Full podcast episode: ua-cam.com/video/oJNvxYEcVAY/v-deo.html Lex Fridman podcast channel: ua-cam.com/users/lexfridman Guest bio: Hikaru Nakamura is a chess super grandmaster and is currently the #1 ranked blitz chess player in the world. He is also one of the top chess streamers on Twitch and UA-cam.
I feel like to answer the last question of the clip. The bot did thousands of calculations every second and decided that for now this was mathematically the safest move and has the most potential for future set up. I could be talking out my ass though.
Hey Lex, I'd love to see a sit down with Yasser Seirawan while you're on this Chess binge. He's such a captivating story teller and has a world of experience of Chess from the days of Garry Kasparov and his ilk
He said that because he had no one on his level during this time, infact, he felt as if he wasted his life because his parents wanted him to be something else
This quote is kind of outdated. Morphy said this because at that time (the 1800s) if you were really good at a board game like Chess you were seen as a loser because you weren't good at something that was gonna further society.
Best thing I’ve learned in chess came from Mikhail Tal quotes. “You must take your opponent into a deep dark forest where 2+2=5, and the path leading out is only wide enough for one.” “There are two types of sacrifices: correct ones, and mine.”
@@sriracha9729 1) this is such a perfect example of chess, at the higher levels of chess, you are constantly trying to take your opponent out of the prep they know (bright forest) into somewhere they dont know/prep they dont know (dark forest). 2+2=5 meaning the conventual rules that everyone knows stop making sense, an example being knights at the edge of board being a better move only during circumstance, hell maybe even a knight on a corner being the best move. "the path leading out is only for one" is self explanatory, but it just shows how some games are very aggressive, and other are very drawish, making the choice of you pushing for a win is also you being pulled for a loss. an amazing quote. that really shows how beautiful chess is.
@@jaytorr6701 honestly i would have expected clips from podcasts to trend immediately, cause most people simply don't have time to watch entire podcasts
I asked ChatGPT this: Why is H2-H4 a good move in a chess opening? In a chess opening, H2-H4 is a good move because it allows the knight to move to the more active f3 square. This move also helps support the pawn on the e4 square, creating a strong center for White. Additionally, H2-H4 can be used as part of a fianchetto, where the bishop is developed on the g2 square, which can lead to a strong kingside attack. Overall, H2-H4 is a useful move that helps White develop their pieces and create a strong position on the board.
@@marnixvlot1868 Yeah, that sounds pretty good. What i meant by Machine Explaining though is...as A.I becomes more intelligent and more self aware will it always choose the best option that benefits the user or will it choose an option that benefits itself? Decisions, decisions...
@@marnixvlot1868 It's actually pretty funny how bad of an explanation that is: pawn to H4 doesn't help the knight to move at all and blocks it from moving if anything, H4 is really far from E4 and I don't see how moving an edge pawn is strengthening the center, and if you move H2 you still can't move the bishop to F2 so everything it said is basically nonsensical It's pretty interesting to look into the ways these chat AI's fail, since they're not designed to think like us or to reason or to come up with correct answers, but rather to write like we do. It often ends up with a lot of "Escher sentences", where at first glance it sounds like it makes sense until you think about it; for example, the sentence "more people have been to Germany than I have" makes literally no sense Funny enough, it's these kinds of results/interactions that show the need for more emphasis on machine explanation like the OP was talking about. On being able to ask Chat-GPT why it thought that moving to H4 would help support E4. Of course, the real reason is just that it's mimicking how people talk about chess, but it itself wasn't trained to play. Unfortunately, explainable AI is far less developed and researched on right now, so we can't know for sure
@@diegotejada55 It's also interesting how they both thought is was a decent explanation. It highlights two problems as far as I can tell at the moment: 1. People are largely content with content that "sounds right" about a given subject. 2. The less a person knows about a subject, the higher the chances for number 1. AI is amplifying this age of information, and we need critical thinking now more than ever.
Here is chatGPT 4o’s explanation for why H4 is a fine move: In the position shown in the video, Hikaru Nakamura is discussing why moving the pawn to h4 might be considered a good move despite seeming unconventional. This is an example of modern chess theory where computers have significantly influenced human understanding. Here are a few reasons why h4 could be considered a good move in this context: 1. **Opening Principles and Flexibility**: While traditionally the advice is to control the center with pawns, developing pieces, and ensuring king safety, modern engines sometimes prioritize piece activity and flexibility. The h4 move might prepare for future attacks on the kingside, potentially supporting an h5 push or opening lines for rooks. 2. **King's Safety and Aggression**: By pushing h4, White could be planning to castle queenside. This move can serve as a preparatory step for an aggressive kingside pawn storm. It signals White’s intention to launch a strong attack on the black king's position if Black castles kingside. 3. **Preventing g5**: The move h4 could be aimed at preventing Black from playing ...g5, which could disrupt White’s control of the center and potentially kick out a knight on f3. By playing h4, White is controlling the g5 square indirectly. 4. **Psychological Aspect**: Unconventional moves can put psychological pressure on opponents, especially if they are not well-versed in handling such positions. It can lead to errors if the opponent overreacts or misjudges the intention behind the move. 5. **Specific Tactical Ideas**: There might be specific tactical ideas that justify the move in this exact position, which aren't immediately obvious. Computers are excellent at finding deep tactical resources that might not be apparent to the human eye initially. By analyzing the position more deeply with a chess engine, specific tactical nuances can be revealed that justify the move. Chess engines evaluate positions on a much deeper level, often seeing the long-term strategic potential of a move that humans might overlook at first glance.
