The WORST Scrabble game ever is a marvel of construction!

No video

The WORST Scrabble game ever is a marvel of construction!

Рет қаралды 3,452

Күн бұрын

Recently I posed a new Scrabble challenge to basically come up with the LOWEST-scoring possible tie game that fully emptied the bag. The lowest score initially stood at a mind-boggling 42-42.
That record has since been obliterated by our renowned Scrabble constructors, Amaranth and lovemathboy. HOW few points did these geniuses score? I'm not gonna spoil it here, you'll have to watch the video and marvel at this brilliance yourself!
Do you dare go lower than the score in this video? If so drop me an email at mackmeller@gmail.com and you might be featured in a future vid!
Previous record: • Think you're bad at Sc...

Пікірлер: 72

@AmaranthRBY Ай бұрын

Playing the same word that your opponent just played is kind of like en passant in chess. Doesn't matter how big the point sacrifice is, you need to do it whenever possible, purely for cool points I do not defend any other play made in this video. Hope everyone's having a good Tuesday, thanks for featuring us again Mack

@mackmeller Ай бұрын

Of course, thanks as always for giving me content with your awesome constructions!

@throwaway1743 Ай бұрын

You may not like it, but this is what peak Scrabble looks like

@bueger Ай бұрын

You may like it, but this is not what peak Scrabble looks like

@halcyonzomboid Ай бұрын

Mack: 2:18 ever-valuable S Also Mack: 7:05 S's are terrible

@mackmeller Ай бұрын

Was hoping someone would notice this! It was intentional, though admittedly I was running out of things I could use to justify these horrific plays so it was inevitable even if I didn't try 🤣

@LeviATallaksen Ай бұрын

LOL, so funny to hear your satiric justifications for things like keeping all power tiles together and sacrificing 100+ points to set up a 60 point play😂 Great content, and great construction by Amaranth and lovemathboy, not just shattering the record, but also giving you racks that set up a lot of those comments!

@AmaranthRBY Ай бұрын

Putting something funny in the racks was pretty quick and easy compared to the rest, it was a no brainer to add in a few things for laughs after spending so much on the board itself. That might be stage 2 of this challenge though - if 29-29 is maxed out, the only way to make this even worse is to maximize the number of points that the players miss out on :)

@LeviATallaksen Ай бұрын

@@AmaranthRBY Yeah, obviously it's the easier part, but I appreciate that you added something that wasn't part of the challenge itself, for us to enjoy! Interesting idea to also look at the number of points sacrificed! Did you already try to optimize that when picking those racks (even if you didn't put a lot of effort into that part)?

@mackmeller Ай бұрын

Hahaha glad you enjoyed it! And yeah all the missed 100+ point bingos definitely made my job easier :)

@AmaranthRBY Ай бұрын

@@LeviATallaksen Haven't really tried to optimize no. I just reverse engineered a bit with Quackle: for example on the OPIOIDS turns I knew that the players 100% needed to have DIIOPS on their rack, so I gave them DIIOPS? and saw what Quackle thought would be the best play - turns out it's EPISODI(C), so I make the last letter an E and job done You could definitely go a lot deeper playing around with the order of the plays themselves, almost certainly you could have several turns in a row of missed triple triples if you set your mind to it

@Riokaii Ай бұрын

i'd propose a 3rd challenge the "unluckiest maximal scoring game" Essentially, rather than play the worst available move each turn, play the best, but still minimizing the score. whats the worst rack of tiles you can get in sequence for a single player (resulting in likely an extremely one sided largest score differential) and separately for both players in combination

@AmaranthRBY Ай бұрын

This would definitely add a whole new layer of difficulty to construction, it would have a ton of depth and require way different approaches from the ones we've used for the two challenges so far. Pretty fascinating idea on the surface, maybe we'll make it come to life one day :)

@mackmeller Ай бұрын

That's a cool idea, though definitely way more complex as Amaranth points out -- would love to see it though!

@emmeeemm Ай бұрын

This is like the time that GM Magnus Carlsen and GM Hikaru Nakamura played a Double Bongcloud chess opening to a draw. It's impressive for entirely atypical, non-traditional reasons.

@mackmeller Ай бұрын

Hahaha that's awesome, I'll have to check that out!

@GwenMason125 Ай бұрын

Very excellent construction, kudos to Amaranth + lovemathboy! It's worth mentioning that in your video, you said that the only way to get to a 28-28 would be to cover two fewer double-letter scores. This is not necessarily true, in fact, most of the what is being optimized for this challenge are the number of intersections between non-blank tiles, since this also causes a tile to be scored twice. For example, if there was a layout that covered one more double letter spot but had three fewer intersections, that would also achieve a 28-28. Best board layout I could come up with was a 30-30 after realizing it was best to play to two of the double letters on the cardinal points. I didn't even think to play down from the other blanks into one of the double letters, since I thought that would make too many intersections.

