I appreciate it too, I can't memorize just bland word lists but if I have definitions I can more easily convince myself that a word I want to play is in fact a word if I remember the definition
My pleasure! As players, it's easy to fall into the "the words are just playing pieces" trap, but for casual Scrabble players or general gaming fans, the idea that Scrabble plays are totally divorced from meanings is bizarre. The first question a normal person will ask is, "what does that mean?"
Yeah, I particularly appreciated this for "ENURN" which on the face of it looks like nonsense but, as soon as I saw the definition, I connected it to words like "ENTOMB" and it made so much more sense. Immediately providing that info in the video just makes for such a better piece of vocabulary education!
DUO also gives Nigel a (small) chance to draw EJIDO from K8 I guess. I remember watching this game live, and after DUO Jesse started audibly laughing in disbelief 😅
Maybe after Jesse didn't block the first two he either thought Jesse wouldn't realize the third, or more likely, he was hoping Jesse would think it was a two tile bingo fish merely bluffing the J.
@@jesseday7925have you thought of becoming a content creator as well? If we can't ever get Nigel's insight, at least we can get the best human players to share their thoughts for the world
I remember that “scrabble players retell their favourite moments video”, Jesse Day said his goal was to be known as the second best scrabble player alive, after Nigel. Seems like he might be on his way to do that!
He is in a very small group of players who can make a strong claim to that distinction. Right now, I'd award that title to David Eldar, the only other player in the world with multiple World Championships.
You make an excellent point about how computer solvers fail to 'know' what the opponent is setting up after plays like DUO. That one was even more obvious than the other two given the relatively few tiles left in the bag AND the fact that there is no reason Nigel would make that play without the J!
@@wanderer15 i think nigel mustve done it for the memes. win or lose, he did make history with three set ups in a single a game, against a very powerful opponent no less. dude is just flexing
It's actually surprising to me that Scrabble solvers are as basic as they are. I suppose there are way too many legal moves and a bit too much randomness in Scrabble to achieve Stockfish levels of gamesense and foresight, but I would expect the program to go at least a few moves deeper into the game tree than it does.
@@tissuepaper9962it simulates the rest of the game but using only random tiles (from the bag) and equity. I think a DL approach would be much stronger
I know, I feel the same way. Contrary to my assertion in this video re: DUO, Nigel is actually excellent at determining exactly when to make his patented setups - that includes holding back when the risk doesn't justify the reward. It would be tricky to unearth them, but showing some positions where an appealing setup was available and Nigel *didn't* play it would put some of these setups into better context.
The most iconic Nigel-Jesse moment is the PIMA play in the world final. Most players seem to agree Jesse made the right move, but I'm with the minority who believe the Q risk was greater than the bingo risk. That one play deserves a video all by itself!
I hear the WU combination described as awful often. It being awful makes some amount of sense but I feel like a video on the topic of bad letter combinations and why they are like that could certainly help. Wink wink.
I made a video for the Scrabble Go channel a couple years ago about this which explains the main principle behind it very quickly: ruclips.net/video/M4v7J9ul768/видео.html
I've recently found your videos Will, and I have to say, I'm hooked for good. Although I don't play Scrabble in English (I'm from Poland) I enjoy your channel thoroughly
I wonder how hard and slow it would be to make an "ouvert" sim feature where the opponent's first response is done with exact knowledge of your leave or rack (with the plies after that being still done in Hastybot style)
your videos have been helping me as a casual too, last week i played a 100 point bingo against my parents and won easily, and this week i would've beat my grandpa if my dad didnt block my bingo by accident
The thing computer analysis can't do that humans can is smell. Jesse possibly found Duo a very fishy play, and thus had more incentive to play there than what a computer analysis would suggest.
A stronger engine would definitely be able to sniff this out. This is like the difference between running sims on your solver in poker and using an actual deep learning engine like Pluribus
Computers always seem to have problems with giving up points in order to win the game. I think that this is a reoccurring theme in computer plays. This is also the case with the earlier computers in chess that have trouble with positional sacrifices (sacrifices that elevate a position into a winning game). I don't know how it works in scrabble, but computers in chess seem to not understand closed boards/positions as well as humans sometimes. Great vid as always!
I disagree. The problem here seems to be hidden information much like in poker. Maybe the next time, Nigel will be able to play duo without a strong follow-up as as bluff. So each move in scrabble are not good or bad on their own, they are the product of a range of possible plays from possible hands which does not happen in chess.
