Insane Solving Times
 

This has been written about ad nauseum, I know. But it really intrigues me.

Could one of you speed demons make a video, with live audio commentary? I know the commentary would slow it down (relatively) significantly, but I am so curious. 4x7 most difficult level.

I bet the skilled players are not solving logically but are doing some sort of pattern recognition, maybe even remembering the puzzles from past efforts.

Thanks to whomever tries.

Wow! great question. can't wait to see replies!:)

This is impossible. There are many puzzles that have the same basic information, but not the same answers. Maybe one name in the puzzle will be changed. I solve maybe 15 puzzles a day, and the really dedicated solvers do many more than that. Usually at least a month goes by before I get the same puzzle again, and there is no way that I could remember a specific puzzle. I know that some have said that they do recognize patterns, though.

i recorded a couple of smaller, easy ones, no sound tho. i tried to use the mouse to track my eyes and order that i went...

https://www.youtube.com/watch?v=GKpeYGAFiIo
https://www.youtube.com/watch?v=mSQK27-ii3I


i agree, i think that it's highly unlikely that anyone is recognizing patterns. there are some clue 'tricks' that you can do.. but in my mind, the time is a matter of the order that you do the clues, how many of them you can maybe keep in your head so that you don't have to go back and re-read the clue, and just how fast and accurate you can click...

and if you ate your wheaties and spinach that day. :-)

.bri.

The logistics of 'pure' logic in a difficult 4x7, to solve in < 4 minutes, there has to be more to it then reading and reasoning.

Yes, the logic reasoning can extend into eyeballing the grid and ferreting things out based on patterns of cells checked/not checked, that is logic, but it isn't tied to the clues directly. But a valid method nonetheless.

I would think one would have to spend at least a minute just reading and marking the no-brainers. I just don't see how the rest can be done that quickly without some bizarre insight into the visuals of the grid. There is just no way to read & think that quickly. There has to be a geometric component or something like that, something that can be done from a distance without picking through minutiae

Couple that with the sheer volume of repetitiveness, there has to be some repeating pattern recognition going on here.

I have read the accounts of people who claim they have cataloged and studied and seen that there are only so many unique patterns and that when one learns to pick up on the clues defining a pattern or family of patterns, one can swoop in quite dramatically to a speedy solution.

It has to be a big picture process, there is no way to digest. You drown in information, from a speed perspective.

Thus the request to see a 'real time' solve process to just see what is happening. I bet more time is spent looking at the grid than the clues.

So the best way to put an end to this is to actually see a documented solve. The process may not reflect the true intuitive insights in play, but if the player could describe what they are seeing/feeling, it would be at worst interesting.

Interesting ideas you have there. I think you're right about spending more time looking at the grid than the clues. Ever since I read your OP, I've been paying closer attention to how I solve these, especially when I get a nice, low number for my elapsed time. I think I pick out the most relevant keywords in the clues and combine that scanning with trying to remember all I can as I work the grid so that I don't waste time going back through the text.

I've just watched the first one. The steps and methods used were EXACTLY as I would have done it, but the speed and seemingly almost simultaneous execution of steps is quite interesting. It looked like you knew what to mark a step ahead of what you were doing: You made an error and continued on to the next mark and then went back to fix the error.

I am an engineer and am programmed to think things through methodically. I am happy with that approach, I enjoy figuring things out that way. Low times are cool but actually being able to untangle one of these bit by bit is what I like most.

I noticed in your first video how quickly you were able to mark the boxes from the 1st pass through the two relative clues. I do the exact same thing but in probably 4x the time. I do them in two passes - I mark the exclusions first (A is more than B, so I mark A != B), then make a 2nd pass and set the actual relative quantity exclusions. But in real time in your video, it was almost instantaneous that you did this (ignoring the A != B part - that being the bottom section).

So the methods you used were virtually identical to what I would have done, but done way more quickly than I, with what appeared to me to be almost lightning fast realizations.

I typically do 2000 sec. on a 4x7 most difficult, using my point-by-point method. I just tried a couple easy small ones, they were done in 100 sec. roughly.

Thanks for sharing, I'll check the 2nd one later.

great information
 

At least now i have a better idea of why my times are so tectonically slow!:)

Others might not find that impossible. Depends on how the memory works.....

i don't do the clues in order, i try to scan for the 'easy' ones - shorter ones, ones that apply to the top sets of rows, anything that marks a 'yes' takes priority over a 'no', etc. once i get to the point where the 'easy' ones are done, i usually start at the bottom and go up, or maybe do the first one or two and then start at the bottom. i assume the clues are ordered 'randomly', but still it 'feels' sometimes that i can pick a 'better' order and get there faster..

i haven't done it, but it would be interesting to look at the clues in depth and come up with the 'optimal' order for those clues. there shouldn't be the case where a clue isn't needed (i would consider that a flaw in the puzzle generation, but that's probably a bit pedantic on my part :-), but there could be the case where the 'correct' order of clues would mean that some clues wouldn't need to be processed two (or more) times to be completed.

.bri.