@mackmeller Ай бұрын

Ah, that's a very good point -- I was thinking number of intersections didn't matter if there were no parallel plays, but that's erroneous as you correctly noted. So 28-28 is definitely still conceivable... who's gonna get it!?

@GwenMason125 Ай бұрын

@@mackmeller haha, I love the optimism, but I'd be willing to call this challenge complete. I could give some numbers as to why I think this: The top and left quadrants that reach the double letter on the cardinal lines are by far the optimal sequences for filling space, not just for that area but for the whole board, since they contain the largest stretches of non-score spots in a straight line. You can only do it for a maximum of 2 quadrants though, since reaching a third would require playing 7 tiles at the start, which is a bingo. The sequence averages 3.5 tiles placed per doubled point, while the crisscross sequence through the bottom and right quadrants average 3.22 tiles placed per doubled point. The next best shape I found from my research kind of looks like a capital F and averages 2.9 tiles placed per doubled point. We need to average 3.74 tiles placed per doubled point to get a 28-28. Not only would we need to find a better crisscross, we would need to find a better quadrant-filling sequence, and significantly better. Even if we added the rule that bingos don't add 50 points to your play, which would allow us to reach a third quadrant and fill its space using the optimal placement, that would still result in a layout that leaves 3 in the bag. If we also allowed these last 3 tiles to just "float in space" and be scored as if they were played, but without incurring any additional doubled points, we would get a 28-29. There are several other variations to this 29-29 layout, for example, symmetries and/or moving one of EAU or MOA to under CEE, but I don't think there are any 28-28 layouts unfortunately.

@polygon586 13 күн бұрын

@@GwenMason125 There is a 28-28 layout; but unfortunately it leaves so many letters in relatively open areas that I don't think it's possible to shed all dangerous vowels elsewhere. Hope I'm wrong :) But doubt it.

@I_like_smashburgers Ай бұрын

I love how Mack tries to justify every move played, even though the moves played were designed to be the worst.

@mackmeller Ай бұрын

Had to make it entertaining in some way!

@nootnootnits Ай бұрын

As someone who played @AmaranthRBY over the board just two weeks ago (and was unable to ebb his ferocious tide of bingo bongoing in our first game) at a cafe in Rome, I can confirm that he is, indeed, an incredible Scrabble player!

@mackmeller Ай бұрын

That's awesome! I'll have to hit him up if I'm ever there :)

@theepicosity Ай бұрын

your commentary is so entertaining and insightful, thank you for explaining such brilliant and ingenius moves that i never could haave comprehended!

@mackmeller Ай бұрын

Thanks! Yeah I really had to dig deep to understand some of these superhuman brilliances

@mpeterll Ай бұрын

Covering fewer premium squares doesn't necessarily allow for a lower score. That would have to be done without increasing the number of turns because each extra turn gains at least one more point from a tile already on the board.

@Anthropolite Ай бұрын

That hurt so much to watch but the 29-29 tie at the end is impressive

@mackmeller Ай бұрын

Hahaha

@thegodofpez Ай бұрын

The sarcasm is over the top. Loooving it.

@Aploplex Ай бұрын

I don’t think it’s possible to beat this lol

@mackmeller Ай бұрын

We'll see!

@Firefly256 Ай бұрын

18:32 I think there's another way, which is to play more tiles in a turn If you play EAT, then (EAT)EN, you'll score 8 points If you play EAT, then (T)EN you'll score 6 points If you play EATEN, you'll only score 5 points

@emmeeemm Ай бұрын

Keeping the 14 highest-scoring tiles on racks seems like the strategy for this challenge. The remaining tiles are worth a collective 110 points, because those 14 highest-scoring tiles are worth a total of 77, and the whole bag is worth 187 in face value. I googled it and the letter and point distribution is actually on Hasbro's website. So, a minimum of 110 collective points must be laid on the board, and a maximum of 77 points could potentially be collectively subtracted at the end. 110-77=33. And each player has to have half of these remaining mandatory points, so under the constraints of this challenge, each player must score at least 16.5. Therefore, I believe the lower bound on this challenge to be a theoretical 17-17 game, because you cannot gain or lose half a point in Scrabble, 16-16 is less than that absolute minimum calculation, and 16-17 is not a tie. I am not the guy to sit here and actually find such a game. But 17-17 would be the score I would shoot for if I was looking, and I encourage the community to get as close as they can to this bound.