@@barakeel yeah, you are right. The problem with Scrabble (which also makes it interesting) is that you are partly dependent on the tiles you receive. This prob makes it so that the computer can’t make an accurate guess of the win percentage of the current board.
The idea of bluff setups is fascinating - we're definitely still in that territory of when you see an obvious setup, your opponent basically "always has it." Balancing your range to induce opponents to sacrifice equity is tough, because it's rare that a setup will be useful for one tile and one tile only. Usually, scoring spots on the board are considered good because they accommodate a variety of tiles, so if you're going to make a setup, either it needs to be extremely specific to one or two tiles (making it a better candidate for a bluff), or the upside for you has to be worth the downside risk.
The #CHL[O]R[O]DyN[E] play (00.06) is legendary, but I’m only just now learning that one of those players got BASENJI down as a natural (no blanks) bingo in the same game! Maybe not chlorodyne level, but still amazing. If it was Nigel, and if he played TIE to set up his S, then even that game is an example of his set up prowess.
What’s the difference between the left side and right side placements for yealdon at 1:30 if the board was still symmetrical? Can’t Nigel play the same heavy letter words even when Jesse played on the left side of howl?
LAH, LAV, LAW, LAX, and LAY all hit that spot super hard with tiles scoring 4 or more points. On the other hand, the worst that can happen with AL is MAL and PAL, which only use 3-point tiles and will score much less. In general, 14 tiles go after LA, but only 8 tiles go before AL.
@@Cloiss_ Oh whoops, I thought that scrabble words could be spelled forwards or backwards. Since players sit opposite each other, there's no one "correct" direction. Is it only in tournament Scrabble that you always have to spell words left to right / top to bottom?
On most boards, the bonus squares have text which describes the nature of the bonus - tiles need to be oriented that same direction, so players usually spin the board around so that they can see it right-side-up and play their words in that same orientation.
It’s surprising that engines don’t take into account expected value of open spots for an opponent. It seems pretty clear that playing on a more open part of the board is better than on a more closed part, and setup plays generally open the board. It seems like the definition of equity needs to be improved.
Engines lump all of the open spots on the board together into one number - average opponent score. Because the board is quite open, DUO doesn't even increase average opponent score by that much, (for example, with Jesse's letters, he had similarly high-scoring plays through the V of VEGIE), but it massively increases Nigel's average score on his next turn, so simulation loves DUO for that reason. The trouble is that engines don't know when human beings will deviate strongly from the usual heuristics that govern most decisions (scoring a lot of points and keeping good letters), which is a trickier problem. Solving it would involve some kind of lookback to the previous move and weighting what tiles would make sense for the opponent to hold when making that play. This process would quickly turn up the J as an extremely likely tile to hold when playing DUO, for example.
In my casual matches, I very rarely open up the triple letter and triple word spaces at all, usually waiting for my opponent to branch out towards them so I can take the bonus points. However, I occasionally risk a setup play like this if it might make a bingo or 60 point (usually ZA or QI) play possible. The trouble is that you can't really hide your intentions when you play a weak setup word next to such a powerful spot. If I'd somehow gotten that far, I don't think I would have risked playing DUO, but my more cautious playstyle isn't likely to win any championships anytime soon. Sometimes the risks pay off - this just wasn't one of those times. Still, after two successful high scoring setups, I think going for a third might have been pushing his luck. 😅 Great video!
Nigel has peerless instincts for making setups when they dramatically favor him compared to his opponent and for when the game situations warrants it - DUO is a rare potential counter-example (even then, a debatable one).
Hi Will, I am just getting back into the game after many years away, largely in part due to inspiration from your channel! I own a Diamond Edition Scrabble set, unfortunately it is missing the "K" tile, which Hasbro is not able to replace (they can only do the regular Scrabble game tiles). Do you have any recommendations for pro-grade Scrabble tile sets, maybe injection molded plastic ones like I've seen used elsewhere? I looked on Amazon and saw some pretty black and silver wooden replacement sets for $15, but the comments indicated that the tiles had uneven width (not great for the Diamond Edition board), and if I'm going to invest in a nice new set, I'd like them to be really good and standardized.
This is a good question that I wish we had a simpler answer to. I know that some Scrabble orgs are looking into producing tournament-ready boards, but I can’t turn up any of them on Google.
One step ahead of the game isn't the plan kid. Two to three steps ahead. Beating an enemies move before it's even made. That's a plan. - Hannibal Smith, the A-Team
Will - I love these videos. Completely addicted! You have a great record against Nigel yourself. Can you make a video about what it's like to play him? I imagine it's a bit like coming up against the great Shane Warne in cricket - some players will be halfway to losing already from the pressure of his reputation.