The Alpha GO movie is great for understanding how AI thinks. It transformed the approach to GO. When they figured out its strange approach to the game they finally realised. AI doesn't care about winning by a huge margin. It figured out that winning by 1 point still means a win. So, once it has pieces in place to win by 1 point, it can move on the another area of the board. Human players were concentrating on capturing as many pieces as they could, wasting time in an area while GO was setting them up for a loss in another area. Moving the Pawn up along the edge makes no sense to a human thinking just 5, or 10 moves ahead, but AI probably figured out that it will help isolate the King in 25 moves, or at least shift the odds from 50:50, to 50.1:49.9 in its favour.
This conversation makes me think about people today growing up with GPS. Listening to the maps voice telling one to turn right, drive for three miles, then turn left, and you will reach your destination. I used to study physical maps and figure out which way to go. This, to me, is the human element difference Hikaru is talking about.
In your case i think it makes you smarter because going from looking at the map to determining location and finding a path and remembering the path requires logical thinking almost as if you are a computer computing something. With automated voice instructions theres no logic Chess is different because regardless of the approach whether that be ideas and plans vs calculation both require logical thinking
@@blinkers88 Different areas from what the AI solves, for instance it can solve math problems for us, meanwhile we contemplate what music is and why we enjoy it, something that AI cannot do. Or getting better at relationships, making deals, instructing the youth etc. But there are areas where both us and the AI can be good at, at the same time, like pattern recognition, without us becoming worse because of the AI.
h4 is an underrated move because you're attacking the kingside and taking up space where there's plenty. If that gets up to h5-h7 in a way where it's protected and can't be attacked, it's close to making a new queen and can create many different issues as an attacking piece.
I woke up on Jan 1 2023 with an urge to learn chess. I can't explain it but yesterday morning I downloaded 2 apps and began watching house of UA-cam videos of chess for beginners. I don't know where this came from but I am having a blast and am now deep diving chess. It's a hobby to round out my list. So, Archery, Cigars, chess. I guess I'm just an old need but I always "wished I knew how to play chess" well, now I will.
I had a seizure on Jan 23 this year (nothing serious) and the next morning, my mind felt so empty and at peace, but for some reason bettering my life was the first conscious thought and Chess was the first step in the right direction that my mind craved for some reason. I started playing, studying, I feel like I’m totally nerding out on it. Now I’m wondering if I downloaded the same 2 apps you did 😂
8:24 Anatoly Karpov, a Russian grandmaster is known for making these moves. It is called, "prophylaxis" (a preventative move that counters your opponent's future moves). Magnus Carlsen started his openings this way sometimes.
how to be better chess player 1. always look for fork opportunity 2. recognize advantage patterns 3. put pieces in optimal square 4. castle-king shall be first priority 5. think to develop pieces (not pawn) 6. bring castle to center 7. find opponent pieces not in optimal square 8. use computer for ultimate move 9. just move pawn forward
Maybe pawn to H4 along the edge is the computer giving the other side another turn to play. It's calculated that any other move by the opponent results in a better position
If it was simply about giving the turn to opponent, surely you would want to push only one square forward. The fact that it's a two square move, it must be about tempo. I would guess it's about tempo getting the pawn closer to promotion for the end game and it's still possible to defend the pawn if needed so it's not a gambit.
For how much of an uber dork Hikaru is in his streams, I’m always surprised at how down to earth and cool he is in situations like this interview, or like when he did the OTB chess hustler videos.
9:13 it kinda of makes sense tho, because this an open Sicilian, and in these kind of Sicilians sometimes white likes to castle queen side and pawn storm the king side to play for mate, so this is just doing things in a weird computer way move order.
pushing the pawn two squares on the side of the board makes sense because it gives you time to understand what your opponent is aiming for without showing your colors. It gives you also extra tempo in the late game
but its a highly committal move, pawns cant go backwards... so u better know dam well taht pushing it that far is positional not damaging in the future
I think it does make sense, black bishop on f8 is blocked by e7 and d6, if you play g6 and then Bg7, white will have a faster contact with h5 treatening to open h file, blacks queen side is not safe for blacks king. And also, if blacks bishop wants to go to e7 you weaken d6 pawn. That´s sense I get from this h4 move, it also allows Rh3-g3 ideas, to throw everything on the king side and to castle queenside.
9:13 doesn't the move make sense because it prepares an h-file pawn push in case black castles on the king side? Given it is the logical side for black to castle? You earn a move ahead of time, black needs to counter with H5 to block
The crossover between poker and chess is very cool. I became a much better chess player when I thought much more deeply about what the opponent is trying to do and what he/she will most likely do next to achieve this. The exact same thing with poker.
Hikaru says this move h4 does not make sense 9:14 I dont understand Hikaru..... grabbing space and thinking long term in chess often makes sense. what kind of sense does he need ?
I love how Hikaru is bringing expanding concepts back down to earth. Lex is working to over-conceptualize and Hikaru is like, “nah, I just got the reps in, lol.”
That pawn is not on the edge of the board. It IS the edge of the board. Moving it up like that is making more space for your own pieces to move while slightly cutting off your opponent from getting into sniping positions.
Slightly different, but I think related take, some of what RL typircally does is implicitly reduce the complexity of future moves. Fundamentally, it is similar to a branch-and-bound concept in an integer optimization problem. If the AI can (perhaps just heuristically) remove a set of moves from being reasonable in the short-term, it combinatorically simplifies its computations for moves that are n-steps out. In this way, changing the edge of the board may not make sense for a human, but makes sense for Stockfish to recommend (I actually don't know much about the engine, so I don't know if this is how it operates). So, it's not just about cutting off the opponent, it's about computationally simplifying the game which benefits the player (be it human or much more likely machine) that can better reason at, say, 80 moves out.
