Is it cheating to use "it can be solved" meta-logic?

[Cvphilo]

I am just curious about people's opinions on this. Obviously, it is a game, played for fun, and you can do whatever you find fun.

So occasionally (not often, but every now and then) you get a situation of the following type: "I know exactly one of these three is blue. Two of them have no clues (including no possible clues that might be revealed later) that distinguish between them. Therefore it must be the third one that has a clue, a clue that must eventually show it is blue or else the puzzle will not be solvable (but I have not gotten far enough on that clue to have it yet)." Going ahead and clicking that third one is 100% deterministic logic, but it does not follow from the rules of the game, per se, but from the meta-rule that it is solvable.

A variation on this occurs often on random levels where the count resolves the last several clue-free cells, you often get to the point where you know that the count must tell you "all the cells that have no clues, and cannot get a clue, are black [or blue]" before you have identified all such cells. Is it cheating to just go ahead and resolve them? You are depending on the knowledge that the final count will resolve all these cells and it could only be in one direction. Granted once you are at that point, it is almost always easier to finish without doing that. But is it even a legitimate option in your opinion?

[SirDmitry]

Hexcells is a puzzle game. Any logical solution is legal and sometimes you can't solve puzzle without meta-logic described above.

[Cvphilo]

Thanks for your input. That seems like a reasonable "ethical" rule. Kinda like when someone crafts a puzzle with a particular pathway in mind and you find another one instead.

I believe that the random puzzles and all the dev-crafted puzzles never require "it can be solved" logic. The rules for creating the random puzzles precludes that. There are some user-crafted levels in the Reddit that were specifically designed to require using this logic. I am not sure I have ever encountered any others that actually *required* it. That is, even when I find myself saying "hey! I can use this", there is always a path that does not require it (I just might not see it!).

[Berahlen]

The only puzzles I've ever seen that require that kind of meta-logic are user levels that specifically call it out as an intended gimmick. If it's in the main game or generator, you just missed something.

[Cvphilo]

Yes, that seems to be true for both the random generator algorithm (I believe the dev has said as much) and all the dev-designed levels (the latter could have required this, but empirically did not). So I take that to perhaps mean "the designer would consider this an exploit, not a legitimate move", though perhaps not. Thus my question.

The situation that got me thinking about it was a random level where I was able to use this logic to get out of being badly stuck. Of course, I was stuck because I was missing what a collection of clues elsewhere told me. But there I was, able to use this to get out of being stuck. (FWIW, after I realized what I had been missing, I went back and redid the puzzle without using this trick.)

[Boksha]

I'm curious though; does this ever actually make a big difference? By the nature of these meta-logic "guesses", they can only happen in places with very few hints, so you probably won't be unlocking any new hints either, and you'll still be stuck after eliminating or filling in the hex; the actual benefit won't come until you've used the hints the generator used to verify the puzzle until you're nearly at the end, at which point it may make counting hexes a bit easier or unnecessary.

I kind of suspect it won't make a huge difference, so I also wouldn't worry about it much.

It would certainly be easier to reason about if there was a clear example involved though. Screenshot time?

[Cvphilo]

Ah, a fair question, one that forces me to concede that for the motivating case I could NOT actually do this, but thought i could. Here is a screenshot: https://steamcommunity.com/sharedfiles/filedetails/?id=2609369284 Look at the 7 flower.

Now imagine that you were badly stuck on this puzzle, as a result doing the next moves in the active clue area at the bottom and then not noticing the last blue in that 5 diagonal. Because that is why I was stuck . So now you have to further imagine that at some point during clicking back to Hexcells to try again you accidentally clicked the cell at the very top (it is non-flowering blue), which apparently I did . The screenshot is a replay and I realized that I could not have really gotten that cell. And that cell (when finally determined by the count-up at the end) could resolve which of the two cells to the left of the flower was which.

So having done all that, you have an example of the phenomenon. An incorrect one, as I concede, but there we have it. It does turn out that clicking the third candidate cell at the bottom of the flower lets you move forward because the other clues for that cell are then resolved.

So that is how it *can* work. It is also a pretty good dissertation on how it happens so rarely, and how there really is always another way out for the generated puzzles, that it really doesn't matter much. But, hey, Hexcells really doesn't matter much either, and here we are.

[Phoil]

It's not clear to me what your meta-logic is at that point in the screenshot. When I solved the puzzle, that entire 7 flower was resolved before needing to do any count-up at the end (my final count-up was at 2 remaining).

[Cvphilo]

I know I goofed in my reasoning (both an errant click, apparently, and just missing the "right" way to solve the puzzle which inevitably existed). So I guess I might have yet another error or it might be that you did it normally and never reached the frustration where you might want to use that reasoning (which, again, requires the errant click). I was trying to present that as my motivating example in response to the reasonable request for an example. It was never a great example, unfortunately.

[Phoil]

I created a custom puzzle (Two ways to solve) where you can use meta-logic as the first step, and using the meta-logic step makes the rest of the puzzle easier to solve. The meta-logic step still isn't easy though (and hopefully I didn't make an error in the meta-logic, it's not something that I can automatically check is correct).

[Cvphilo]

Thanks, Phoil. Is that posted in the Reddit that makes the puzzles easily accessible, or in some other form I can try it. (Note: I admit I have no idea how to do either of those. I just rely on the kindness of strangers to provide them :-)

[Phoil]

Yes it's posted in the subreddit, so it appears under "User made levels" in the game.

[Cvphilo]

Great. Thanks.