Thank you for the kind words and for watching! I definitely plan to do a video about playing Nigel at some point. My record is pretty easy to explain - I'm juuuust good enough that if I draw insanely well I can win more games than him in a small sample.
Players typically have 25 minutes to make all of their moves in a game. Some moves might take you 10 seconds, others might take you 10 minutes - as long as you spend no more than 25 minutes combined, you're good. If you go over your allotted time, you get docked a 10 point penalty for each minute you go overtime. After 10 minutes overtime, you instantly forfeit the game (this is exceedingly rare).
Nigel’s earlier setup plays aren’t super obvious as setups. They score decently and Jesse could reasonably assume Nigel is just dumping some tiles while saving strong bingo tiles so he doesn’t look closer and see the setup. Duo on the other hand is so obviously a setup move a blind person could see it and Jesse knows he should block Nigel’s scoring opportunities off of duo.
I wonder if it would be too hard to compute something like weights for tiles based on plays made. If the average rack doesn't lead to DUO being the best play maybe it could weight J heavier in the "random" selection of tiles.
This is definitely possible in theory and I believe it would represent a huge step forward for engines against the top human players - their ability to brute force calculate something like this would outstrip human instinct, especially in non-obvious situations, unlike DUO.
There are occasions where the other player plays a phoney*, and you can play something normally considered to be crazy that favours you. Putting a vowel next to a triple letter score square when you have an X or something, and there are no heavy tiles unseen is not uncommon. This dude decided to learn the entire dictionary before playing his first tournament from what I heard. Seems to have the memory of a GM Chess player.
It's true that he usually does win in the games I showcase! But I posted a video about Joel Sherman's epic comeback against Nigel in 2018 here: ruclips.net/video/kKySlV_VpRg/видео.htmlsi=ntAQj4goEuTDMxj_
Why does the Scrabble engine algorithm assume that players will always make their highest equity play like this instead of using a standard minimax strategy?
Probably for ease of calculation/computation. Things quickly get hairy when you try to do full simulations-within-simulations, and the current approach actually yields good results most of the time.
It seems to be an archaic synonym! I see that it's more commonly defined as one twentieth of a ream of paper, but the Scrabble dictionary always defaults to the definition with more inflections to capture more overall words (in other words, even an obscure verb definition would yield QUIRED whereas the more common noun would not)
I'm shocked that scrabble engines are still so limited. With modern machine learning techniques, it should be fairly trivial to write an engine that doesn't have the issues you describe.
At the very end of the video? That was a joke that this video is "setting up" future videos about Jesse and Nigel's epic showdowns. If you mean the Scrabble game, DUO places the O in just the right spot for Nigel to play JOINS and JO for 81 points.
Computer evaluation ignores the information that the opponent might have on your rack. The suggested replies for Jesse after DUO assume that it is very unlikely for Nigel to have a good scoring word that connects with DUO. Jesse on the other hand knows that the DUO setup is very dangerous, otherwise Nigel would have not played such a low scoring move. Therefore the evaluation is lacking information ad as a consequence is not to be trusted completely.
@@wanderer15 i'd imagine it'd excell at the kind of positions like the u showed in the video. i can see the comparison to certain positions in chess which chess engines couldn't solve before the alphazero era of chess engines. i know stockfish and leela chess zero are both opensource neural network based engines, so it shouldn't be super extremely difficult in setting up. seems exciting :D
They can do so quite effectively, I just glossed over that process for the purpose of explanation. Lots of players prefer simulating 4 or even 6 turns ahead, but I prefer to keep it simpler because I the flaws with the process that I described in the video compound when you go multiple turns ahead - many of the suggested plays just aren’t all that realistic.
I would consider DUO if Jesse had played something very short, signaling low-point tiles, but he just bingoed, so his rack is completely random...hard to criticize the GOAT, but I agree
Is there not an AI engine that learned scrabble by self-play? The only scrabble engines are algorithmic? Surely a self play AI would eventually "learn" to guess the opponent rack composition
hidden information is a massive hurdle for the speed at which engines will learn, there's a reason all the games AlphaZero has learned have no hidden information. progress on poker AI was much slower than chess, and scrabble suffers from both being more niche as a serious competitive pursuit and more importantly having an almost unfathomably larger search space than poker.