@@himynameisdavenicetomeetyou the engines are set to perform at specific move depths, which should automatically limit computations. From what I remember most engines extensive databases of played games and have opening theory. Perhaps the Google engine does do what you suppose, as it learned mostly by play. Could possibly explain why it makes very alien moves. Ofcource when you input those moves into stockfish it's gonna agree that it's not losing (if given enough depth for analysis), but I'd be surprised if it would come up with such strategies itself.
Lex seems to have a great understanding of the “meta” of chess. Not sure how much he plays, but I bet he’d become a pretty strong player if he invested time into training
Unlikely. It's almost impossible for adults to become strong players if they weren't already strong players as children. At best he could become an average player. It's kind of like piano or guitar. It has to be ingrained while your brain is still developing.
@@MrFreeGman it is definitely possible to become great at something with enough time on your hands of you're already an adult. The misconception as to why it isn't as easy for adults comes from having to work, having to pay bills, having to do chores, raising kids etc. If you factor out all of that and only compare skill or talent or brainpower, it is most definitely possible to become as great at something as you would become if you'd have started as a child.
@@Olibaby12 Nah, it's really not. You can become good at many things with practice, but you will never be on par with a super grandmaster no matter how much time you dedicate. Again, it has to do with brain development. Children's brains are much more elastic than that of an adult.
@@MrFreeGman untrue some NBA players picked up basketball in high school or even after and still made it to the NBA they are most certainly the minority but it is possible, of course most the people in the league all started playing when they where young and it helped them since kids absorb information much better then adults but your statement of people “can’t be strong” is incorrect it’s just much harder or more unlikely to come in fruition.
Lower level of chess you can look or focus your attention to one area of the opponents board. He will only be focused on what you are looking at and make the wrong move. My rule of thumb is look at the least active part of the board.
Funnily enough, back when I was first learning chess I also didn't have/use computers (this was a few years before I got my first PC In the late 90s), but I always loved pushing my pawns on the edge. I think the "deep wisdom" behind why it's good is manifold and pretty easy to explain: 1. It gains space on that side of the board. 2. It's easily protected by the rook and hard to attack. 3. It allows you to get a head start on a potential "pawn storm" if the position later allows. 4. If it reaches your opponent's pawns they have three less-than-ideal options: a) Take it, in which case they've ruined their pawn structure on that side, created an endgame weakness with the isolated pawn, and allowed the rook to get active; or b) Push the g-pawn, which will usually create a hole on f3/f6 (always in king's pawn games) and allow that pawn to be a permanent thorn in their side, by permanently controlling the g7/g2 square; or c) Ignore it, allowing you to take and advance a pawn to your opponent's second rank, which will then force them to defend against promotion. It's basically a move with no drawbacks but with many potential strengths, especially later in the game. It's also a good move in a lot of positions, so especially in low-time formats it can be a great move to play quickly knowing it's usually going to be safe.
This is the level of chess that I love. This showcase of very strategic *ideas* that you put into reality on the board. It makes the game so much deeper than just calculation. Love the explanation!
It has the drawback of losing a tempo, which could be used to fight for the center, for example. It would be considered ridiculous before computers because, from a human perspective, there are other moves that achieve more meaningful goals and provide more tangible benefits
@@zzzinga I wouldn't consider it losing a tempo as it's hardly a useless move (as I outlined). Now, I would very much agree that there are other priorities that are generally higher, but there are plenty of openings in which the center is fought for early and then you have time to make such moves and fight on the flanks. As for the rest, well, I consider myself a human (last I checked) and the move always made sense to me. I probably didn't do it as much as computers suggest, but in certain games in which there was a lack of more tangibly beneficial move I always found it useful.
If you look at the position that Hikaru is talking about around 8:40, the issue is speficially the situation where king side knight has already moved over to other side of the board. That's what makes that move so weird to still make sense for a computer.
@@alfaseeds13 Not even close. He've definitely cheated OTB as well, I fully support Magnus on that. Just getting tired of that teenege boys' joke about "vibrating devices", hahaha, anal beads, hahaha, soo funny. For me the funniest part is the fact that there's grown up people that STILL finds the vibrating joke so funny 😁
i think for me i have set opens where im looking to attack an area of the board and when that doesnt work which i know str8 away, i set up a defence and then kind of just wait for them to make an error
As a fan of Irving Chernev's classic book: Logical Chess Move by Move, I was intrigued with Lex mentioning that none of the chess engines explain the logic behind each or its moves. So I typed this prompt into ChatGPT: "I'd like you to play a game of chess with me and explain why you made each of your moves". And voila, it played a game of chess with me and not only explained its moves, but mine as well. Cherrnev would be gobsmacked to realize that such a valuable tool existed.
Best beginner advice: Learn opening, middle, & end game theory. Sure it will take you longer to start winning but when you do win, you will win at higher levels. Beginners & intermediates memorize gambits & openings. Grandmasters focus heavy on theory. Notice how no grandmasters tell you to learn theory simply bec it IS the secret.
how do you learn opening, middle, end game theory? and what is the difference between learning/memorizing openings and learnings opening theory? thanks.
@@cartown406 I’m not sure what the other guy means xd but the idea is that we should learn the opening principles (control center, develop pieces, castle before attacking, remember to keep everything protected etc.) instead of memorizing moves without actually understanding the ideas behind them. This also helps you develop instincts on where your pieces should go to give you the most advantage. Understanding openings boosts your knowledge, while memorizing lines might end backfiring if you accidentally switch the order of moves or blindly go for a setup without looking at what your opponent is playing (for example playing the same London setup against anything). An example I can give you of understanding the openings is in Exchange Caro-Kann (1. e4 c6 2. d4 d5 3. exd5 cxd5) you can try to memorize why exactly you put the bishop on d4 as white right after this line, but the whole point is that the only good square for the opponents bishop is on f5 (same diagonal) or a potential pin on g6, so you want to occupy the diagonal first and then find a way to not let the opponent pin you with Bg6 (so you’re pretty much playing in a way that makes it hard for your opponent to do something about that bishop). Also throwing in the move pawn to c3 builds a strong pawn chain and protects the pawn on d5. Obviously this is only the beginning, the whole variation can go in different directions, but just understanding this idea helps you understand the position and you end up remembering the opening better
It's so funny to see Hikaru talk about "the computer says it's fine so we know that it is but I and other GMs can't explain it." :D It makes absolute sense that we have gotten to the point a while ago where the engines produce good moves that humans wouldn't consider but it must be so weird for someone who learned chess 20+ years ago.
