Was thinking about covering this myself and am happy you did it. What I'd like to add is that Nigel gave out some of the prizes to the young champions. I find that so wholesome given his general avoidance of prize ceremonies 😃
Wolfram and I were undecided whether Nigel operated on intuition or brute-force computation. I think the amount of precision and certainty to get such optimal play comes only from brute force, while Wolfram thinks human brain doesn't work that way and Nigel must have intuited his way through this. In his favour, Nigel took just 4 minutes of thinking time after I played my bingo to arrive at this sequence. I cannot imagine that is enough time to brute-force compute this option, and technically he can only start considering this after he knows of my bingo that will provide the N for CI(N)Q. Unless he foresaw that already when he played his fishy (ETA)T ...
Even when one is programming a computer, if the goal is 98% accuracy within a limited amount of time, strong heuristics ("intuition") will be crucial. But in any case, "intuition or brute-force computation" is a false dichotomy; both humans and computers use a combination of both. Even endgame solvers with 100% accuracy rely on heuristics to cut down the expected size of the search space.
thanks for making the commentary on this! apologies I didn't make it explicit when I put up the annotation in cross-table that the racks are correct for both players: I record my racks, and Wolfram Poh recorded Nigel rack which I borrowed to Quackle. yes Nigel's plays were sus eg UNS and ETAT, would have been interesting to hear other expert opinions on that. if only you knew in advance those were his actual rack.
I was privileged to be standing by the board watching this endgame after I finished early. Even after OCH I wasn't sure what he was up to. ETAT is also vet interesting; I think it relates to him ensuring he had a high scoring Q play. You often see content regarding Nigel that focuses on his word knowledge in expansive games but in my opinion, and I've played him more than 50 times, it is closed boards and set ups like this where he is at his most mesmeric. His accuracy is of course legendary. In all our games he only once challenged me incorrectly (DURABLES). In one game he also admitted.postgame that he nearly played the phoney RECOT that would have given me a surprising endgame win. The key word there is "nearly"!! Nice video Matt.
I saw this on cross-tables and I’m beyond psyched that you made a video for it. All the sequences are so close, it’s like it was created in a lab for endgame instruction. And I think that it also shows how if Nigel starts some sequence like this that looks like it can’t be right, you’re better off just trusting his math and not trying to block it 😂. Same reason why, if I’m ever lucky enough to play him, I’m not gonna waste five points challenging an obscure 12 letter word.
Hey man nice vid as always. Just wondering (idk how easy it would be in your editing software) would you perhaps consider mixing the mic audio using a de-esser or some other mixer? A lot of your s and sh sounds are really sharp and offputting for me and I think it would be feasible to apply some filter to the audio to reduce those frequencies
Still fairly new to Scrabble but have seen heaps of Nigel vids and it is accepted that he is the GOAT. But...wtf is that table? Why is the difference so insane? Damn.
_adroit_ and its derivatives are typically pronounced uh-DROYT (sorta rhymes with _Detroit_ ) in US English (per Merriam Webster). The word does come from French but it's not super modern; first recorded use 1652 per same source. Cool vid and it was fascinating to watch (tho I admit I did so at 1.25× speed)
Yeah that's a pretty sick setup. My setups usually backfire and make a winning game a loss because I miss that the setup I'm making has more hooks than just the one I'm thinking of
As far as whether all of the endgames in that study were brute-forced with 100% confidence, I don’t personally know, but to answer the related question of how you know if an endgame is better than Quackle’s (and what types of endgames it often gets wrong) - the way to check is pretty simple, you can put in the starting play that you think is better and let Quackle play the other side, and if it can’t find a way to stop or alter the sequence to make it worse for you (judging by final score differential) than the sequence it suggested, then you know that it wasn’t right (though this doesn’t prove that your sequence is best). It will usually make mistakes in scenarios like this where someone gets stuck with a tile, because like Josh says in the video it’s USUALLY best to slow play and maximize the points you get per tile, but you also have to be aware of other things, like how they can make setups and how they might be able to get more points in some spot than you, so you have to take the spot rather than let them use it even if it’s not the most points you can get for your tile. So play order and even passing your turn can be important. Like one scenario I remember where Quackle played a suboptimal (by quite a bit) endgame against me was when it had a tile it was going to be stuck with (the Q iirc) and played another tile for a good score, but that was the only tile it had that could hook onto EX, and I was able to play EX->LEX->ILEX->SILEX, scoring the X four straight turns (and I think one was a double word).