@amconners Ай бұрын

it's not possible to score only 110 collective points with the 86 lowest scoring tiles because you can't avoid scoring at least some amount of points from tiles already on the board. this obviously doesn't matter on turn 1 and you can avoid extra points for up to 3 more turns by only playing through or overlapping the blanks, but it takes a minimum of 15 turns to play 86 tiles without bingoing (86/6=14.33) and those other 11 turns will each have to score at least one point from a tile already on the board. adding those 11 points to the 33 points from your calculation results in a minimum combined point total of 44, making the lower bound at least 22-22 (I think bonus squares also make 22-22 impossible in practice but I don't know by how much)

@emmeeemm Ай бұрын

@@amconners Ooh, great point. I did forget about the connectedness of play. Thank you for the correction on the theoretical lower bound.

@shichengrao1387 Ай бұрын

I think we can go a bit higher here. On the board, there is no spot for 6 consecutive letters. So any 6 letter play necessarily involves at least one extra point (either something goes on a double letter, or you go through an existing double letter). There's also only 4 spots where 5 consecutive letters can be played without hitting a bonus and 8 (I think) places where 4 can be played. Thus the theoretical optimum is to use all of those 4*5 + 8*4 = 52 tiles over 12 turns and then either use 6 letters scoring a bonus point or 3 letters without the bonus point (equivalent since 6/2 = 3/1) for the remaining 34, which gives 34/3 = 11.33 = at least 12 additional points. So you have 12 turns (4 of which can be pointfree with blanks) and 12 additional points, giving 8+12+33 = 53 points. So the minimum I think is 27-27 (given 26+26

@Iridescence93 Ай бұрын

Took me a minute to remember NYE is CSW only as it would play on the last board for (relatively) big points

@AmaranthRBY Ай бұрын

Yeah, CSW would make this challenge a little harder - most notably you need to make sure your Ns are blocked because of NY, which is the most 'trivial' fail (below ENGS). Hilariously on this board CH also ends up relevant (you could play AH/CH at the intersection of ABOARD/CROUTON). Then there's a bunch more 3s that fit - NYE like you said, but also E(X)O above IBOGAINE, and that's just for this particular construction. I have no doubts a CSW board that ties 29-29 is also constructible even with all these obstacles, it would just take a little bit of searching

@craiglarimer1173 Ай бұрын

Incredible video. Hilarious. Loved the phrase about the “bad” rack. Roflol

@nicholasgott3590 Ай бұрын

Just a thought for a challenge I'd be curious to see. Can a game be constructed where player 1 can take their turns normally, but player 2 can never legally make a move? Player 2 can only trade or pass, but based on player 1's moves they never can make a move. If not fully possible, I wonder how many tiles player 1 could play off before player 2 could make a move. Love these videos, I'm an absolute fan of puzzle/brainteasers.

@rohitg1529 Ай бұрын

I did a quick calculation and obtained a theoretical lower bound of a 17-17 tie. This assumes all but a single double letter score is used (to ensure a tie). Of course this is very optimistic, since purposefully avoiding all the bonus squares forces much shorter plays, which means it will take a longer number of moves to empty the bag, and increase the likelihood of players running out of legal moves before the bag is empty. I would be surprised if 29-29 is truly the optimum, but it is likely not more than a couple points off from the true optimum. Calculation: The scrabble bag contains: - 0 points worth of 0 point tiles (2 blanks) - 68 points worth of 1 point tiles (68 tiles) - 14 points worth of 2 point tiles (7 tiles) - 24 points worth of 3 point tiles (8 tiles) - 40 points worth of 4 point tiles (10 tiles) - 5 points worth of 5 point tiles (1 tile) - 16 points worth of 8 point tiles (2 tiles) - 20 points worth of 10 point tiles (2 tiles) As in this construction, we want the players to have all the 5+ point tiles (5 tiles) and 9 of the 4 point tiles on their racks at the end. This results in a total deduction of 9*4 + 1*5 + 2*8 + 2*10 = 77 points. Supposing the players manage to avoid ALL bonus squares, they must have to play all the remaining tiles, which would net them at least 2*0 + 68*1 + 7*2 + 8*3 + 1*4 = 110 points. The combined score would then be 110 - 77 = 33 points. Since the game must end in a tie, the actual lowest possible combined score is then 34, with a 17-17 tie at the end (with one player using one double letter score, or from an overlap play). It would be mathematically impossible for both players to do worse than this.

