[Kde-games-devel] Kmines

[Matthew Woehlke]

Ethan Anderson wrote:
>> I'd rather see an option 'don't allow unsolvable puzzles' where
>> "unsolvable" is defined as indeterministic (i.e. impossible to complete
>> without random-guessing). Of course this applies to *any* board, so I'd
>> add the condition that we assume any contiguous area of at least X
>> 0-neighbor squares is initially revealed when making that determination.
>>
>> Partly that's because having a border feels like it would affect the
>> style of play adversely... but maybe not. I certainly wouldn't object to
>> it as an option.
> 
> It would alter it, yes.  Adversely, maybe-- I can sure see it decreasing the
> length of the game when all the 0's on the perimeter dig cavities into the
> map from every side.

Yes, and avoiding those games that start with 0-3 random clicks that 
turn up 3 neighbors followed by an unlucky one on a mine :-).

>> Well, mines is a two-part game, where one part consists of finding
>> "large" 0-neighbor areas by random guessing, and the second part is
>> wholly deterministic. There isn't really any "AI", just a series of
>> logic rules to be applied.
> 
> Artificial Intelligence is... a series of applied logic rules.   ..and
> tubes.

I guess I think of "AI" as necessarily having some random factor as 
well, or at least "complicated" rules or some form of "learning". 
Otherwise to me it's just a rule engine.

@Richard: I *would* like to see an algorithm that checks for 
deterministic solvability :-). Having no choice but to guess sucks ;-).

> If you can do it, your machine can do it.

Shouldn't that be the other way around? ;-) I can visually identify 
objects. Computers aren't very good at that yet (and it's arguable how 
good they are at OCR or speech recognition, both of which humans are 
much better at).

-- 
Matthew
FOSS: Giving people the cage and it's blueprints, gratis. The "L" matters.