Does anyone here play Solitaire Battleships? Samples can be found here:
There is also an app (at least for Android, I imagine IOS as well) and these puzzles show up monthly in Games World of Puzzles.
I find that I can get through easy and some medium difficulty puzzles without issue, but at some difficulty level, I just get stuck. Conceptis says that no guessing should be required, but I often can’t figure out where to go without putting in a ship, seeing where that leads, and if it fails, backing it all out again.
I’m probably missing some trick (like X-wings in Sudoku) or tip or something, but I can’t find it.
Anyway, if anyone does do these puzzles at higher than medium difficulty, I’d love some advice.
It looks like the rule that ships can’t touch, even diagonally, is the big rule to pay attention to. For example, if I have a horizon space of 2 that I know one square will get filled in, that means the two spaces above and below will be water no matter what.
but that’s not guessing - that’s investigating a possible result.
The Conceptis ap allows you to make “pencil” marks (as opposed to permanent guesses) - similar to Sudoku. Even pencil and paper, you try something and if it doesn’t work, you back it out.
How is that different from guessing? I guess I’d like it to be like Sudoku, where you should be able to proceed logically from one step to the next, even if some of the relationships are difficult to see at first.
When I put in a ship, it doesn’t fail immediately – I have to go multiple steps, put in multiple pencil marks for ships and water, before I end up with some contradiction.
Surely you would expect some of that from an NP-complete puzzle
That includes Sudoku, and in an old thread I did wonder whether puzzle designers have any standard metrics for this category of puzzles (there are programs that will compute that a given Sudoku is “4.7 — Extremely difficult”) and whether they are firmly correlated to solvers’ subjective experiences.
I agree that ultimately a puzzle for humans ought to be designed to be solvable by a human.
I remember a Sudoku thread about whether there should be any guessing, and there was some argument over whether this or that was guessing. Then, I watched one of those Cracking the Cryptic guys solve a “vicious” puzzle and he just went along, step by step, and solved it. He used some techniques that I wasn’t familiar with, but none of it involved going forward five or six steps and then having to back out.
That’s what I’m trying to figure out with these puzzles – is there some technique that I’m missing, because moving forward a bunch of steps and backing up really feels like guessing to me, and Conceptis claims there’s no guessing.
Ten internet points for mentioning NP-complete, though!
What you call guessing, the creators probably call looking forward a few steps. Say I know that my 4-box boat can be in one of two places. I look at the board and can see that if I put it in Column A, that will fill the rest of Row 2, which means my 3-box must be in Row 5, which means I can’t place anything in Row 6, so I know my 4-box goes in the other spot. Was I guessing that the 4-box was in Column A, or was I using logic to deduce that it couldn’t be there? The difference might just be in your head depending on if you need to use pencil marks or not.
I can solve the medium and hard puzzles. I’d guess I use pencil marks on maybe a third of them. You probably know any advice I’d give - start with the 3s and 4s, fill in the known squares for 3s and 4s even if you don’t know their exact location, pay attention to the boats you have left, etc. You need most of those techniques even for the simple puzzles, the main difference for me is how many moves ahead you have to consider.
If you want, post a few examples and I can try describing my thought process.
Troutman, yeah, that’s basically what I do. If it really is about looking ahead multiple moves, then I’ll probably just stop playing them – I find that less interesting than solving it step by step.
Chronos, yes, I think I remember you saying basically the same thing in some Sudoku thread. So, if the creators of these Solitaire Battleships game say that no guessing is needed, what do you think they mean?
When I do a Sudoku, I can look at each square and figure out what can and cannot go there. From there, I see that, for example, if two squares only have two possibilities, then nothing else in their row (or column, or 9-box) can have those two numbers, so I can eliminate more possibilities. Eventually, I find squares that can only have a specific value, and that eliminates that value from the relevant spots. At no time have I had to say, well, this can be a 3 or a 9 – what if I put a 3 here. OK, that means this, which means that, which means this…uh oh, there’s no place for 5 – undo, undo, undo, etc. That’s what I call guessing.
Yeah, I’m pretty sure that I have said it before. And I still maintain that all of those “non-guessing” methods still involve guess-and-check, just on a small enough scale that the solver is able to store all of the relevant steps in their own memory, rather than on paper. I imagine that when the designers say “it doesn’t require guessing”, they mean “it requires a small enough amount of guessing that you can do it in your head”. But that depends on the individual head of the solver, since some have more memory capacity than others.
I agree. I just see it as an example of proof by contradiction.
I thinking something like “This square has to have a four or a six in it. I’ll think about what would would happen if I put a four in it. If I did I would have to put a nine here and that means this square would have to be a two. But this square can’t be a two because I already have a two over here. So if I can’t put a two here that means I can’t put a nine there and that means I can’t put a four in my original square. So it has to be a six.”
I guess I consider 1 step to be logic, and more than that to be guessing. So
“Let’s say I put in a one. No, try two. No… Ok, it can only be a three or four. … Ok,by filling in other possibilities in other squares, I know this is a four.”
Vs.
"I know this can be a three or four, but I don’t have enough information to decide. Let’s try three. That means this must be a four, that must be a five… (Ten steps later)… No room for an eight. I guess that one was a four. Now what did I fill in based on that three… "
One difference is that I never have to remember the path I went down. In Sudoku, I can put in the possibilities and then they can get eliminated as more is filled in.
Of course, some would also say that putting in multiple possible numbers in a square is also a method that is “beyond logic”, in that it’s something that isn’t done entirely in one’s head.
