CTC is fairly good about providing time stamps for that sorta thing. All the fluff is labelled as such in YT-player compatible format. I'm going to have to give this one a go. It's not usually a...

CTC is fairly good about providing time stamps for that sorta thing. All the fluff is labelled as such in YT-player compatible format.

I'm going to have to give this one a go. It's not usually a great sign if Simon can pull them off in 30 minutes, means the rules are those his Sudoku brain is overfitted to. I have more fun (and more of a chance beating Simon) if the rules are complete oddballs.

Initial thoughts for a solve after playing around for 2 minutes and not looking at the video

I'm going to guess that the "nearby must differ by 3" is a massive restriction on what a 3x3 box can even contain, and I'm going to guess that Simon knows what those combinations are. [I vaguely remember a CTC where such a neighbor rule combined with a knight's move constraint produced a global, tiling pattern and you just needed to figure out the offset and orientation of the tiling.] Particularly with those thermometers in there. I think that's going to be my first attempt, figuring out the combinations that satisfy the thermometer. The dot nearby (Top Left box) further restricts the next box over, but only slightly. That might end up being the tie breaker when choosing out of the few combinations available.

Well then have I ever got a doozy for you. From yesterday: The First Hardest Sudoku! (2h42min). :P

It's not usually a great sign if Simon can pull them off in 30 minutes, means the rules are those his Sudoku brain is overfitted to. I have more fun (and more of a chance beating Simon) if the rules are complete oddballs.

Oh man, that is a phenomenal ruleset. I'm not sure when I'll get to it, but that is one lazy saturday morning right there. Might be the afternoon too. And parts of sunday. We'll see. Edit: Tuesday...

Oh man, that is a phenomenal ruleset. I'm not sure when I'll get to it, but that is one lazy saturday morning right there. Might be the afternoon too. And parts of sunday. We'll see.

Edit:

Tuesday night, approx 3h in, I think I have a good grasp of the rules. The 5th column and 5th row are filled, 7 clumps are localized exactly. With the walls formed by the cross made from the 5th row and column, I feel like the upper left or lower right quadrants are where I should be looking now. Both seem close to collapsing, but not quite there.

Edit2: Approx 30 minutes more, and the top left corner collapsed. Into a contradiction. Dangit. Option one, there's a detail in the rules I've neglected a bit: You could theoretically make a "horseshoe" clump of 5 cells that intersects an arrow's line twice. It's a very very long shot, and I don't think it would actually be useful the way most of these things were constrained, but I can't positively rule it out. Option two, I just messed up.

It took two saturdays. Holy shit that is an extremely tough one. I don't think a sudoku ever took me this long. I had to reset partially after the previous issue. Wasn't a horseshoe that caused...

It took two saturdays. Holy shit that is an extremely tough one. I don't think a sudoku ever took me this long. I had to reset partially after the previous issue. Wasn't a horseshoe that caused it, just a regular old messup.

Yeah, you can usually locate when the solving starts if you read the video chapters names. But it's kind of a PITA, and so IMO it's way easier with SponsorBlock, or a direct timestamp.

Yeah, you can usually locate when the solving starts if you read the video chapters names. But it's kind of a PITA, and so IMO it's way easier with SponsorBlock, or a direct timestamp.

I have definitely noticed that quicker doesn't necessarily mean easier for these sorts of videos! I find I can usually do well if the video is about 45-60 minutes long, longer and it usually...

I have definitely noticed that quicker doesn't necessarily mean easier for these sorts of videos! I find I can usually do well if the video is about 45-60 minutes long, longer and it usually requires some really clever stuff going on, shorter and it usually requires knowing a trick.

Here though, I found I was able to derive the trick fairly easily, although I think that comes from watching enough of Simon's videos to know what some of the different interesting groups of numbers are (odd even, divisible mod N, high low etc). After that, I needed to spend a bit longer to figure out what the impact of that trick would actually be, but that first step was the key, I think.

