Possible To Make Clones With Deterministically Solvable Boards?

Area to discuss development and bugs of official clones (Clone, Arbiter, MSX, Viennasweeper)
Post Reply
Double X
Posts: 3
Joined: Fri Aug 26, 2011 3:35 am
Location: China, Hong Kong

Possible To Make Clones With Deterministically Solvable Boards?

Post by Double X » Sat Jul 07, 2018 11:39 pm

As we all know, sometimes we've to guess when playing minesweeper. Unfortunately, as least in rare cases, it causes some otherwise very good records to be thrown away. Since there are already some minesweeper versions that can guarantee deterministically solvable boards, I've been thinking this for a while.
As far as I can imagine, the main problem of this will be the drastic reduction of the board cycles, causing those trying to go for the dreamboards having a much, much easier time. So how about using 1, or both of the below setup as well to mitigate this?
1. Make a separate minesweeper record set for clones with deterministically solvable boards. That way, the records of these games with a lot smaller board cycles won't mix with the original ones. Also, we know that we can somehow know how good a player is just by inspecting how he/she handle guesses.
2. Make a new variant of deterministically solvable boards - Rather than filtering out boards having to guess, make boards that, upon detecting a guess, let players to always have the right guess. This can be done by changing the position of the involved mines at that moment. As they're forced guesses anyway, this won't impact the other areas of the board. Also, this setup won't shrink the board cycle at all, even though it might cause some boards to become dreamboards due to the effective disappearance of those forced guesses.

Posts: 9
Joined: Thu Aug 24, 2017 1:16 pm

Re: Possible To Make Clones With Deterministically Solvable Boards?

Post by sugarsugarsugar » Wed Jul 25, 2018 5:40 am

I played 100 exp board with 'the first cell is blank' mode, and 20 of them are deterministically solvable. I think this rate is not so low and the board cycle may keep long.

The former idea looks good, but the latter looks not good for me. Example is here.
t.png (8.06 KiB) Viewed 542 times
Case 1-1 and 1-2 are easy examples of that the cells are affected by other areas.
Case 2 is an examples of the big difference of the strategy between this rule and the original. The probability of that the cell marked star is mine is 7/8, but to open this cell is optimal to solve this board in the rule.(This pattern extend infinitely)

Double X
Posts: 3
Joined: Fri Aug 26, 2011 3:35 am
Location: China, Hong Kong

Re: Possible To Make Clones With Deterministically Solvable Boards?

Post by Double X » Tue Jul 31, 2018 12:45 pm

Then maybe the 2nd idea can be changed a bit - only letting the players always have the right hard guess, like those obvious 1/2 guesses. For soft guesses, just leave them as they are.
I don't know how to clearly differentiate hard guesses from soft guesses, and I don't know whether there's a way for any official clone to implement this, but I'll still try to explain:
1. In case 1-1, assuming that there are no other unsolved areas:
- If there's 1 or 4 mines left, then there's no guessing.
- If there's 3 mines left, then there are 3 possibilities. If the grid at the upper left corner or between the two 2 is clicked, then the revealed number will cause the remaining click deterministic. If the grid above the 3 is clicked, then the next click must be the one between the 2 two. As the former involves reaction time for checking the newly revealed number while the latter doesn't, I agree that the 2nd idea will change the strategy in such cases.
- If there's 2 mines left, then there are also 3 possibilities, and similar analysis applies. In this case, clicking the grid above the 3 would be optimal, as the remaining 2 mines must form a diagonal, and choosing either one will be right, due to it being a 1/2 guess anyway(strictly speaking, that's not true due to corners having a slightly larger chance of having a mine, but I think that's negligible here).
2. Similar cases can be applied to case 1-2.
3. For case 2, I don't know how to elaborate, but it seems obvious to me that no hard guesses are revealed yet, as least when compared to those obvious 1/2 guesses.
So for case 1-1 and case 1-2, if it's possible for official clones to detect guesses involving better strategies that would not be better without the 2nd idea, then those guesses should be forcibly regarded as soft guesses anyway :D

Posts: 20
Joined: Thu May 08, 2014 8:27 am

Re: Possible To Make Clones With Deterministically Solvable Boards?

Post by senfti » Wed Aug 08, 2018 5:33 pm

MySweeper (can be found in official clones section) has a "lucky mode".

In this mode the mines are rearranged if
- a mine was clicked
- the probability of a square being a mine can be calculated for the whole board (computational expensive, therefore in many difficult situations impossible, but works for all simple guessing situation if the rest of the board was solved)
- the square with the highest probability of being no mine was clicked

So if the best square was clicked you cannot loose.

Post Reply