You can find more information if you search for Francis Desjardins ortograf. Desjardins collaborated with Gilles Blanchette to write an optimized, custom solver which can solve endgames perfectly. But the software can take a long time to run, so they have analyzed only a select set of endgames.
This is covered, nigel will go out with qis instead, scoring 9 less points himself but denying etats (10) and getting 2 from the s. So it's 3 points worse to hold the s that long. Good practice to hold the s for a while just in case Nigel messes up, but eventually you have to play it.
do you know if scrabble engines take advantage of transpositions (a chess term that indicates positions that can be reached by multiple different move orders)? i kinda feel like "moves that score the same amount and don't interfere with my plays" would be an interesting transposition heuristic for scrabble engines, if only in endgames
Great question. Transposition heuristics don't really seem useful to me in Scrabble outside of the endgame, since the goal of Scrabble is not to checkmate, but to outscore. The concept of "interference", similarly to chess, is complex, since maybe your opponent isn't interfering with your plan, but their plan is stronger and must be stopped. The equivalent in Scrabble is creating a threat that must be addressed, regardless of score. The algorithm would have to understand the danger levels specific to the position at hand, and although there are clearly plays that "don't accomplish anything new", I don't know how useful it would be to track for those plays as an engine. As for the Scrabble endgame, transposition is extremely important, since some endgames have billions, if not trillions of possible moves in them. So an engine very much needs to parallelize as efficiently as possible to get to the correct answer in a reasonable amount of time, after an exhaustive search. Currently there is only one engine that might have been able to solve this endgame as fast as Nigel did, although it's possible for engines to be fairly confident that something is solved in a shorter period of time. That's similar to how humans, Nigel included, process these things.
I hope these competitions have huge cash prizes, because I can't think of any other justification for spending countless years learning words you will never use in real life.
Hope away... passion overpowers many of life's constants, money being one of those. I won around 16k playing scrabble competitions in 2023, and that was a lucky year. I also spent maybe around half of that just to be able to play.
I think it mostly had to do with endgame timing. If Ricky did in fact get a bingo down, he would likely not be able to threaten an outplay in the endgame and therefore not be able to win
I feel like you need to speak a little faster and brighter, and cut out the “um’s” and “uh’s”. But the content is fantastic and the editing is really good. And you explain things really well
trying to find the right balance between genuine and flowing... don't want to do too many takes, but also don't want to be too rough around the edges... thanks for the feedback
Wow the commentary in this video is.... lethargic. Depressing. Suicidal-sounding.
x1.25 speed is your friend
im gonna say it before anyone else does, pin of shame.
@@gibhib Thanks bro. Actually, you should put it at 1.5x. He sounds like a normal human only then
I couldn't agree more.
didnt have to do him like that bro
Crazy to think that Nigel saw all that before his opponent even sat down.
Was thinking about covering this myself and am happy you did it. What I'd like to add is that Nigel gave out some of the prizes to the young champions. I find that so wholesome given his general avoidance of prize ceremonies 😃
You should do so anyways. Your first video about him was so entertaining, I watched it twice
Great guy!
Btw Josh did a fantastic job covering it, but it's so interesting an end game that I'd enjoy other perspectives on it.
@@YayComity I don't think I could add to Josh's analysis 🙂
Wolfram and I were undecided whether Nigel operated on intuition or brute-force computation. I think the amount of precision and certainty to get such optimal play comes only from brute force, while Wolfram thinks human brain doesn't work that way and Nigel must have intuited his way through this.
In his favour, Nigel took just 4 minutes of thinking time after I played my bingo to arrive at this sequence. I cannot imagine that is enough time to brute-force compute this option, and technically he can only start considering this after he knows of my bingo that will provide the N for CI(N)Q. Unless he foresaw that already when he played his fishy (ETA)T ...
Even when one is programming a computer, if the goal is 98% accuracy within a limited amount of time, strong heuristics ("intuition") will be crucial. But in any case, "intuition or brute-force computation" is a false dichotomy; both humans and computers use a combination of both. Even endgame solvers with 100% accuracy rely on heuristics to cut down the expected size of the search space.
thanks for making the commentary on this! apologies I didn't make it explicit when I put up the annotation in cross-table that the racks are correct for both players: I record my racks, and Wolfram Poh recorded Nigel rack which I borrowed to Quackle.
yes Nigel's plays were sus eg UNS and ETAT, would have been interesting to hear other expert opinions on that. if only you knew in advance those were his actual rack.
