Adjustable AI strength

[Skizzaltix]

hey why not even copy the default ai 3 times & call them easy medium & hard. Gives the illusion of choice (which people want) doesn't leave the negative first impression that an easy AI would leave..

...Actually a pretty good idea
I suppose you could give medium +1 income and hard +2 or something like that, if you were hung up on morality.

But on the other hand, it's easier and IMO just as effective to leave it as-is, and just let players toggle income and starting gold when they want a harder match.

[Dragon280]

It may be hard to create a good AI for wesnoth, but is it that hard to create a worse one?

Heck could just change the current ais to be called easy, medium, hard. With "hard" corresponding with default.

Would keep new players happy.

But you would end up getting threads like "do you think normal difficulty is harder than hard difficulty?"

Or if an easier AI were made new players may start multiplayer even less-well prepared, by being deluded that they are competant & average by being able to beat the game on "normal" easier than what we have today!

I don't think it would be such a good idea to incorporate easier difficulties because of the problems above.

As for my opinion of the difficulty of the wenoth AI as it stands... its not really that hard, its competant in small areas (and makes good moves). If you made it dumber when I had picked "easy" as an eager noob, I think I would have cast aside the game thinking it was too easy & therefore crap.

So yeh its ok at the moment...

hey why not even copy the default ai 3 times & call them easy medium & hard. Gives the illusion of choice (which people want) doesn't leave the negative first impression that an easy AI would leave..

Forking the AI is a good idea. However, extensive play-testing is necessary to see how each AI behaves in various situations, since the AI can behave much differently in different situations, and it would be difficult to have a consistant difficulty level for each AI. Creating an illusion won't work; the AI needs to truly be easier at lower levels. As I said, I don't expect this to be ready until version 1.8 or equivalent, so there's no rush.

On the other hand, there is virtually no way to adjust the AI in MP without editing the Python code. Since many players don't know Python (or any other programming language), there needs to be a way to easily adjust the AI via the game's normal interface (that is, through the menus and the GUI and not the internals) for MP.

[PROTEUS]

Um....I think people have started repeating themselves -_-
Unfortunately, I can only reinstate what has been already said
If you truly want an EASIER game, then just give yourself more gold/income and/or give the computer less gold/income.

I might also add that it is probably for all intensive purposes impossible to write an AI that with exact resources on a MP game can even come close to the better players. Xianqi (chinese chess) has many similarities to modern western chess (fixed board size, units, sureity). It has a game tree complexity for 10^150 and we still have yet to design an AI that can surpass top players. Now, with the game complexity of Wesnoth easily reaching a few thousand times more than Xianqi I do not believe that it's fair to really expect that out of a FREE game.

It's nice to dream....but I'm sure that the devs have worked fairly hard to get the AI up to the point of where it is and as it has already been stated...

For MP, there are different AI you can choose while you're assigning the sides. However, as far as I know, none of them are that much better than the original one we've had for years. Turning up the "strength" of the AI isn't all that easy.

...developing further AI may not necessarily even bear anything "harder" (or "easier") for that matter strategically with noticeable effect. HOWEVER gold is a good way to 'tweak' the difficulty. I would definitely have to agree with Dave that

There are parameters available to scenario designers to allow them to tweak AI behavior and analysis depth, hence essentially tweaking AI difficulty.

The biggest problem, however, is that tweaking the AI to adjust difficulty is much harder than tweaking the resources available to the AI. In some scenarios, changing the AI's analysis depth might make a huge difference in difficulty, while in others it may make almost no difference at all. It is very difficult for a scenario designer to predict how much exactly changing the AI's analysis depth will alter the difficulty of their scenario, and would require a great deal of play testing to work out.

For these reasons, I think most scenario designers stick to tweaking resource availability to determine difficulty, rather than trying to tweak AI behavior.

David

So as far as I've gathered from this and other similar posts, that's the general consensus.