It’s simply because a very low percentage of 7 and 8 letter words contain a J compared to other letters. (That’s a big reason why it’s worth 8 points. The tiles that are harder to use in long words get a boost in score to balance them against more commonly seen letters like S, E, R, and so on)
This is one of the few areas that engines can improve. In fact, the engine I used here (Quackle) allows you to simulate positions using inferences. You can specify that your opponent will always have a certain tile and the random combinations it picks will always include that tile. However, it can't perform this process on its own - that would be the next step forward.
To limit the influence of luck Scrabble tournaments generally include way more games than, say, chess, so losing one isn't necessarily a death sentance
Is "grandmaster" in Scrabble actually defined as rigorously as it is in chess? If not (and I don't think it is) it shouldn't be used to describe a player's abilities.
Unlike chess, different countries have different standards for awarding GM titles. In North America, titles are determined based on holding a sufficiently high ranking for a period of time. (www.scrabbleplayers.org/ratings/titlists.shtml) However, you have a good point. I use the term much more loosely than I probably should. Still, within the world of Scrabble, everybody I've referred to as a grandmaster is either a titled player within their home country or somebody who is obviously at that level (or higher).
Thankfully, there's enough luck that if you get really, really good (though still not as good as Nigel), you'll win sometimes! At some point I'll do some videos about my own experience, but I've actually won 6 out of my 8 games with Nigel in my career, primarily because I've gotten by far better letters.
@@wanderer15 Definitely helps with my confidence then. I really should have gotten involved with Scrabble championships, I would have been good enough to compete back in the day, had I known about them.
I absolutely love how you included the definitions for all these obscure words they're playing
I appreciate it too, I can't memorize just bland word lists but if I have definitions I can more easily convince myself that a word I want to play is in fact a word if I remember the definition
My pleasure! As players, it's easy to fall into the "the words are just playing pieces" trap, but for casual Scrabble players or general gaming fans, the idea that Scrabble plays are totally divorced from meanings is bizarre. The first question a normal person will ask is, "what does that mean?"
Yeah, I particularly appreciated this for "ENURN" which on the face of it looks like nonsense but, as soon as I saw the definition, I connected it to words like "ENTOMB" and it made so much more sense. Immediately providing that info in the video just makes for such a better piece of vocabulary education!
The guy is such a computer he even makes computer mistakes.
DUO also gives Nigel a (small) chance to draw EJIDO from K8 I guess. I remember watching this game live, and after DUO Jesse started audibly laughing in disbelief 😅
Maybe after Jesse didn't block the first two he either thought Jesse wouldn't realize the third, or more likely, he was hoping Jesse would think it was a two tile bingo fish merely bluffing the J.
I meant to throw the EJIDO possibility in and forgot, good point!
there was no way I was going to let him cash three setups in one game!!
@@jesseday7925have you thought of becoming a content creator as well? If we can't ever get Nigel's insight, at least we can get the best human players to share their thoughts for the world
I love how Jesse played "gee" after drawing five E's. So funny!!
I remember that “scrabble players retell their favourite moments video”, Jesse Day said his goal was to be known as the second best scrabble player alive, after Nigel. Seems like he might be on his way to do that!
He is in a very small group of players who can make a strong claim to that distinction. Right now, I'd award that title to David Eldar, the only other player in the world with multiple World Championships.
You make an excellent point about how computer solvers fail to 'know' what the opponent is setting up after plays like DUO. That one was even more obvious than the other two given the relatively few tiles left in the bag AND the fact that there is no reason Nigel would make that play without the J!
Agreed - not to mention Jesse would already have felt the burn twice from the previous two setups and been on his guard even more.
@@wanderer15 i think nigel mustve done it for the memes. win or lose, he did make history with three set ups in a single a game, against a very powerful opponent no less. dude is just flexing
It's actually surprising to me that Scrabble solvers are as basic as they are. I suppose there are way too many legal moves and a bit too much randomness in Scrabble to achieve Stockfish levels of gamesense and foresight, but I would expect the program to go at least a few moves deeper into the game tree than it does.
@@tissuepaper9962it simulates the rest of the game but using only random tiles (from the bag) and equity. I think a DL approach would be much stronger
I love how at 0:34 the brilliant moves roughly correspond to the positions of the setup plays (I have no idea if that was intentional or not)
Yes, that was intentional - nice catch!
i try to set up stuff like this all the time and it never works but hey if nigel does it
I know, I feel the same way. Contrary to my assertion in this video re: DUO, Nigel is actually excellent at determining exactly when to make his patented setups - that includes holding back when the risk doesn't justify the reward. It would be tricky to unearth them, but showing some positions where an appealing setup was available and Nigel *didn't* play it would put some of these setups into better context.
