Your little nice words wont get you anywhere, "Brother" Lal. By now everybody knows youre just a mealy-mouthed hypocrite who pretends to help others while infiltrating their colonies and building planet busters behind their backs. So fuck off.
ITZ may yet happen in our lifetime.
kyrub said:
I don't see AI build much of anything. I'll try to pay more attention next time I'll play.
NO. In fact, they are pretty poorly thought out, as is most of SMAX content. They are 1) too costly for the technology tier they occupy 2) don't affect resource restrictions 3) not very useful, as land terraforming is always superior to the sea one. I can see them being used in sea bases with no access to land tiles, but those have a very low mineral output, making building any structures - safe from basic necessities - a long and arfuous task.
There are different AIs. I have noticed that Miriam and sometimes Morgan/Gaia play the builder game, with Miriam dabbling into momentum and thus being more efficient. Yang is the most successful of AIs, and he plays more as a momentum player (don't bother with buildings, expand, expand, then expand some more, then swarm everything), but then his bonuses/civics support this kind of play.
The tricky part is that terraforming land tiles of a nearby ocean base is very hard operation for the AI, unless it owns a base on that same continent. Sadly, projecting colonizing sea=> land seems really hard. The game was not built for this idea, it came only later with Svensgaard. AI thinks about colonizing other continents once its continent is nearly full. The sea is considered as a continent as a whole.. You see the outcome.
This could actually help the AI as well for pop-booming.
(Miriam for builder? I have not seen that one.)
Hey, I just share my observations. You decide what to do with this.
AI already prioritizes food far, far too much. It ends up with bloated bases full of drones it can't manage, sitting on a lot of farms and not producing anything of value.
She builds a lot, I think, when she isn't at war. The latest PBEM we played with your patch, she seemed to neglect an army in favor of stuff like Energy Banks. With not a lot of energy to go around, too.
Yup. It seems that you can assign similar priorities to two factions, and one would still fare better than the other. In fact, I can try and dissuade Yang from expansion, and he still fares better than a highly encouraged Morgan!
Hehe. I have gone too far with Energy Banks in general.
Won't matter as long as AI probes regularly end their turn next to enemy units/bases.
Gaians. +2 police and cheaper native life is nice.
Virtual World - University. Obviously.
It protects from rockets and direct buster hits (this feature is bugged, though). I am not sure how important it is, but it is certainly 'nice to have'. Not a priority, though. It would be better to concentrate on a Tachyon Field.
What do you mean by 'self-shooting'?
Well, I had Miriam with -4 Planet build those in every base. You tell me if it's any good. +2 POLICE and cheaper wild life - I think at least Greener AIs need to give it a go.
Never used those. Subspace Generator is an alien victory condition, no?
This would explain a lot. Every once in a while I see AI bases that just riots for years, most visible with Uni bases.
It manages, though it does so terribly. It seems like it converts citizens to doctors/empaths if there is a riot and if there is enough food. Specialists would make for a better choice most of the time.
Obviously, I am going by the original + expansion rules.To say the truth, it is hard and it requires a lot of space to code flexible options for Computer players, I said it on ac2 formus from the beginning. If you want to improve the AI (with primitive tools like hex editing) and you face changing rules, your task gets exponentially bigger. What I see as a definite possibility, however, is to make adjustment for the AI for a typical set of rules, agreed and used by the community. It's not hard, I would just switch off parts of the AI code, maybe I would recode a few locations. I suggest making a "terranx.exe" file adjusted for a community set of rules.
I worked on both issues in the last version of the patch. AI did not value +PLANET at all, if it did not have massive EcoDamage or a "Psi,25" in faction.txt file (this was probably intended for Gaians as I find it in all the AI code, but it does not switch on). It should be better now and it can be further improved. I am quite sure I saw AIs with FM defending during a war in game simulations.
Some of these were tackled in the patch (I remember putting a lot more weigh to Wealth). The biggest problem was the overpricing of the + MORALE and +SUPPORT, and +GROWTH partially. I suggest you look again at the result after I post the patch and we talk about it further, I am open to a lot more improvement in the SE department. And yes, the AI always analyzes the values only. So if it runs SIMPLE instead of GREEN, it means that --GROWTH is too much of a penalty and it is not outweighed by ++PLANET and ++EFFI.
If SUPPORT is overpriced, shouldn't we see more Police States? Instead they run Fundy during wars and Democracy during peace time, almost exclusively so.
Not caring about Planet I can understand, but EFFIC is what kills AIs, and they already have a major emphasis on food, so a little penalty should not hurt them too much. The importance of --GROWTH is definitely overstated - unless they somehow hit -3, that one I can at least understand.
That's an interesting question. At some point in my experiments with faction bonuses and custom units I have made every other AI build mind worms for some reason. I don't know what I did to cause such a behavior, and I can't reproduce it anymore. Did I make worms cheaper? I don't remember. But they were definitely preferring those over regular units.
I think I got that wrong. They simply count support minerals. Similar with effiency, they count the energy saved. They don't care about bureacracy drones, I think. With morale, they quantified the combat units, per base... and since the units number grows with time and AI likes to churn many units, MORALE becomes the overwhelming choice.
I changed the approach to GROWTH a bit, if I recall, putting emphasis on Growth = 6, or Growth =4-5 + Creches to enahnce pop-booming at some stage. I may tone it down further and add the Growth=-3 line.
The unit price is a factor in all AI decisions, if you changed that, it surely helped. Other things, I would have to look up, maybe later.
It is, if AI can be taught to spam bases. Its +2 research bonus, even though it looks insignificant, should not be underestimated. Could work well for Yang, for example.
Reduce their cost to 3 mineral rows, and suddenly they are a pretty decent unit. Still situational, but worthwhile. That is a work for modders, though.
Scout patrols.
I quite like the idea, I reworked them for AI bases with lack of Energy output.
Yep. Maybe things like this could make it into a community driven patch, then we would make the AI build less worms.
Hey, I remember this from my first and last game on SMAX, the moment when I realized that natives build normal scout patrols. It felt so wrong, thematically.
Kyrub the only problem I see with your patch (I'm not really sure if it was a vanilla problem), is that the AI doesnt defend their cities worth a shit. 1, 2 and at the very most 4 units per city (Which isnt horrible but its reserved for capitals etc). Some with no defense at all.
I did a lot of testing by letting the AI duke it out on their own for about 200+ turns, and it seems to be a problem every game.
I havent tested this newest version, but unless the AI builds more units based on a better economy, I doubt it would change much
edit: And I have no idea how I missed the part about buying less probes. Maybe that will have an effect. Testing now
edit2: seemed a bit better after running about 3 games. Capitals were sometimes fully stocked. Still, it seemed like the cities were way to weak, in some cases being garrisoned by a single 1-1-1 scout unit at year 2300.