Then I goofed the whole thing up by just writing the wrong number into a cell and ending up somewhere completely wrong. I went back to try and figure out what I did wrong, but I've got no idea what my thought process was at the time. I didn't get very far, though, so I'll probably give it another go eventually, just for the satisfaction of solving it.

Rules:
Normal sudoku rules apply. Adjacent digits must always differ by at least three. Digits along a thermometer must increase from the bulb end. Digits separated by a black dot must be in a 2:1 ratio (ie one digit must be double the other).

For those who don't have SponsorBlock, the Rules + Solving starts at 5m51s.

CTC is fairly good about providing time stamps for that sorta thing. All the fluff is labelled as such in YT-player compatible format.

I'm going to have to give this one a go. It's not usually a

greatsign if Simon can pull them off in 30 minutes, means the rules are those his Sudoku brain is overfitted to. I have more fun (and more of a chance beating Simon) if the rules are complete oddballs.## Initial thoughts for a solve after playing around for 2 minutes and not looking at the video

I'm going to guess that the "nearby must differ by 3" is a massive restriction on what a 3x3 box can even contain, and I'm going to guess that Simon knows what those combinations are. [I vaguely remember a CTC where such a neighbor rule combined with a knight's move constraint produced a global, tiling pattern and you just needed to figure out the offset and orientation of the tiling.] Particularly with those thermometers in there. I think that's going to be my first attempt, figuring out the combinations that satisfy the thermometer. The dot nearby (Top Left box) further restricts the next box over, but only slightly. That might end up being the tie breaker when choosing out of the few combinations available.

Well then have I ever got a doozy for you. From yesterday: The First Hardest Sudoku! (2h42min). :P

Oh man, that is a phenomenal ruleset. I'm not sure when I'll get to it, but that is one lazy saturday morning right there. Might be the afternoon too. And parts of sunday. We'll see.

Edit:

Tuesday night, approx 3h in, I think I have a good grasp of the rules. The 5th column and 5th row are filled, 7 clumps are localized exactly. With the walls formed by the cross made from the 5th row and column, I feel like the upper left or lower right quadrants are where I should be looking now. Both seem close to collapsing, but not quite there.

Edit2: Approx 30 minutes more, and the top left corner collapsed. Into a contradiction. Dangit. Option one, there's a detail in the rules I've neglected a bit: You could theoretically make a "horseshoe" clump of 5 cells that intersects an arrow's line twice. It's a very very long shot, and I don't think it would actually be useful the way most of these things were constrained, but I can't positively rule it out. Option two, I just messed up.

It took two saturdays. Holy shit that is an extremely tough one. I don't think a sudoku ever took me this long. I had to reset partially after the previous issue. Wasn't a horseshoe that caused it, just a regular old messup.

I honestly don't think I ever could have solved it, even with a bunch of hints and infinite time, so kudos! 👏

Yeah, you can usually locate when the solving starts if you read the video chapters names. But it's kind of a PITA, and so IMO it's way easier with SponsorBlock, or a direct timestamp.

I have definitely noticed that quicker doesn't necessarily mean easier for these sorts of videos! I find I can usually do well if the video is about 45-60 minutes long, longer and it usually requires some really clever stuff going on, shorter and it usually requires knowing a trick.

Here though, I found I was able to derive the trick fairly easily, although I think that comes from watching enough of Simon's videos to know what some of the different interesting groups of numbers are (odd even, divisible mod N, high low etc). After that, I needed to spend a bit longer to figure out what the impact of that trick would actually be, but that first step was the key, I think.

Then I goofed the whole thing up by just writing the wrong number into a cell and ending up somewhere completely wrong. I went back to try and figure out what I did wrong, but I've got no idea what my thought process was at the time. I didn't get very far, though, so I'll probably give it another go eventually, just for the satisfaction of solving it.

"Secret Whispers" by Fafrd