The most iconic Nigel-Jesse moment is the PIMA play in the world final. Most players seem to agree Jesse made the right move, but I'm with the minority who believe the Q risk was greater than the bingo risk. That one play deserves a video all by itself!
Oh, man, I remember that one. I'll toss it on my to-do list! (Though I don't want to oversaturate with only Nigel content)
Tbh you should. For me it was one of those articles about Nigel winning in French that brought me to your channel.
@@IBAndreas nah, what subscribing to Will's channel has showed me is that there's much more to pro Scrabble than just the best player
Yeah Duo probably was a misplay. Doesn’t change the fact that he played legendarily the same whole game except maybe that one move.
Another fantastic video. I find it amazing how well high-level opponents can read each other and counter their opponents' anticipated strategy.
I hear the WU combination described as awful often. It being awful makes some amount of sense but I feel like a video on the topic of bad letter combinations and why they are like that could certainly help. Wink wink.
Very few words have WU
I made a video for the Scrabble Go channel a couple years ago about this which explains the main principle behind it very quickly:
ruclips.net/video/M4v7J9ul768/видео.html
I've recently found your videos Will, and I have to say, I'm hooked for good. Although I don't play Scrabble in English (I'm from Poland) I enjoy your channel thoroughly
I'm amazed how he can remember so many words and also read the opponents mind to one of the best players in Scrabble History.
to be one of the best*
You had me at "Nigel" :)
I'll never tire of Nigel Richards stories, the strategy of this level of gameplay is so engaging.
I am Loving the series :) hope this goes on forever
Great stuff. We just learned QUIRE in my group, glad to see it used on the big stage :)
I wonder how hard and slow it would be to make an "ouvert" sim feature where the opponent's first response is done with exact knowledge of your leave or rack (with the plies after that being still done in Hastybot style)
Have you ever played Upwords? It’s similar to scrabble but it’s on a smaller board and allows players to stack tiles on top of each other.
I have. It's one of the better Scrabble variants.
your videos have been helping me as a casual too, last week i played a 100 point bingo against my parents and won easily, and this week i would've beat my grandpa if my dad didnt block my bingo by accident
The thing computer analysis can't do that humans can is smell. Jesse possibly found Duo a very fishy play, and thus had more incentive to play there than what a computer analysis would suggest.
A stronger engine would definitely be able to sniff this out. This is like the difference between running sims on your solver in poker and using an actual deep learning engine like Pluribus
Thank you for making this video. My middle-school self used to compete at the King's cup (2000-2002) 😂
Computers always seem to have problems with giving up points in order to win the game. I think that this is a reoccurring theme in computer plays. This is also the case with the earlier computers in chess that have trouble with positional sacrifices (sacrifices that elevate a position into a winning game). I don't know how it works in scrabble, but computers in chess seem to not understand closed boards/positions as well as humans sometimes. Great vid as always!
I disagree. The problem here seems to be hidden information much like in poker. Maybe the next time, Nigel will be able to play duo without a strong follow-up as as bluff. So each move in scrabble are not good or bad on their own, they are the product of a range of possible plays from possible hands which does not happen in chess.
@@barakeel yeah, you are right. The problem with Scrabble (which also makes it interesting) is that you are partly dependent on the tiles you receive. This prob makes it so that the computer can’t make an accurate guess of the win percentage of the current board.
@@barakeel correct
The idea of bluff setups is fascinating - we're definitely still in that territory of when you see an obvious setup, your opponent basically "always has it." Balancing your range to induce opponents to sacrifice equity is tough, because it's rare that a setup will be useful for one tile and one tile only. Usually, scoring spots on the board are considered good because they accommodate a variety of tiles, so if you're going to make a setup, either it needs to be extremely specific to one or two tiles (making it a better candidate for a bluff), or the upside for you has to be worth the downside risk.
The #CHL[O]R[O]DyN[E] play (00.06) is legendary, but I’m only just now learning that one of those players got BASENJI down as a natural (no blanks) bingo in the same game! Maybe not chlorodyne level, but still amazing. If it was Nigel, and if he played TIE to set up his S, then even that game is an example of his set up prowess.
What’s the difference between the left side and right side placements for yealdon at 1:30 if the board was still symmetrical? Can’t Nigel play the same heavy letter words even when Jesse played on the left side of howl?