@tender_rain Ай бұрын

Going further, because we need to play 86 tiles for each player to be stuck with 7 tiles, and we can play a maximum of 6 tiles each turn, we have at least 15 turns in the game. On each turn, we get one extra point by playing through a letter that's already on the board, except on the first turn and on two other turns when we play through each blank. This means we get 12 extra points this way, making the theoretical minimum a total of 45 points, or 23-22.

@tender_rain Ай бұрын

This is, of course, not possible in a real game because bonus squares get in the way. In fact the only way to play six tiles in a turn without encountering a bonus square is if we play through an existing tile that's on a bonus square. This means on 7 of the 14 turns where we play 6 tiles, we need to set up a tile on a bonus square (racking up 7 more points at least) and the other 7 turns we play through it, for a theoretical minimum of 26-26 if we play six tiles a turn.

@rohitg1529 Ай бұрын

@@tender_rain Great analysis. I think this bound can still be improved upon to 28-28. With the exception of the first move, when both blanks are used on the double word score, we actually don't want to hit any double word scores. By my count, the only places to play 6 tiles while hitting only a single double letter score are: - row 8 through the initial play (1) - columns H and I through one of the blanks (2) - rows 2 and 14 between the double letter scores (2) - exactly one of columns D or L (1)

@DIPHENHYDRAMINEMAN Ай бұрын

You can also do better by playing more 6 letter words, aka by having less overlaps, since overlaps add extra points

@mackmeller Ай бұрын

Ah true, good point!

@Slaydrik Ай бұрын

and also if those double letter spots on the top and left can be avoided it'd bring the score down very slightly

@Slaydrik Ай бұрын

oops, i didn't watch the video enough -- he mentioned that, lol

@richards31415 Ай бұрын

Here's a challenge idea: by total # of tiles played, what's the shortest possible game that must end in a 6-pass (due to no possible valid plays from either side)? The idea is similar to your previous VAV/v(A)v short with 5 tiles played, but this would not count as one player could hook an S or play AK(VAV)IT, for example. I feel like the answer would involve creating a 3x3, 4x4, or 5x5 block of tiles for which no word can possibly be hooked or extended; not sure.

@jellomochas Ай бұрын

H, X, and Y are playable in CSW

@ekcrisp1 Ай бұрын

Appreciated the funny commentary showing the big scoring plays but would have preferred giving insight into the game construction to explain why these were chosen over other low scoring options

@jahonasnapkins Ай бұрын

they must know something we don’t

@Slaydrik Ай бұрын

It took me a while to figure out how they would get such low point values, until I recalled how they'd lose so many points from not being able to play such high-scoring tiles

