OT What I really want

[Numdydar]

is a game design that provides the following difficulty against an AI

Very easy - "I win" button. In other words, the player will never lose [:)]
Easy - Player wins about 80-90% of the time
Normal - 50/50 chance of winning against the AI
Hard 25-35% chance of winning
Very Hard - <10% of winning

I would not want the AI to 'cheat' either by starting with vastly improved units/locations/resources/etc. The AI should have the same options at all times as the player(s) but increasing the difficulty just adds in additional routines that increase the number and depth of things that the AI looks at. I would be willing to accept a slowdown in the AI turn processing as a trade off for this.

While I love the Civ series (especially Brave New World), adding extra units/etc. as the level increases always bothered me. At least in DW, the amount of control you have over game setup can adjust things to an absurd degree. What would be nice is an initial screen with the levels above that automaticly sets all the game setup options and THEN allow you to go through and adjust as needed.

I wonder when we will be able to get to a game design like this? I know some are getting closer but I think we still have a ways to go.

[Canute0]

Very easy - "I win" button. In other words, the player will never lose

Thats not possible, you allways find someone who didn't even manage to power on the PC.

I would not want the AI to 'cheat'

Oh the dream of each programmer to create a real AI that can beat a human.

[mSterian]

Heh.
AI's are actually pretty hard the first times you play the game.
The AI "knows" the game and can do some pretty damn quick calculations. A lot faster and more accurate than you can.
And it has implemented a strategy tested by the developers.

The problems with the AI's is that they're not dynamic. They don't adjust much to counter what they're up against. They don't adapt, they don't anticipate your strategy.

They have fixed decisions, taking into considerations just a handful of factors which are even sometimes wrongly interpreted.
And you get to know them the more you play. You start knowing how to manipulate them and counter them.

So.. in the long run, the only thing that's left to adjust for an AI to have a different levels of dificulty are numbers.
Give extra % tax income here... decrease maintence by % of that..... etc. If they wouldn't "cheat", all levels of difficulty would just seem the same to you after a while.

We can just hope that in the future, smarter AIs might be created. (but please not one like Skynet that becomes sentient and tries to erradicate all life from earth )

[Numdydar]

Yes an adaptive AI would be nice as well [:)] Of course if someone else started to play the game, then it could get really confused [:D]

[Jeeves]

What you really want is multiplayer mode. Maybe version two will have that, but it has the downside of being more turn based rather than pause-able realtime...

Lonnie Courtney Clay

[Icemania]

ORIGINAL: Numdydar
is a game design that provides the following difficulty against an AI

Very easy - "I win" button. In other words, the player will never lose [:)]
Easy - Player wins about 80-90% of the time
Normal - 50/50 chance of winning against the AI
Hard 25-35% chance of winning
Very Hard - <10% of winning

I would not want the AI to 'cheat' either by starting with vastly improved units/locations/resources/etc. The AI should have the same options at all times as the player(s) but increasing the difficulty just adds in additional routines that increase the number and depth of things that the AI looks at. I would be willing to accept a slowdown in the AI turn processing as a trade off for this.

I completely agree. I can't think of a better competitive advantage against the bulk of other 4X strategy games out there than to focus on this in the next expansion.

While an experienced human player will always defeat an AI without cheats, at the moment, there is a lot that could be done in DW within the normal limits of an Expert System e.g. take some of the most experienced players on the forum ... watch them play ... and build up as many rule sets / styles as you can ... then implement them.

By all means after that consider a more adaptive system but you've got to get basic AI right first. I've just spent some time watching the game on Auto and to be honest was pulling my hair out at the complete and utter stupidity of almost every AI decision.

Personally I'm not interested in multiplayer (tend to play an hour here, an hour there).

[dostillevi]

An adaptive AI is often a holy grail for developers but it has a number of factors against it, not least of which are how difficult it is to program such an AI and once programmed how to balance it against the player's skills. It is generally much easier, less expensive, and really generally more fun for the player to build an AI that is competent and is able to respond to and enact a number of scenarios.

I of course would love a game with an adaptive AI and I've got 2 in mind that I would love to make right now if I had the time and resources.. generally speaking such games must have a "dungeon master" AI that works behind the scenes to regulate the emergent behavior of learning AI in order to provide an entertaining experience for the player. Truly competent AI would obliterate human players without regulation.. imagine the kind of overwhelming attacks you use against your neighbors being used against you, for example. Consider if your actions actually had political consequences that mattered and how complicated it would be maneuvering through the political minefield in order to wage a war without having the "allies" come crashing down on your doorstep. Imagine well designed fleets with ships built for specific purposes, some kiting your missiles while others sit at range unloading their fighter bays. The AI can truly be a grandmaster at a level that very few people are able to achieve, and figuring out how to dumb down such an AI to be challenging but not impossible to beat is at least as hard as making the AI in the first place.