If I had talked about his plays this video would have lasted way too long... lol
Why on Earth did Nigel play ETAT
"The best end-game engine we had was Nigel".
I was privileged to be standing by the board watching this endgame after I finished early. Even after OCH I wasn't sure what he was up to. ETAT is also vet interesting; I think it relates to him ensuring he had a high scoring Q play. You often see content regarding Nigel that focuses on his word knowledge in expansive games but in my opinion, and I've played him more than 50 times, it is closed boards and set ups like this where he is at his most mesmeric. His accuracy is of course legendary. In all our games he only once challenged me incorrectly (DURABLES). In one game he also admitted.postgame that he nearly played the phoney RECOT that would have given me a surprising endgame win. The key word there is "nearly"!! Nice video Matt.
I saw this on cross-tables and I’m beyond psyched that you made a video for it. All the sequences are so close, it’s like it was created in a lab for endgame instruction. And I think that it also shows how if Nigel starts some sequence like this that looks like it can’t be right, you’re better off just trusting his math and not trying to block it 😂. Same reason why, if I’m ever lucky enough to play him, I’m not gonna waste five points challenging an obscure 12 letter word.
I love these videos. Always so interesting to watch. Thank you for upload as always
I officially never get tired of Nigel videos. 🙏
That was a great video, thanks for sharing it! Hope you get to play Nigel soon!
Can we please get Nigel to play against BestBot on Woogles?
everything has a price :)
Just became your 2000th subscriber - well deserved and congratulations man
Thank you!
Great video! Thanks for making it
His beard is an extraterrestrial neural net. Make him shave and see what happens!
Great video and congrats on 2k subscribers!!!!
Beautiful ending and analysis
UH-DROYT-LEE
take it out of the dictionary, it has 2 anagrams anyway
@@axcertypo c'est votre droit
In the turn where "ab" wa played, I saw the bingo of "abrading", but it didn't work because there was no open D.
Hey man nice vid as always. Just wondering (idk how easy it would be in your editing software) would you perhaps consider mixing the mic audio using a de-esser or some other mixer? A lot of your s and sh sounds are really sharp and offputting for me and I think it would be feasible to apply some filter to the audio to reduce those frequencies
I think my new mic should fix this. I actually manually fixed one of the "S" peaks in this video, so I know exactly what you're talking about.
@@axcertypo nicecream
Still fairly new to Scrabble but have seen heaps of Nigel vids and it is accepted that he is the GOAT. But...wtf is that table? Why is the difference so insane? Damn.
_adroit_ and its derivatives are typically pronounced uh-DROYT (sorta rhymes with _Detroit_ ) in US English (per Merriam Webster). The word does come from French but it's not super modern; first recorded use 1652 per same source.
Cool vid and it was fascinating to watch (tho I admit I did so at 1.25× speed)
God bless him. Nigel is a fabulous inspiration and a wonderful genius.
Yeah that's a pretty sick setup. My setups usually backfire and make a winning game a loss because I miss that the setup I'm making has more hooks than just the one I'm thinking of
how do we know the percentage of endgames misplayed if we have no way of knowing for sure what the perfect endgame is?
Engines have progressed, and we can always use foresight and anaylze them for hours if we need, so it's more accurate now.
As far as whether all of the endgames in that study were brute-forced with 100% confidence, I don’t personally know, but to answer the related question of how you know if an endgame is better than Quackle’s (and what types of endgames it often gets wrong) - the way to check is pretty simple, you can put in the starting play that you think is better and let Quackle play the other side, and if it can’t find a way to stop or alter the sequence to make it worse for you (judging by final score differential) than the sequence it suggested, then you know that it wasn’t right (though this doesn’t prove that your sequence is best). It will usually make mistakes in scenarios like this where someone gets stuck with a tile, because like Josh says in the video it’s USUALLY best to slow play and maximize the points you get per tile, but you also have to be aware of other things, like how they can make setups and how they might be able to get more points in some spot than you, so you have to take the spot rather than let them use it even if it’s not the most points you can get for your tile. So play order and even passing your turn can be important. Like one scenario I remember where Quackle played a suboptimal (by quite a bit) endgame against me was when it had a tile it was going to be stuck with (the Q iirc) and played another tile for a good score, but that was the only tile it had that could hook onto EX, and I was able to play EX->LEX->ILEX->SILEX, scoring the X four straight turns (and I think one was a double word).
