With the DLL coming out, is anyone planning to update the AI?

I'm not a programmer, so I wouldn't be able to do it on my own, but if anyone is planning to do this work and needs game design help (I am a professional game designer) or planning/organizational work done, I would be interested. I would like to see a Deity level AI that doesn't need three free techs at game start and half a dozen free units.

 

Modders *do* wish to update the AI, but I don't think that there are specific plans for doing so in a fully-laid-out fashion.

 

Indeed, and knowing the structure of the code. Some things may be quite simple to change, and others quite hard, and we can only know how hard it would be to make a particular change once we get a chance to see the actual code itself.

 

Is it too early to start organizing? Of course the DLL might be a neutered version compared to Civ 4, and it might not be possible, but presumably fixing it won't be a 1-person thing (unless there's an engineering prodigy hanging around to do it), and setting up the contacts a month or so in advance doesn't strike me as premature.

It's of course premature to talk about exactly how to fix it, though.

.shrug

 

That being said, the offer may interest some people. I am a programmer myself with absolutely no desire to delve into game design. I will probably be too bothered with another non-civ5 project by that time but it may interest someone else.

 

I would love to see many AI mods with this release, but I can just imagine the difficulty of such a task. The idea of a smarter AI seems to be on everybody's minds. I will probably still be working on my scenario by the time this comes out though, so any delving into such code will have to wait till later.

 

I shall expect all diplo screens to begin with: "Hi, I'm Troy McClure - you may remember me from such Civs as..." :

 

It's an enormous effort for very little reward, and this is coming from someone who has the technical skills, game knowledge, and interest to conceivably work on something like this.

It's quicker to design and implement a turn-based strategy game from scratch and then write an amazing AI for that. It really is that much work. It doesn't excuse how poor civ5's AI is, but we're talking about one of the hardest problems in all of gaming.

Firaxis has designed an excellent strategy game where the strategy comes from the way different game mechanics come together. I saw a lot of people crying about the "combat AI" like it's some isolated thing. Here's everything that influences the "combat AI":
- roads
- gold (upgrades / upkeep )
- technology
- diplomacy
- production / resource management
- units
- terrain
- city placement
- exploration
- social policies

that's every mechanic in the game, especially if you look closer at something like "production / resource management". the interplay between different aspects of the game and the resulting tough choices are what makes civ great. But that hurts the AI tremendously.

Realistically, it should be possible to improve individual aspects of the AI so that it needs to cheats less or can cheat in less obnoxious ways. But getting the AI to the level where it can actually play the game is a big project. It's too big for casual modding. I believe Civ4 (a simpler game in terms of AI programming) tried something similar... it still needed a ridiculous amount of cheating.

 

As I said in another post (before discussion veered off into settling Mars and other topics), there is a lot of focus on some really hard AI topics. Of course, I'll be very happy when someone with the skills takes on this hard stuff. I'd surely welcome some actual tactical AI.

But in the meanwhile, why not knock off a difficulty level or two on some easy stuff? We don't have to raise it to human-level in one go after all. As Biz_ says above, it should be possible to reduce the AI need for cheats or at least the obnoxious level of cheats that it currently gets (at Emperor+).

I say "easy" when I know full well that nothing is easy. My point is that there is a lot of relatively "low hanging fruit" when it comes to AI improvement. If anyone buys into that thesis, I'd be happy to lay out a few (I think others could think of more).