@mathematicskid Ай бұрын

this is hilarious

@maxlimit9129 Ай бұрын

Wait, Amaranth? As in, the Pokemon player Amaranth? Damn that crossover's wild

@AmaranthRBY Ай бұрын

@@maxlimit9129 That's me sir

@violetasuklevska9074 Ай бұрын

29-29 is likely the minimum, if you remove the tied game constraint the minimum should be a combined score of 57, baring dictionary nuances. I'll try to give an idea of why this is. Throughout I'll be assuming that both players are stuck with 7 tiles at the end of the game, worth a maximum combined face value of 77. The difference between combined score of the players and the sum of the values of the tiles on the board at any given moment (before the 6-pass) is the number of double-letter squares the have been used and non-blank word intersections (assuming no other bonus squares were used, the tiles on double-letter squares and the tiles that are a part of word intersections are worth 1 point or less and the first play uses just blanks, which is the case in this game and the 42-42 tie). For instance, in the final position before the 6-pass, the players have a combined score of 135, the tiles on the board add up to 110, 7 double-letter squares were used and there are 18 non-blank word intersections, giving us: 135-110=25=7+18, I'll refer to this difference as DIF. To get a 28-28 tie the DIF of the final position must be 23, and to get a final combined score of 57 DIF must be 24. After the first 5 opening plays in this game 26 tiles are played and DIF=4 from the 2 double-letter squares and the 2 intersections. You can get DIF=4 with 29 tiles played in 9 moves in the 7x7 central squares similar to the 42-42 tie, however it's not a great board shape. If we are aiming for a 28-28 tie and assume a 26 tile opening (these 26 tiles are part of the center) with DIF=4, similar to this game, then the remaining 60 tiles must be played with DIF=19 for the rest of the game or an average of 60/19≈3.16 tiles played per 1 DIF. In the game the sequence (e)ERIE, ELO(I)N, A(L)MNER, LEA(N)T, F(E)NS, plays off 20 tiles with DIF=6 from 2 double-letter squares and 4 non-blank tile intersections for an average of 20/6≈3.33 tiles played per 1 DIF (tiles remaining 40, DIF remaining 13). A similar potential 5 word sequence of (E)AST, TOU(S)E,... on the right side of the board would play off 19 more tiles with DIF=6 for an average of 19/6≈3.17 tiles played per 1 DIF (tiles remaining 21 DIF remaining 7). With each of the plays of AER(O), OIL(S), INT(O) and ENG(S), 3 tiles are played per 1 DIF (tiles remaining 9 DIF remaining 3). The board has been completely exhausted and no spots on the board allow for 3 tiles to be played per 1 DIF, the best we can do at this point is something like AB(O)ARD on the top right that plays off 5 tiles with DIF=2 and an average of 5/2=2.5 tiles played per 1 DIF, much less than the required 3.16 average. The remaining 4 tiles can be played off for DIF=2 for a final DIF=24 and a potential combined score of 57, but no 28-28 tie. Triple and double word squares are taboo, and when playing through a triple-letter square at best you are getting to play 12 tiles with DIF=4, barely breaking even with 12/4=3 tiles played per 1 DIF and an insufficient board shape. Now if we aren't playing through triple-letter, triple and double word squares, the board can be separated into the center and the 4 quadrants where plays can be chained without playing through massive bonus squares. The center gives a nice head-start with 26/4=6.5 tiles played per 1 DIF, of course with the aid of the blanks. The aforementioned chain of (e)ERIE, ELO(I)N, A(L)MNER, LEA(N)T, F(E)NS with 3.33 tiles played per 1 DIF in the bottom quadrant is the absolute best ratio per quadrant, not counting the opening plays. In most cases you will struggle to play more than 6 tiles in a quadrant while staying afloat with 3 tiles per 1 DIF (like in the top quadrant just counting the plays of AER(O) and OIL(S)). Now 6 tiles per quadrant is criminally low, making the pace of 3.16 tiles per 1 DIF completely unsustainable. With the 29 tile opening with DIF=4, there are 57 tiles left and to get a 28-28 tie we need DIF=19 for the remaining plays or 57/19=3 tiles played per 1 DIF. However, in a given quadrant, there are NO sequences that break even with 3 tiles played per 1 DIF, the best is 11 tiles with DIF=4 or 11/4=2.75 tiles played per 1 DIF; most openings create bad boards and this one is no exception. One last thing, on the final board, if a blank tile switches places with a 1-point tile that is both on a double letter-square and part of a word intersection then the combined score would increase by 1 point, this is because the face value of the blank gets multiplied by 4 over the course of the game while the value of that 1-point tile gets multiplied by 3. In this case switching them is counterproductive and in other cases switching them might break even, which is to say the blanks are best used on the opening move (unless you are trying to use the triple-letter squares because you've exhausted all other ideas). This is by no means a complete proof that 29-29 is the minimum tied score under these circumstances, rather it should give you an idea of the many problems that exist when aiming for lower. If you are trying to disprove a 28-28 tie conclusively you need to examine specific cases of how the center and the quadrants might look like, which I don't feel like doing. If you are trying to improve on this challenge in any way I suggest going for the 57 combined score, one of the players might even get a negative score. You can also try to get a 29-29 tie in CSW, but someone has beaten you to it ;)

@tender_rain Ай бұрын

I personally don't think this can be beaten. The problem with playing more six letter words is that you also then hit a double or triple letter score, thus negating the advantage gained by more turn around. It's a matter of optimizing the number of letters played vs bonus squares hit. I couldn't get a score below 29-29 and that was without playing legal words, I was just trying to create a board setup. Creating a valid board while also keeping in mind that no vowel can be exposed out in space and even some consonants such as M, B aren't allowed because they'll form words with the Y isn't easy. Also nice touch keeping the H and Y on separate racks to avoid 6K (T)HY. This was a very well constructed board!

@tender_rain Ай бұрын

PS: I believe the lowest theoretical score possible on a board with zero bonus squares would be 23-22, playing six tiles each turn (33 pts from the letter point values and 12 points for each turn where we need to play through an existing non-blank tile already on the board). Edit: did some calculation to account for bonus squares in another comment, and it's not possible to go below 26-26 in an actual scrabble board playing six a turn, and playing fewer each turn would rack up more points because more turns, so I think 26 should be the theoretical minimum. Please correct me if I'm wrong!

@MrCheeze Ай бұрын

I'd be interested as well in the records for max and min points WITHOUT making the game a tie. Especially if it turns out the tied record really can't be improved.