Not even computers can evaluate the full game so we cannot know if the computer program actually evaluates the position correctly using heuristics (AI or otherwise). However, computers are already much stronger in Chess so we must simply trust the computer to be probably correct - or at least more correct than humans.
I think Hikaru and Yasser Seirawan are the best when it comes to teaching chess and making the game interesting to even people who aren't even fans of the game.
When i was younger without computers i would push the h pawn because my dad would alway castle and i realized the pawns on the edge of the board can be quite powerful because the king usually ends up in one of the corners and they can be very tricky
Started using computer at 38, I agree it’s so much better to wait, you devellp intuition and you might didcover domething new along the way (not that I did)
Lex you gotta have Derren Brown on the show. . . this made me think of the chess episode he played against 9 the pros all at once. (please and thank you lol)
Full podcast episode: ua-cam.com/video/oJNvxYEcVAY/v-deo.html
Lex Fridman podcast channel: ua-cam.com/users/lexfridman
Guest bio: Hikaru Nakamura is a chess super grandmaster and is currently the #1 ranked blitz chess player in the world. He is also one of the top chess streamers on Twitch and UA-cam.
0:39 why capture the rook if checkmate is right there? :O
I feel like to answer the last question of the clip. The bot did thousands of calculations every second and decided that for now this was mathematically the safest move and has the most potential for future set up. I could be talking out my ass though.
Great Hikaru is attempting to train an AI how to play chess
Hasnt AI beaten champhions decades ago? already
This is the best comment ever
@@baqirhusain5652 it’s a joke
@@baqirhusain5652 he's a 100 rated AI
Did you see chatGPT play lol
0:38 Editor taking rook instead of mate in one, classic
I want the king to suffer psychologically as I eat all of his friends
Hell slowly kill everyone and then checkmate the king
@@ruslo_b yes there is, queen with the pawn checkmates after the king moved
@@ruslo_b The king didn’t move to c7 lol
when you see mate in 1, look for better
Hey Lex, I'd love to see a sit down with Yasser Seirawan while you're on this Chess binge. He's such a captivating story teller and has a world of experience of Chess from the days of Garry Kasparov and his ilk
Would be a great guest, I’d also like to see Naroditsky but Yasser would be legendary
Seirawan is the funniest guy, has tons of stories and has played at the highest leven. Great idea!
@@bvraeqvete Lex has already interviewed Kasparov.
Would love for him to sit down with Eric Hansen
@@booni5114 although Eric seems like the coolest guy to hang out with, Idk if he'd be of great relevance.
"The ability to play chess is the sign of a gentleman. The ability to play chess well is the sign of a wasted life"
- Paul Morphy
This one I've seen before it disturbed me
Well of course Morphy would say that...
Shut up and puch the pieces around paul
He said that because he had no one on his level during this time, infact, he felt as if he wasted his life because his parents wanted him to be something else
This quote is kind of outdated. Morphy said this because at that time (the 1800s) if you were really good at a board game like Chess you were seen as a loser because you weren't good at something that was gonna further society.
Best thing I’ve learned in chess came from Mikhail Tal quotes. “You must take your opponent into a deep dark forest where 2+2=5, and the path leading out is only wide enough for one.” “There are two types of sacrifices: correct ones, and mine.”
what did you learn from them? They are more like inspirational and funny.
@@sriracha9729 1) Chess is more exciting 2) AlphaZero plays like Tal
@@sriracha9729 1) this is such a perfect example of chess, at the higher levels of chess, you are constantly trying to take your opponent out of the prep they know (bright forest) into somewhere they dont know/prep they dont know (dark forest). 2+2=5 meaning the conventual rules that everyone knows stop making sense, an example being knights at the edge of board being a better move only during circumstance, hell maybe even a knight on a corner being the best move.
"the path leading out is only for one" is self explanatory, but it just shows how some games are very aggressive, and other are very drawish, making the choice of you pushing for a win is also you being pulled for a loss.
an amazing quote. that really shows how beautiful chess is.
Proceeds to blunder the queen and a rook*
@@AltairCreedZ Ask Kasparov about his last match before Tal passed
Lex's folks utilizing the "make content, then slice the content into smaller content" thing. Good job, team. Thousands and thousands of views.
He learned from Rogan and JRE Clips
Danaher Clips Danaher Clips Danaher Clips
@@jaytorr6701 Andrew Schultz was the og
@@jaytorr6701 honestly i would have expected clips from podcasts to trend immediately, cause most people simply don't have time to watch entire podcasts
@@ImHeadshotSniper I watch most clips but never watched anything on the main channel lol
Machine learning is one thing, machine explaining is another.
I asked ChatGPT this: Why is H2-H4 a good move in a chess opening?
In a chess opening, H2-H4 is a good move because it allows the knight to move to the more active f3 square. This move also helps support the pawn on the e4 square, creating a strong center for White. Additionally, H2-H4 can be used as part of a fianchetto, where the bishop is developed on the g2 square, which can lead to a strong kingside attack. Overall, H2-H4 is a useful move that helps White develop their pieces and create a strong position on the board.
Pretty decent explanation as far as I can tell.