LAH, LAV, LAW, LAX, and LAY all hit that spot super hard with tiles scoring 4 or more points. On the other hand, the worst that can happen with AL is MAL and PAL, which only use 3-point tiles and will score much less. In general, 14 tiles go after LA, but only 8 tiles go before AL.
@@Cloiss_ Oh whoops, I thought that scrabble words could be spelled forwards or backwards. Since players sit opposite each other, there's no one "correct" direction. Is it only in tournament Scrabble that you always have to spell words left to right / top to bottom?
On most boards, the bonus squares have text which describes the nature of the bonus - tiles need to be oriented that same direction, so players usually spin the board around so that they can see it right-side-up and play their words in that same orientation.
It’s surprising that engines don’t take into account expected value of open spots for an opponent. It seems pretty clear that playing on a more open part of the board is better than on a more closed part, and setup plays generally open the board. It seems like the definition of equity needs to be improved.
Engines lump all of the open spots on the board together into one number - average opponent score. Because the board is quite open, DUO doesn't even increase average opponent score by that much, (for example, with Jesse's letters, he had similarly high-scoring plays through the V of VEGIE), but it massively increases Nigel's average score on his next turn, so simulation loves DUO for that reason. The trouble is that engines don't know when human beings will deviate strongly from the usual heuristics that govern most decisions (scoring a lot of points and keeping good letters), which is a trickier problem. Solving it would involve some kind of lookback to the previous move and weighting what tiles would make sense for the opponent to hold when making that play. This process would quickly turn up the J as an extremely likely tile to hold when playing DUO, for example.
Never EVER stop doing this 😂😂😂😂 Amazing
Was about ro make a snarky comment about Nigel losing, That was before you informed me that he won the tournament for the 15th bloody time 💀
7:14 I was looking for an open "I" so he could play Jillinois
Another great video! Makes my day whenever one pops up in my notifications 😊
Thanks so much for this enjoyable and instructive video.
"Jesse draws five E's..." Certified Random Distribution Moment™. Bad luck doesn't care how good you are.
3:00 What does the # symbol mean in the word list?
It means that a word is valid in the Collins dictionary (used outside of North America), but invalid in the North American dictionary
Yup, well said
In my casual matches, I very rarely open up the triple letter and triple word spaces at all, usually waiting for my opponent to branch out towards them so I can take the bonus points. However, I occasionally risk a setup play like this if it might make a bingo or 60 point (usually ZA or QI) play possible. The trouble is that you can't really hide your intentions when you play a weak setup word next to such a powerful spot.
If I'd somehow gotten that far, I don't think I would have risked playing DUO, but my more cautious playstyle isn't likely to win any championships anytime soon. Sometimes the risks pay off - this just wasn't one of those times.
Still, after two successful high scoring setups, I think going for a third might have been pushing his luck. 😅
Great video!
Nigel has peerless instincts for making setups when they dramatically favor him compared to his opponent and for when the game situations warrants it - DUO is a rare potential counter-example (even then, a debatable one).
Hi Will, I am just getting back into the game after many years away, largely in part due to inspiration from your channel! I own a Diamond Edition Scrabble set, unfortunately it is missing the "K" tile, which Hasbro is not able to replace (they can only do the regular Scrabble game tiles). Do you have any recommendations for pro-grade Scrabble tile sets, maybe injection molded plastic ones like I've seen used elsewhere?
I looked on Amazon and saw some pretty black and silver wooden replacement sets for $15, but the comments indicated that the tiles had uneven width (not great for the Diamond Edition board), and if I'm going to invest in a nice new set, I'd like them to be really good and standardized.
This is a good question that I wish we had a simpler answer to. I know that some Scrabble orgs are looking into producing tournament-ready boards, but I can’t turn up any of them on Google.
great vid as always
Another great video! Keep up the good work
One step ahead of the game isn't the plan kid. Two to three steps ahead. Beating an enemies move before it's even made. That's a plan. - Hannibal Smith, the A-Team
Will - I love these videos. Completely addicted!
You have a great record against Nigel yourself. Can you make a video about what it's like to play him? I imagine it's a bit like coming up against the great Shane Warne in cricket - some players will be halfway to losing already from the pressure of his reputation.
Thank you for the kind words and for watching! I definitely plan to do a video about playing Nigel at some point. My record is pretty easy to explain - I'm juuuust good enough that if I draw insanely well I can win more games than him in a small sample.
