version is smaller and takes significantly less time to train. MENACE (the tic tac toe version) is cool, but this version could actually be used as a classroom demonstration of machine learning.
I really love your videos. I have always wanted to make a learning computer ( AI ) And now i can create my own with only matchboxes and beads. I just want to say thank you so much for all the incredibly smart videos and keep up the good work!
Also btw. i just subscribed! Yay!
Standupmaths did the 300+ matchbox MENACE a couple years ago
I knew I had seen something similar on youtube already, but I didn't remember who had done it, thank you.
*Hey Vsauce, Kevin here.*
Thanks!
Woah, where’d you get the Shrek amiibo?
It's not an Amiibo, just a figurine with a base.
_or am I missing the joke?_
Ogres are like onions. They’re _smart_
uh
huh
h
u
h
During quarantine I built this and I am now going to build the big MENACE
Standupmaths made a copy of MENACE and had it play tic-tac-toe with many people during the International Maths Festival (I think). He did both punishment and rewarding- if the next turn lost, remove that bead. If that move won, add another bead. The experiment was done, reset, and done again the next day, and the results were different.