@@marnixvlot1868
Yeah, that sounds pretty good.
What i meant by Machine Explaining though is...as A.I becomes more intelligent and more self aware will it always choose the best option that benefits the user or will it choose an option that benefits itself?
Decisions, decisions...
@@marnixvlot1868 It's actually pretty funny how bad of an explanation that is: pawn to H4 doesn't help the knight to move at all and blocks it from moving if anything, H4 is really far from E4 and I don't see how moving an edge pawn is strengthening the center, and if you move H2 you still can't move the bishop to F2 so everything it said is basically nonsensical
It's pretty interesting to look into the ways these chat AI's fail, since they're not designed to think like us or to reason or to come up with correct answers, but rather to write like we do. It often ends up with a lot of "Escher sentences", where at first glance it sounds like it makes sense until you think about it; for example, the sentence "more people have been to Germany than I have" makes literally no sense
Funny enough, it's these kinds of results/interactions that show the need for more emphasis on machine explanation like the OP was talking about. On being able to ask Chat-GPT why it thought that moving to H4 would help support E4. Of course, the real reason is just that it's mimicking how people talk about chess, but it itself wasn't trained to play. Unfortunately, explainable AI is far less developed and researched on right now, so we can't know for sure
@@diegotejada55 It's also interesting how they both thought is was a decent explanation. It highlights two problems as far as I can tell at the moment:
1. People are largely content with content that "sounds right" about a given subject.
2. The less a person knows about a subject, the higher the chances for number 1.
AI is amplifying this age of information, and we need critical thinking now more than ever.
Here is chatGPT 4o’s explanation for why H4 is a fine move:
In the position shown in the video, Hikaru Nakamura is discussing why moving the pawn to h4 might be considered a good move despite seeming unconventional. This is an example of modern chess theory where computers have significantly influenced human understanding. Here are a few reasons why h4 could be considered a good move in this context:
1. **Opening Principles and Flexibility**: While traditionally the advice is to control the center with pawns, developing pieces, and ensuring king safety, modern engines sometimes prioritize piece activity and flexibility. The h4 move might prepare for future attacks on the kingside, potentially supporting an h5 push or opening lines for rooks.
2. **King's Safety and Aggression**: By pushing h4, White could be planning to castle queenside. This move can serve as a preparatory step for an aggressive kingside pawn storm. It signals White’s intention to launch a strong attack on the black king's position if Black castles kingside.
3. **Preventing g5**: The move h4 could be aimed at preventing Black from playing ...g5, which could disrupt White’s control of the center and potentially kick out a knight on f3. By playing h4, White is controlling the g5 square indirectly.
4. **Psychological Aspect**: Unconventional moves can put psychological pressure on opponents, especially if they are not well-versed in handling such positions. It can lead to errors if the opponent overreacts or misjudges the intention behind the move.
5. **Specific Tactical Ideas**: There might be specific tactical ideas that justify the move in this exact position, which aren't immediately obvious. Computers are excellent at finding deep tactical resources that might not be apparent to the human eye initially.
By analyzing the position more deeply with a chess engine, specific tactical nuances can be revealed that justify the move. Chess engines evaluate positions on a much deeper level, often seeing the long-term strategic potential of a move that humans might overlook at first glance.
The Alpha GO movie is great for understanding how AI thinks. It transformed the approach to GO.
When they figured out its strange approach to the game they finally realised. AI doesn't care about winning by a huge margin. It figured out that winning by 1 point still means a win.
So, once it has pieces in place to win by 1 point, it can move on the another area of the board. Human players were concentrating on capturing as many pieces as they could, wasting time in an area while GO was setting them up for a loss in another area.
Moving the Pawn up along the edge makes no sense to a human thinking just 5, or 10 moves ahead, but AI probably figured out that it will help isolate the King in 25 moves, or at least shift the odds from 50:50, to 50.1:49.9 in its favour.
Glad you brought that movie up. I had never even heard of "GO." Great movie indeed.
That cover image is on point. The added graphics are on point. All the points are on point as well.
You're on point on your points.
@@champ8605 good point
no the cover image of the board is wrong. they missed the castling for black
@@thegoonist point taken.
The board overlay graphic is incorrect, but A for effort.
This conversation makes me think about people today growing up with GPS. Listening to the maps voice telling one to turn right, drive for three miles, then turn left, and you will reach your destination. I used to study physical maps and figure out which way to go. This, to me, is the human element difference Hikaru is talking about.
In your case i think it makes you smarter because going from looking at the map to determining location and finding a path and remembering the path requires logical thinking almost as if you are a computer computing something. With automated voice instructions theres no logic
Chess is different because regardless of the approach whether that be ideas and plans vs calculation both require logical thinking
AI is making us dumber and eventually absolete
@@libertas12 I can flip the script and say that it gives us more room to think, at least in different areas.
@@Steven_Kiagiadakis Which areas?
@@blinkers88 Different areas from what the AI solves, for instance it can solve math problems for us, meanwhile we contemplate what music is and why we enjoy it, something that AI cannot do.
Or getting better at relationships, making deals, instructing the youth etc.
But there are areas where both us and the AI can be good at, at the same time, like pattern recognition, without us becoming worse because of the AI.
3:58 Lex, your editor forgot to castle for black and the move h3 for white on the 2D board.
super annoying
I try to never miss the opportunity to let a master teach the fundamentals of/in any skill or field, generally seems like a good idea.
The board showing during the "class" is wrong, some pieces are in wrong positions
I noticed. The colors are reversed
Chess is a puzzle that has a solution that changes with every move the other player makes
it's not a puzzle. Puzzles are meant to be solved. There's nothing to solve in chess. Sports are not puzzles. Sports are meant to be won. Not solved.