@@wanderer15 I'm sure you're being far too modest! But that puts you in rarefied company anyway. Take care mate
I'm not clear on how much time they have to make these moves? Are they even on a timer?
Players typically have 25 minutes to make all of their moves in a game. Some moves might take you 10 seconds, others might take you 10 minutes - as long as you spend no more than 25 minutes combined, you're good. If you go over your allotted time, you get docked a 10 point penalty for each minute you go overtime. After 10 minutes overtime, you instantly forfeit the game (this is exceedingly rare).
Bro Is The Hikaru Of Scrabble
Nigel’s earlier setup plays aren’t super obvious as setups. They score decently and Jesse could reasonably assume Nigel is just dumping some tiles while saving strong bingo tiles so he doesn’t look closer and see the setup. Duo on the other hand is so obviously a setup move a blind person could see it and Jesse knows he should block Nigel’s scoring opportunities off of duo.
Ahhh the goat uploaded🎉
I wonder if it would be too hard to compute something like weights for tiles based on plays made. If the average rack doesn't lead to DUO being the best play maybe it could weight J heavier in the "random" selection of tiles.
This is definitely possible in theory and I believe it would represent a huge step forward for engines against the top human players - their ability to brute force calculate something like this would outstrip human instinct, especially in non-obvious situations, unlike DUO.
Me: "CHILDREN!"
Nigel: "Chlorodyne"
Me: "...oh"
There would be no chance that half or more of these words would be accepted on any house board I’ve ever played on lol
There are occasions where the other player plays a phoney*, and you can play something normally considered to be crazy that favours you. Putting a vowel next to a triple letter score square when you have an X or something, and there are no heavy tiles unseen is not uncommon. This dude decided to learn the entire dictionary before playing his first tournament from what I heard. Seems to have the memory of a GM Chess player.
Is this the first video you've posted of a Nigel game where he didn't win?
Lots of crazy overlaps this game!
It's true that he usually does win in the games I showcase! But I posted a video about Joel Sherman's epic comeback against Nigel in 2018 here: ruclips.net/video/kKySlV_VpRg/видео.htmlsi=ntAQj4goEuTDMxj_
Why does the Scrabble engine algorithm assume that players will always make their highest equity play like this instead of using a standard minimax strategy?
Probably for ease of calculation/computation. Things quickly get hairy when you try to do full simulations-within-simulations, and the current approach actually yields good results most of the time.
Indeed Jesse Day is a strong human player.
Might’ve been able to phrase that better. Haha
I swear the bigger the beard the more insane his plays
So a trio setup containing of the words di, tomb and duo? Mmh, could there have been a superstitious sign that this wasn't the best idea?
Thought about bringing some of the players on for a cocommentary on a video?
Definitely need to try this sometime! I did it once with Josh on a game we played at a past national championship and I think it came out well.
Isn't the "quire" definition you put up for "choir"?
It seems to be an archaic synonym! I see that it's more commonly defined as one twentieth of a ream of paper, but the Scrabble dictionary always defaults to the definition with more inflections to capture more overall words (in other words, even an obscure verb definition would yield QUIRED whereas the more common noun would not)
@@wanderer15 Ah! I had no idea. Thank you!
I'm shocked that scrabble engines are still so limited. With modern machine learning techniques, it should be fairly trivial to write an engine that doesn't have the issues you describe.
How would a top human (say, Nigel) do against a computer engine playing the "equity strategy"?
Nigel would beat the computer engines we have today, but it might take a large sample of games for his edge to manifest.
Will with the huge setup at the end of the video. What play is he setting up next?
At the very end of the video? That was a joke that this video is "setting up" future videos about Jesse and Nigel's epic showdowns. If you mean the Scrabble game, DUO places the O in just the right spot for Nigel to play JOINS and JO for 81 points.
why does the thumbnail look like something you'd find in ancient Egypt 😭
Computer evaluation ignores the information that the opponent might have on your rack. The suggested replies for Jesse after DUO assume that it is very unlikely for Nigel to have a good scoring word that connects with DUO. Jesse on the other hand knows that the DUO setup is very dangerous, otherwise Nigel would have not played such a low scoring move. Therefore the evaluation is lacking information ad as a consequence is not to be trusted completely.
After making 2 really small words that turned out to be setups Jesse picked up on the pattern and stopped him this time
i'm a complete noob to scrabble, but i do play chess. are there any neural network based engines in scrabble, and are they any good?
Not yet, but I think some of our brightest minds have been considering how such an approach might work.
