try reading more carefully; because i never said that.  your reading comprehension is poor.  I said I cannot do it ALONE.  With others in a project I can do significant coding and maybe sound assets.  Without others I cannot.  I do not have the will to work on a large project without others also working on it.  It is easier for me to work when others are relying on me than if it is only myself to whom I answer.  Please read words moer carefrully before you assume.

Nobody ever said my people skills were good.

If there is a reason why the use of genetic algorithms and the other things I have mentioned would not work I would very much like to know.  As far as I can see the changes I suggest are feasible; only moderately time consuming to code; improve the ai over time; and use very little processing time.  They are also things that are almost never done in game ai design; hence why I believe its' an inadequately explored area.  The advantages of using large scale distributed processing for difficult problems are documented; but such are rarely used for game ai.

Hmm, I think i'll ponder what alternative ways of designing units might look like.

I actually think the foundations for this type of behavior is already in-game. You have initial faction hostility (Kingdom vs Empire) + Sovereign personality quirks (Unstable, Warmonger, Paranoid, ect?). If these personality quirks were more prominent then it seems feasible that such things could be implemented by simply modifying what is already in-game.

Reminds me of the Civilization Ghandi jokes. If these personality quirks were more evident in the diplomacy screen; such as effecting their posture or generated dialogue, I'm guessing it would be all the more awesome. It would be nice if we could modify these in Sovereign creation as well, as the only way to create your own unstable/warmonger/(ect) Sovereigns is to make XML modifications.

It would also allow the access to "experimental" Sovereign AIs that are not associated with the default Sovereigns. I would be surprised if there were not interesting AI behaviors observed in development that were shelved simply because it would not be appropriate with the "core" game. Letting players access this overflow might be really fun.

"Designing a game alone I can do, but all the coding, network, artwork; that's not something I can do alone."

The implication seems the same to me, especially since there is an assumption that you can work in a team and design the game in complete isolation from the input of your team, play testers, audience, or extraneous circumstances (exp: scope). On reflection if it is truly different, this may actually even be worse than the former interpretation.

There are plenty of examples of decent games that have been created with extremely small teams. Some that come to mind aside from Dwarf Fortress being Minecraft (initially) and Mount and Blade (original, husband + wife). Note also that I didn't even specify a completed game but merely a computational prototype. A prototype treated as a "large project" is... Odd. If you are not willing to even give a proof of concept with a prototype, why should others allow you to lord over them as their "designer?"

We already have that, it's called "novice" difficulty.

Don't get this. I have to design units for a faction in one game and then later play against that faction in a new game. To have a better experience. I don't play that many FE games! Please improve the game for me by giving the AI additional useful unit designs.

I made a bunch of units for FE if you want them...
Look at my reply here: https://forums.elementalgame.com/440288

Goes for the unmodded game though, 15ish units for each faction.

Sincerely
~ Kongdej

It appears there are some of these units in the 1.29 beta.  Seeing more things pop up than Spear Maidens.

The idea is that the AI will use what you've used in the past so that it ultimately plays the game in a manner similar to you.  It was a fairly non-trivial effort to make it do that and it's one of the best bang/buck features in FE (and something I think makes it stand out).

There are no "better" units objectively. There are only units that you find more effective and therefore, having the AI use those units too will make it more effective against you.

That is, of course, a straw man argument.  Nobody is asking for an AI that just rolls over and let's the player win.  All we're asking for is that the AI doesn't base its decisions on whether or not another player is a human.

Well we could start discussing some of the weaker traits... for the cost/effect anyhow

I never use the defender trait for once, especially not in the unit design package I made, because I know the AI won't just be standing around waiting to use its counter attack. (only real reason I see for getting more defense when I skip my turn)

I for one do like the feature that the AI grabs some unit designs and throws them at you, makes games much more interesting, I could argue that the game could have shipped with more unit designs, but then again, when to stop adding unit designs? And I experienced first hand how tedious and boring it is to add unit designs for each faction

Sincerely
~ Kongdej

I don't get this either. What happens if I'm bad at making units? The AI is getting worse.

Or, what if I design a unit to work a cetain way under certain conditions. The AI doesn't know this, and so pumps out units are are all but useless.

I like the sound of diplomatic recalculations, as described. Seems like a good first step to help remedy a strong game's worst feature.

I can agree with that.  They should play to win, and should not have a bias against any one player, AI or not.

My one remaining gripe, besides diplomacy which is getting an overhaul here, is the instant city razing.  This is really preventing me from enjoying the game because I cannot allow a single city to fall or to make a single mistake or I lose hundreds of turns worth of buildings and upgrades.  I hope this will be looked at for this patch.

It should take a few turns to destroy the city or maybe only a few buildings can be destroyed per turn and when no buildings remain can a city be razed.  I also don't necessarily mind the loss of fertility, I think that could be an interesting feature.  A destroyed city of a certain size, maybe level two, should be replaced with a "city ruins" tile that prevents rebuilding and monsters can spawn from to represent the wild reclaiming the land.

I usualyl make a new faction and sovereign for every game I play.  This means the ai tends to not have many new unit designs available to it for each faction, especially if I didn't play that game for too long.  Adding cross-faction unit designs would help; as woudl choosing optimal unit designs from the community on an ongoing basis.  This lets the ai pull from a wider pool for optimality.  Some of it also has to do with the fact that taking a base design; and upgrading it to more developed weapon/armor types may not be sensible; as it violates the point of some builds.

I'd say some units are "better" objectively.  While there's no BEST units; hill-climbing optimization indicates that some forms of a unit are markedly less efficient.  The AI also seems to not understand some of the objectives to be considered when designing units:  space-efficiency vs cost-efficiency for example.  The AI often seems to add things to cost-efficient units that make them no longer cost-efficient; and violate the original purpose of the other build choices; this is most notable when you see Slaves base units equipped with lots of expensive gear; if you're gonna add tons of gear and other cost-increasers, it's better to use the regular unit chassis than the slave chassis; unless you're doing it to make no maintenance garrison units.

Admittedly making a universal algorithm for designing units would be difficult due to the potentials of modding; so design copying is very useful; it's just be nice to increase the scope of copying to go beyond simply copying the units you previously used in that faction; but common unit designs in general.