@@frankzane5913 chess is an ever changing 2 player puzzle
Yet sometimes the result is a draw,
the man has a point
h4 is an underrated move because you're attacking the kingside and taking up space where there's plenty. If that gets up to h5-h7 in a way where it's protected and can't be attacked, it's close to making a new queen and can create many different issues as an attacking piece.
I woke up on Jan 1 2023 with an urge to learn chess. I can't explain it but yesterday morning I downloaded 2 apps and began watching house of UA-cam videos of chess for beginners. I don't know where this came from but I am having a blast and am now deep diving chess. It's a hobby to round out my list. So, Archery, Cigars, chess. I guess I'm just an old need but I always "wished I knew how to play chess" well, now I will.
I feel exactly the same way, Nakamura is the first player I gravitated to as well
Hey me too. I just started this year. If you want to play me let me know. I’m level 700
I had a seizure on Jan 23 this year (nothing serious) and the next morning, my mind felt so empty and at peace, but for some reason bettering my life was the first conscious thought and Chess was the first step in the right direction that my mind craved for some reason. I started playing, studying, I feel like I’m totally nerding out on it. Now I’m wondering if I downloaded the same 2 apps you did 😂
Thats great 👏
Oh, the same as me. I started around the same time. Now at 700
8:24 Anatoly Karpov, a Russian grandmaster is known for making these moves. It is called, "prophylaxis" (a preventative move that counters your opponent's future moves). Magnus Carlsen started his openings this way sometimes.
LMFAOOOOOOOOOOOOOOO 1:35 hikaru trying so hard not to laugh
how to be better chess player
1. always look for fork opportunity
2. recognize advantage patterns
3. put pieces in optimal square
4. castle-king shall be first priority
5. think to develop pieces (not pawn)
6. bring castle to center
7. find opponent pieces not in optimal square
8. use computer for ultimate move
9. just move pawn forward
Maybe pawn to H4 along the edge is the computer giving the other side another turn to play. It's calculated that any other move by the opponent results in a better position
If it was simply about giving the turn to opponent, surely you would want to push only one square forward. The fact that it's a two square move, it must be about tempo.
I would guess it's about tempo getting the pawn closer to promotion for the end game and it's still possible to defend the pawn if needed so it's not a gambit.
Nakamura is so pedagogical. So understanding and does the ''tests'' just like a teacher would.
He’s arm wrestling with Devin n Brzenk and learning chess with Hikaru.. well rounded
Main character stuff
@@roachies4242 indeed 🦸🏻♂️
Did Lex just ask the computer to "show your working" ?
Maths teachers everywhere would be proud. LUL
Lex: Ok, but how do you actually play? What constitutes a move?
@Super Mario I know how to play, it was a joke in the voice of Lex.
@Super Mario wooooooosshhhhhh
@Super Mario you have to be trolling 😭😭
For how much of an uber dork Hikaru is in his streams, I’m always surprised at how down to earth and cool he is in situations like this interview, or like when he did the OTB chess hustler videos.
‘Uber dork’ you are definitely American 😂
@@BL4CK0UT17 you can use it 🙏🏼
@@StephenRahrig no I’d rather say he’s just a smart cunt 🤷🏼♂️😂
@@BL4CK0UT17 well being smart doesn’t make you a dork. He’s awkward asf with a nasally voice
@@StephenRahrig I’m British, we don’t use the word dork. I like Hikaru though, he’s a cool guy.
I don't think those 2 chess boards match. I think you missed that black castled when you brought up the other board
Actually in the discovered attack example that Lex added to win a rook, there's checkmate in 1 xD
Yeah, if the king does that, but it could've taken the pawn.
If you see M1, look for better, right?
The chessboard overlay has the pieces in the the wrong position.
So he castled the black side but the graphic doesn't show that.
9:13 it kinda of makes sense tho, because this an open Sicilian, and in these kind of Sicilians sometimes white likes to castle queen side and pawn storm the king side to play for mate, so this is just doing things in a weird computer way move order.
Ii appreciate the fact that he actually put the chess board on the screen like the other chess videos on youtube
Lex Fridman, the guy who can survive on only vitamins 4 hours at a time!!!!
At 4:05 the overlay shows the black king with out castling which is wrong. Lex do you wanna check this?
Lex one of the few who could actually interview someone like this and make it interesting
pushing the pawn two squares on the side of the board makes sense because it gives you time to understand what your opponent is aiming for without showing your colors. It gives you also extra tempo in the late game
but its a highly committal move, pawns cant go backwards... so u better know dam well taht pushing it that far is positional not damaging in the future
8:00 what he’s doing is amazing. He’s reversing the game move my move.…
Fascinating interview. Thanks!
4:52 he has castled already.
No one noticed that on the graphic it doesn't show that Lex castled?
I think it does make sense, black bishop on f8 is blocked by e7 and d6, if you play g6 and then Bg7, white will have a faster contact with h5 treatening to open h file, blacks queen side is not safe for blacks king. And also, if blacks bishop wants to go to e7 you weaken d6 pawn. That´s sense I get from this h4 move, it also allows Rh3-g3 ideas, to throw everything on the king side and to castle queenside.
9:13 doesn't the move make sense because it prepares an h-file pawn push in case black castles on the king side? Given it is the logical side for black to castle? You earn a move ahead of time, black needs to counter with H5 to block
The crossover between poker and chess is very cool. I became a much better chess player when I thought much more deeply about what the opponent is trying to do and what he/she will most likely do next to achieve this. The exact same thing with poker.
He. Let's be real.
He/she? So how many times have you played with a different 'she' is real life
@@memo.a937 Twice.
@@cornelius8617?? There are many players of both genders
Hikaru says this move h4 does not make sense 9:14
I dont understand Hikaru..... grabbing space and thinking long term in chess often makes sense. what kind of sense does he need ?
