Apologies if this is a little obvious.

One of the design goals of this project is high modability, right? Well, that's one more reason to use adaptable, flexible AI rather than a more rigidly defined "Federation AI" "Klingon AI" etc.

Every race in the game (and any additional ones) are going to need AI, so rather than making hundreds of different AI scripts, why not make one all-inclusive AI system, capable of figuring out when to expand, when to start wars, when to focus on science or infrastructure...

The AI should have multiple interacting parts. First, the system-level AI should handle things like building food building to prevent starvation and such, and have multiple 'priority profiles' for science, industry, morale, warships, expansion ships, etc. This would double as the 'System Manager' for players who don't want to micromanage it. The mission-level AI would figure out which systems or ships are needed to carry out the goal and set their priority profile to the ones required. The empire-level AI would decide what missions the empire as a whole needed to accomplish.

The unique personality of each race can be distilled into a number of ratings of different 'mission' priorities.

Expansion determines how frequently the race sets their sights on a nearby system and tries to add it to the empire, either through colonization, conquest, or diplomacy.

Science determines how frequently individual systems are set to build and staff science buildings, as well as how likely ships are to go on science missions.

Exploration determines how likely ships are to go on exploration missions, and the 'depth' of exploration outside of the races territory.

Intelligence determines how frequently individual systems are set to build and staff Intel buildings, and if I get my way and Intel missions (things like parking your ship just outside (or inside, if you're brazen enough) another races territory and trying to decode their transmissions and scan inside their space, etc, also made easier by cloaking of course) are added, how likely ships are to go on them.

Fleet Size determines how willing the race is to build and keep ships above the 'free support' limit.

Aggressiveness determines how willing the race is to go to war to meet their goals.

Independence raises the threshold needed for the race to join another.


I think that ratings on those categories can adequately capture the spirit of just about every races' decision making, greatly reduce the workload of coding AI, make the game more compatible with mods, and also make the AI more moddable. Of course, the other side of the race personality coin is their morale bonuses and penalties, but if it matches up decently well with the ratings that should prevent it from having to be taken into account directly by the AI.

Let's talk diplomacy AI.

Firstly, I strongly feel that every race in the game should be treated the same by the game--IE, the Vulcans and Andorians should be able to form an alliance together, and the Klingons should theoretically be able to join as members of the Cardassians, even if it's terribly unlikely.

With that in mind... Inter-empire relations should track two scales--Like and Fear/Respect. Like would start out at a number set by how similar the other empire's AI personality ratings are, would slowly increase with time, and hopefully would be further impacted by ships on diplomatic missions (although this is conjecture). It would decrease (including into the negatives) based on hostile actions by the other empire against the first or its allies, or by 'interference' with the AI's missions, such as colonizing a system they were gearing up to colonize, thus causing their mission to be canceled. Fear/Respect would be calculated by the empire's relative strength, militarily, industrially, scientifically, and intelligencically (forgive the made up word, I was on a roll). Equal footing is a 0, weaker empires are rated with negatives, and stronger with positives. Each diplomatic action would have a threshold, achieved by adding those two numbers together, over which it would be successful. So two empires on equal footing would sign treaties essentially only from liking one another, but a mismatch could lead to the stronger being able to diplomatically strong-arm the other--to a point. Membership has an additional hurdle, that of the independence rating. So therefore, a race with fierce racial pride like the Klingons or Romulans or Cardassians would be correspondingly far less likely to join as members even if they would otherwise be inclined to.

Further thoughts outside the scope of AI but related is that member race systems which build colony ships should colonize planets with the member race, member races should keep their own morale modifiers, and if unhappy (and strong) enough member races should leave and take all their systems with them.

That's all for now.

Comments welcome and thanks for reading.

I have absolutely no idea what it takes to write any sort of programming, so i'm no expert on this. I do know that Mike hasn't even started writing the AI code yet though, albeit he has looked into different ways of implementing the code he needs. This also means that I don't know whether he plans to implement a static AI or a flexible one. That said, he's made some very good decisions about the game in the past, and he *knows* that we want the game to be moddable. Having a static AI would likely make it harder for people to create mods for particular races or whatnot. There's also the question of how the Borg AI will work as well. Mike isn't here right now to answer these questions right because his boss kidnapped him over the summer lol, but we're hoping he will be returned to us within the next few days or weeks.

We've always though said that we expect the AI to be with us within a year. The truth is that not even Mike knows how long precisely it will take, so this is merely a guesstimation; It could be sooner, it could be later. The AI is the one area that everyone wants the game to be better than BOTF though - that practically goes without saying. BOTF had an awful AI because it was unfinished and untested. In the case of Supremacy though, we've got thousands of potential playtesters and coders, so I don't think it's going to be too big a problem. There will likely be a long period of time between when the AI is first released, and the final version is complete though because people will need to playtest the game, notify us about any situations where the AI didn't behave as expected, and for changes to be made and the game retested, but that's half of the fun of making a game like this.

Since Mike is the only person who really knows what the Supremacy AI plans are, you could either wait for him to post here, or you could send him a PM. Either way, it might take a few days/weeks for him to reply, but he will once he gets the time to do so. I'll send him a nudge PM to at least let him know people are enquiring about his AI plans.