@@wanderer15 i'd imagine it'd excell at the kind of positions like the u showed in the video. i can see the comparison to certain positions in chess which chess engines couldn't solve before the alphazero era of chess engines. i know stockfish and leela chess zero are both opensource neural network based engines, so it shouldn't be super extremely difficult in setting up. seems exciting :D
imagine if someone intentionally made up a word like QUZ with the sole intent of trying to make it popular enough to get into the Scrabble dictionary
Good idea, quz
@@wanderer15 MIREPOIXS
Duo!
Any games where someone makes an obvious setup, but then someone blocking it somehow activates their "trap card"?
why don't scrabble engines calculate lines to a given depth instead of just next move?
They can do so quite effectively, I just glossed over that process for the purpose of explanation. Lots of players prefer simulating 4 or even 6 turns ahead, but I prefer to keep it simpler because I the flaws with the process that I described in the video compound when you go multiple turns ahead - many of the suggested plays just aren’t all that realistic.
Love me some nigel content
yep
I wonder if there's a game which is all about setups and guessing whether what your opponent just did is a points play or a setup
A variation of Scrabble, or any game? Either way, sounds tough!
Sounds strategic, and maybe a little bit like go? You could call it Stratego. 😛
This is soncooool
The game is a lot easier when you have a photographic memory and can even name the page the word is on in the dictionary lol.
Beat your scrabblegrams today
Can't stop, Won't stop, Never gonna stop stop
Yeah, even if I saw DUO, I'd play SOJU instantly
I would consider DUO if Jesse had played something very short, signaling low-point tiles, but he just bingoed, so his rack is completely random...hard to criticize the GOAT, but I agree
All my setup plays just end up getting blocked.
Imagine drawing 5 E's
Is there not an AI engine that learned scrabble by self-play? The only scrabble engines are algorithmic? Surely a self play AI would eventually "learn" to guess the opponent rack composition
hidden information is a massive hurdle for the speed at which engines will learn, there's a reason all the games AlphaZero has learned have no hidden information. progress on poker AI was much slower than chess, and scrabble suffers from both being more niche as a serious competitive pursuit and more importantly having an almost unfathomably larger search space than poker.
Why is the J such a bad bingo tile?? It’s worth 8 points.
It’s simply because a very low percentage of 7 and 8 letter words contain a J compared to other letters. (That’s a big reason why it’s worth 8 points. The tiles that are harder to use in long words get a boost in score to balance them against more commonly seen letters like S, E, R, and so on)
Has Nigel ever played an invalid word and gotten away with it?
Yes. Those instances are entirely inadvertent (and hilarious) and I should definitely make a video about them sometime!
Sounds like we need better engines
This is one of the few areas that engines can improve. In fact, the engine I used here (Quackle) allows you to simulate positions using inferences. You can specify that your opponent will always have a certain tile and the random combinations it picks will always include that tile. However, it can't perform this process on its own - that would be the next step forward.
It took long to these noobs to learn they were allowed to play horizontal words too
ZA
wait he still won in the end????
To limit the influence of luck Scrabble tournaments generally include way more games than, say, chess, so losing one isn't necessarily a death sentance
Bro, just marry Nigel already
1. I'm taken 2. I wish
Is "grandmaster" in Scrabble actually defined as rigorously as it is in chess? If not (and I don't think it is) it shouldn't be used to describe a player's abilities.
Unlike chess, different countries have different standards for awarding GM titles. In North America, titles are determined based on holding a sufficiently high ranking for a period of time. (www.scrabbleplayers.org/ratings/titlists.shtml)
However, you have a good point. I use the term much more loosely than I probably should. Still, within the world of Scrabble, everybody I've referred to as a grandmaster is either a titled player within their home country or somebody who is obviously at that level (or higher).
Thank you for that reference. I did not know there are criteria in place for awarding titles.
No point playing this game, when Nigel is just going to win everything
Thankfully, there's enough luck that if you get really, really good (though still not as good as Nigel), you'll win sometimes! At some point I'll do some videos about my own experience, but I've actually won 6 out of my 8 games with Nigel in my career, primarily because I've gotten by far better letters.
@@wanderer15 Definitely helps with my confidence then. I really should have gotten involved with Scrabble championships, I would have been good enough to compete back in the day, had I known about them.
ILYSM PLEASE MAKE MORE EDITED BOARD THINGY-MAJINGY. IM SO HAPPY WATCHING YOUR VIDEOS!❤❤❤❤❤❤
SCRABBLE!
hamburger
im EARLY EARLY! LFG!