Because short and long term it doesn’t appear to develop pieces in any significant way compared to other potential moves is why I presume.
5:18, it's driving me bonkers that the virtual board doesn't match the game. Black castled like 3 or 4 moves ago.
Still keeps me up at night. Or should I say at knight 😂
I love how Hikaru is bringing expanding concepts back down to earth. Lex is working to over-conceptualize and Hikaru is like, “nah, I just got the reps in, lol.”
Nice crossover.. Hikaru is an elite GM and teaches with patiente.. really interested in bringing knowledge
That pawn is not on the edge of the board. It IS the edge of the board. Moving it up like that is making more space for your own pieces to move while slightly cutting off your opponent from getting into sniping positions.
Slightly different, but I think related take, some of what RL typircally does is implicitly reduce the complexity of future moves. Fundamentally, it is similar to a branch-and-bound concept in an integer optimization problem. If the AI can (perhaps just heuristically) remove a set of moves from being reasonable in the short-term, it combinatorically simplifies its computations for moves that are n-steps out. In this way, changing the edge of the board may not make sense for a human, but makes sense for Stockfish to recommend (I actually don't know much about the engine, so I don't know if this is how it operates). So, it's not just about cutting off the opponent, it's about computationally simplifying the game which benefits the player (be it human or much more likely machine) that can better reason at, say, 80 moves out.
@@himynameisdavenicetomeetyou the engines are set to perform at specific move depths, which should automatically limit computations. From what I remember most engines extensive databases of played games and have opening theory. Perhaps the Google engine does do what you suppose, as it learned mostly by play. Could possibly explain why it makes very alien moves.
Ofcource when you input those moves into stockfish it's gonna agree that it's not losing (if given enough depth for analysis), but I'd be surprised if it would come up with such strategies itself.
Lex seems to have a great understanding of the “meta” of chess. Not sure how much he plays, but I bet he’d become a pretty strong player if he invested time into training
Unlikely. It's almost impossible for adults to become strong players if they weren't already strong players as children. At best he could become an average player. It's kind of like piano or guitar. It has to be ingrained while your brain is still developing.
@@MrFreeGman it is definitely possible to become great at something with enough time on your hands of you're already an adult.
The misconception as to why it isn't as easy for adults comes from having to work, having to pay bills, having to do chores, raising kids etc.
If you factor out all of that and only compare skill or talent or brainpower, it is most definitely possible to become as great at something as you would become if you'd have started as a child.
@@Olibaby12 Nah, it's really not. You can become good at many things with practice, but you will never be on par with a super grandmaster no matter how much time you dedicate. Again, it has to do with brain development. Children's brains are much more elastic than that of an adult.
@@MrFreeGman untrue some NBA players picked up basketball in high school or even after and still made it to the NBA they are most certainly the minority but it is possible, of course most the people in the league all started playing when they where young and it helped them since kids absorb information much better then adults but your statement of people “can’t be strong” is incorrect it’s just much harder or more unlikely to come in fruition.
@@Nova-6-Ping Can you name me a single chess grandmaster who picked the game up after the age of 25?
A live example of machine learning
I bet that taught Lex a lesson in chess.
developing an opening is the doorway to greatness.
Projected board is incorrect @ 5:20 +/-. King should be castled already, see 3:50 +/-. Lex + Hikaru = Awesome.
What chess board are they using?
Wood
Im pretty sure h4 is a good move because it kind of starts the process of squeezing the opponents position slowly
I thought he was teaching chess how to lex Freedman.
Deep "wisdom", explainability, statistical predictions .... Coming from computer science background can help but focus on these aspects and smile.
Who set up the board didn’t transfer correctly missing h3 for white or 0-0 for black etc
Just saying that in the example at minute 0:40 you can deliver a checkmate in 1 and taking the rook would be a mistake
The chess board over lay was not the position being played on the board.
Correct. Very frustrating.
Name of chess set?
Lower level of chess you can look or focus your attention to one area of the opponents board. He will only be focused on what you are looking at and make the wrong move. My rule of thumb is look at the least active part of the board.
I started modern day chess.
Funnily enough, back when I was first learning chess I also didn't have/use computers (this was a few years before I got my first PC In the late 90s), but I always loved pushing my pawns on the edge. I think the "deep wisdom" behind why it's good is manifold and pretty easy to explain:
1. It gains space on that side of the board.
2. It's easily protected by the rook and hard to attack.
3. It allows you to get a head start on a potential "pawn storm" if the position later allows.
4. If it reaches your opponent's pawns they have three less-than-ideal options:
a) Take it, in which case they've ruined their pawn structure on that side, created an endgame weakness with the isolated pawn, and allowed the rook to get active; or
b) Push the g-pawn, which will usually create a hole on f3/f6 (always in king's pawn games) and allow that pawn to be a permanent thorn in their side, by permanently controlling the g7/g2 square; or
c) Ignore it, allowing you to take and advance a pawn to your opponent's second rank, which will then force them to defend against promotion.
It's basically a move with no drawbacks but with many potential strengths, especially later in the game. It's also a good move in a lot of positions, so especially in low-time formats it can be a great move to play quickly knowing it's usually going to be safe.
This is the level of chess that I love. This showcase of very strategic *ideas* that you put into reality on the board. It makes the game so much deeper than just calculation. Love the explanation!
It has the drawback of losing a tempo, which could be used to fight for the center, for example. It would be considered ridiculous before computers because, from a human perspective, there are other moves that achieve more meaningful goals and provide more tangible benefits
@@zzzinga I wouldn't consider it losing a tempo as it's hardly a useless move (as I outlined). Now, I would very much agree that there are other priorities that are generally higher, but there are plenty of openings in which the center is fought for early and then you have time to make such moves and fight on the flanks. As for the rest, well, I consider myself a human (last I checked) and the move always made sense to me. I probably didn't do it as much as computers suggest, but in certain games in which there was a lack of more tangibly beneficial move I always found it useful.
