AI favor Nationalism branch TOO much!

[Risa]

In my last game (two about-same-size continents), when I finally conquered my continent and enter the industial era, AI's England had united her continent and enjoyed her industial era for a long time. But I caught up in no more than 20 turns. That is all because AI wasted its time in Nationalism branch.

That is an extreme scene, but AI tech choice in that scene is not uncommon. I think it's because Nationalism gives rifleman and a bunch of game options, Espionage offers covert missions, and Communism and Fascism are government techs, these factors weigh too much in AI's algorithm. Unfortunately these techs are all optional, and playing without these techs is not hard at all.

[alexman]

On the other hand, Nationalism is very important to the AI. It gets to draft and upgrade to riflemen, which ends the easy human Cavalry rampage.

But you're right. These optional techs (along with the Theory of Evolution) make it quite easy for the human to jump out in the tech lead in the Industrial era.

[Stuie]

If the AI just went after Nationalism and then moved on to required techs, that would be ok. Problem is, the AI seems to insist on researching Communism and Fascism while I beeline to Scientific Theory and grab ToE.

[dexters]

Communism was AI's preferred war government. Now, it's fascism. It does open up human tech trade possibilities.

The PROBLEM isn't AI preference for Nationalism, but rather AI predicatbility. If one AI civ has nationalism, you'd almost be guaranteed almost all AI civs will be researching along the same path.

The AIs not researching iron clad tech for example, is so predictable Firaxis/BA advertised it as a 'feature' or as a freebie strategy for human players.

I think a better solution is to give each AI civ more flavour by having them favour one path over another.
An AI civ may decide to go Steam- Electricity route and then have them reevaluate their tech paths every 2 or 3 techs. This process may certainly have an element of randomness with an rng diceroll as well.

An alternative, and this could be an easier fix is to use
The relative predictability to help the AI by writing special algorithms dictating special behaviors once certain conditions are met. For example, an AI civ that ia behind in tech, will be more prone to research techs that the other AI civs don't usually research, like Iron clad, in hopes of trading it away for something else.

A middle of the pack AI may also, through an rng dice roll, pick a tech from a list of techs that the AI is known to place a lower preference on, opening up the possibilities of advanced AI civs researching less popular techs.

[Risa]

In fact, I didn't mind predicability in some degree, if AI opted for a relatively best choice. After all, we human also play in some more-or-less-predicable patterns, don't we.

As a student of computer sciense, I reach the conclusion from my observation that AI's tech-choosing algorithm do have a random factor (or it contain much more factors than I can image), and it works fine in other circumstances, only gets messed at this early-industial time. So increasing weight of random factor (or those factors I haven't consider) will perhaps make AI too unfocus on a goal. That leaves us only rebalancing weight of early-industrial techs. Perhaps we should teach AI the importance of railroad (Steam Power), getting infantry and artillery asap and increase of worker's rate (Replacable Parts), production and defensive boost of metropolis (Sanitation), and the amazing bueaty of Theory of Evolution (Scientific Methods). Or we can de-evaluate factors given in Nationalism branch, though I have no idea how to.

BTW, in my last game the English did research Iron Clads. That gave me some time to visit Communism (for police office, of course).

[joncnunn]

Nationalism itself is very highly useful, especally for players who are playing building and hybrid styles and need their realitvely few military units that are concentrated as defended upgraded to Riflemen.

As for Communism, that tech is also useful for everyone.
Builders love the Police Station for loss corruption & waste.
Hybrids love the Police Station for staying a Democracy during their limited wars.
And some war-monglers love the Communism govt itself.

The fascism tech is probably overvalued by the AI. I don't see a thing about this tech that would apeal to players not intending to use this govt.

[Risa]

If I want to defend my country with limited units, I'd choose infantries and railroad system by beeline to Replacable Parts. Riflemen aren't strong enough to rely on. Of course it needs dozens more turns to research, but I can keep myself from total annihilation by diplo skills in that time. It works in SP, I haven't any experience in MP, though.

[steven8r]

In my last game on Monarch level, I fell behind the AI's in techs. The AI spent time researching the Nationalism branch (and Ironclads). I couldn't even trade for Nationalism. I finally got some Saltpeter so that I could use Cavalry to defend against Infantry.

I took the low road and bee-lined for Scientific Theory. I got it and ToE while researching Replaceable Parts (which gave me Atomic Theory). I was finally able to trade some of these techs for all of the others that the AI wouldn't trade to me before. (I still didn't have Rubber, but at least I could make massive stacks of Arty to defend against Infantry.)

Anyway, it seems that the AI DOES favor the Nationalism branch AND even Ironclads, but (for me) that's not always a bad thing.

[vmxa1]

The AI has long loved Nationalism, nothing really chanced, except they have Ironclad as well.