You can find more information if you search for Francis Desjardins ortograf. Desjardins collaborated with Gilles Blanchette to write an optimized, custom solver which can solve endgames perfectly. But the software can take a long time to run, so they have analyzed only a select set of endgames.
the tournament took place in Trang btw, i was there playing the main event
Edit: nvm i didn't read the frame! haha
I Won The Youth Plate Event With Tharin Kariyawasam (it was a side team tourney)Proud to be SL
Actually it's Adheesha Dissanayaka
And Kavindu Mallawarachi
We also received the best team award as we got the championship of the main event and the championship of the youth side event..❤🎉
what if ricky block with id, then de, then he pass so he saves his s so nigel can't play qi so he passes and so on so nigel could never play qi ?
This is covered, nigel will go out with qis instead, scoring 9 less points himself but denying etats (10) and getting 2 from the s. So it's 3 points worse to hold the s that long. Good practice to hold the s for a while just in case Nigel messes up, but eventually you have to play it.
I thought the answer would be to play very slowly and hold the IQI until Ricky plays his S and D and then play the 60 bomb
It's funny how AI beats the best Chess players, even Go players, but got nothing on Nigel Richards yet.
I saw playing qi and gi and cinq but couldn’t think of the word och (no idea it was a word)
do you know if scrabble engines take advantage of transpositions (a chess term that indicates positions that can be reached by multiple different move orders)? i kinda feel like "moves that score the same amount and don't interfere with my plays" would be an interesting transposition heuristic for scrabble engines, if only in endgames
Great question. Transposition heuristics don't really seem useful to me in Scrabble outside of the endgame, since the goal of Scrabble is not to checkmate, but to outscore. The concept of "interference", similarly to chess, is complex, since maybe your opponent isn't interfering with your plan, but their plan is stronger and must be stopped. The equivalent in Scrabble is creating a threat that must be addressed, regardless of score. The algorithm would have to understand the danger levels specific to the position at hand, and although there are clearly plays that "don't accomplish anything new", I don't know how useful it would be to track for those plays as an engine.
As for the Scrabble endgame, transposition is extremely important, since some endgames have billions, if not trillions of possible moves in them. So an engine very much needs to parallelize as efficiently as possible to get to the correct answer in a reasonable amount of time, after an exhaustive search. Currently there is only one engine that might have been able to solve this endgame as fast as Nigel did, although it's possible for engines to be fairly confident that something is solved in a shorter period of time. That's similar to how humans, Nigel included, process these things.
@@axcertypo cool!!! thank you so much for the answer :)
The GOAT is back :)
I hope these competitions have huge cash prizes, because I can't think of any other justification for spending countless years learning words you will never use in real life.
Hope away... passion overpowers many of life's constants, money being one of those. I won around 16k playing scrabble competitions in 2023, and that was a lucky year. I also spent maybe around half of that just to be able to play.
Almost everyone loses money on it
hivin dilimith is from our school
I neeed to know why Nigel played ETAT
I think it mostly had to do with endgame timing. If Ricky did in fact get a bingo down, he would likely not be able to threaten an outplay in the endgame and therefore not be able to win
I don't play Scrabble mutch (chess is my mind game) but i can really appreciate this analysis. Tfp..
are you half french ?
Not quite! I did my last 10 years of school in French, so that's probably what you're detecting.
What does EAU means in English ? As a French native speaker, I'm confused ^^'
eau is used in English when referring to "eau de cologne" perfume, as well as in reference to a certain liqueur.
Latest scandal! Scrabble champions test positive for steroids! The average hat size is 9.5!
I feel like you need to speak a little faster and brighter, and cut out the “um’s” and “uh’s”. But the content is fantastic and the editing is really good. And you explain things really well
I disagree with you on the way that he speaks, this is genuine. I would unsubscribe if he started talking like a youtuber
trying to find the right balance between genuine and flowing... don't want to do too many takes, but also don't want to be too rough around the edges... thanks for the feedback
Sup headphones?
Scrabble babble.
Lol, that person seems a little jewie(fishy) 🤣