I think pawn to H4 is good because it supports your king-side knight to advance and opens up the kingside rook
If you look at the position that Hikaru is talking about around 8:40, the issue is speficially the situation where king side knight has already moved over to other side of the board. That's what makes that move so weird to still make sense for a computer.
all my chess pieces have an existential crisis
Vibrating device inside of me 😂😂😂😂 thx Hans
so, is it really that funny? I was facepalming, like really Lex, is that your level!?
@@edddo4314 hans fanboy?
@@alfaseeds13 Not even close. He've definitely cheated OTB as well, I fully support Magnus on that. Just getting tired of that teenege boys' joke about "vibrating devices", hahaha, anal beads, hahaha, soo funny. For me the funniest part is the fact that there's grown up people that STILL finds the vibrating joke so funny 😁
@Super Mario absolute smegma male
i think for me i have set opens where im looking to attack an area of the board and when that doesnt work which i know str8 away, i set up a defence and then kind of just wait for them to make an error
Chess is in the come up. Happy to be in the wave since i was young!
4.08- onwards Lex’s “castle” isn’t displayed on the illustration graphic!!!!
@0:39 blunder... checkmate was right there
my first time hearing hikaru talk, thought he would have japanese accent
.At 3:57 Lex's castle does not get shown on the screen. its distracting me and i cant move on 😀
What are these pieces called that are on the board? Are those fischer ultimates?
The virtual display wasn't synced with what was on the board. at 5min black already castled but the game screen still showed pre castling. weird.
As a fan of Irving Chernev's classic book: Logical Chess Move by Move, I was intrigued with Lex mentioning that none of the chess engines explain the logic behind each or its moves. So I typed this prompt into ChatGPT: "I'd like you to play a game of chess with me and explain why you made each of your moves". And voila, it played a game of chess with me and not only explained its moves, but mine as well. Cherrnev would be gobsmacked to realize that such a valuable tool existed.
Best beginner advice: Learn opening, middle, & end game theory. Sure it will take you longer to start winning but when you do win, you will win at higher levels. Beginners & intermediates memorize gambits & openings. Grandmasters focus heavy on theory. Notice how no grandmasters tell you to learn theory simply bec it IS the secret.
how do you learn opening, middle, end game theory? and what is the difference between learning/memorizing openings and learnings opening theory? thanks.
@@cartown406 I’m not sure what the other guy means xd but the idea is that we should learn the opening principles (control center, develop pieces, castle before attacking, remember to keep everything protected etc.) instead of memorizing moves without actually understanding the ideas behind them. This also helps you develop instincts on where your pieces should go to give you the most advantage.
Understanding openings boosts your knowledge, while memorizing lines might end backfiring if you accidentally switch the order of moves or blindly go for a setup without looking at what your opponent is playing (for example playing the same London setup against anything).
An example I can give you of understanding the openings is in Exchange Caro-Kann (1. e4 c6 2. d4 d5 3. exd5 cxd5) you can try to memorize why exactly you put the bishop on d4 as white right after this line, but the whole point is that the only good square for the opponents bishop is on f5 (same diagonal) or a potential pin on g6, so you want to occupy the diagonal first and then find a way to not let the opponent pin you with Bg6 (so you’re pretty much playing in a way that makes it hard for your opponent to do something about that bishop). Also throwing in the move pawn to c3 builds a strong pawn chain and protects the pawn on d5. Obviously this is only the beginning, the whole variation can go in different directions, but just understanding this idea helps you understand the position and you end up remembering the opening better
This video is for people who understand chess, non-chess people aren’t even watching this, incase you were wondering.
"Bobby Fischer Teaches Chess" is an excellent way to learn optimal places for pieces.
This guy is asking such great questions, jesus
0:39 why win a rock when you can checkmate with Qd8++ ?
2:38
It's so funny to see Hikaru talk about "the computer says it's fine so we know that it is but I and other GMs can't explain it." :D
It makes absolute sense that we have gotten to the point a while ago where the engines produce good moves that humans wouldn't consider but it must be so weird for someone who learned chess 20+ years ago.
Not even computers can evaluate the full game so we cannot know if the computer program actually evaluates the position correctly using heuristics (AI or otherwise). However, computers are already much stronger in Chess so we must simply trust the computer to be probably correct - or at least more correct than humans.
Congrats to the new Fischer Random World Champ!!!
Has lex, or any guest, recommended any books on chess theory/techniques?
at 00:36 why cant king move to c7
I think Hikaru and Yasser Seirawan are the best when it comes to teaching chess and making the game interesting to even people who aren't even fans of the game.
When i was younger without computers i would push the h pawn because my dad would alway castle and i realized the pawns on the edge of the board can be quite powerful because the king usually ends up in one of the corners and they can be very tricky
Hikaru just made it to the scc final
3:25 That joke bzzzzzzz's for itself. 😂
Did anyone notice that 00:37 is actually a checkmate in 1?
Black King Castle didn't show on image
Such a great conversation!!
I read this as "Hikaru Nakamura teaches chess how to Lex Fridman"
Started using computer at 38, I agree it’s so much better to wait, you devellp intuition and you might didcover domething new along the way (not that I did)
Listening to chess masters is so unbelievably cool.
if the programmers of chess algorythm in the computer were chess players everything would be different today.
I wonder if it's possible to attach an LLM to the way chess engines (newer ones anyway) evaluate, then it could tell you.
Lex you gotta have Derren Brown on the show. . . this made me think of the chess episode he played against 9 the pros all at once.
(please and thank you lol)