The AI is very pragmatic

[MarkShot]

This is not a knock, but just an observation.

Rome fought a short but decisive war against Macedonia (the only really challenger to a Roman future). It was a decisive war as it was not intended to defeat Macedonia through directly inflicting losses. Instead the war was intended to create an unstoppable divergence between the Roman and Macedonian trajectories.

At the start of the first war both Rome and Macedonia were Glorious (Golden Age) Empires. Rome's strategic objective was to kick start the de-evolution of Macedonia while maintaining Rome's current trajectory. This worked quite nicely.

By the time Macedonia would start the second war, Rome would still be a Glorious (Golden Age) Empire, but Macedonia has already slid back to an Old Decadent Empire, and still sliding.

Rome's strategy was after War #1 to demobilize. To what point/purpose? Mainly cost savings. Rome's military was powerful enough to buy one year of security for Italy. That meant one full army, two security forces, and one full fleet. Italy had the ability to recruit one full army and one full fleet in a year. And the Roman economy has the means to support 3 full armies and 2 full fleets in the field continuously.

Which brings me to the point about the AI being "pragmatic". The AI seems to consider a faction's current combat strength as to whether it should declare war, but not its future potential. So, on paper in static numbers, the Macedonian attack was reasonably calculated. But adding in the Roman economy and infrastructure, it was an epic blunder by Macedonia/AI.

Again this is an observation, and not a knock on the AI, Empires, or Pocus.

Comments?

[poesel71]

I think you are quite right.
In one of the diplomatic screens there is a force comparison modifier. IME if this is positive for you the AI won't attack. This does not take into account, as you said, the economic situation but also not the position of your army or its current situation.

Example: the AI is in the north and your army is far in the south and occupied with another war. If your army is still bigger than the AI it won't attack even if it would be advantageous for it in this situation.

Maybe the AI could weigh in the distance of armies in its calculation. Not by a huge factor - but it could tip the scales in some situations and create nasty surprises.

Another idea for AI attack probability would be to factor in the army sizes of current neighbouring enemies. Those armies could be deducted (with a factor) from your army size for the calculations. More enemies would make AI attacks more likely. Caveat: this has potential to snowball.

[MarkShot]

If we are adding considerations ... then:

(1) Your number of enemies who are free to declare war (and their actual and potential forces).

(2) Your ability to recruit based on cash/manpower/metal reserves and per turn income.

---

My general approach is to avoid huge standing armies and instead invest in development. Only have a response force sufficient to buy time to mobilize.

Another thing that a human would have spotted right away ... my target (Rome) was Macedonia. In 10 turns, Rome and Epirus went from spitting insults to full alliance. Thus Rome was now in easy cross border striking distance of Macedonian objectives. Guaranteed supply. And allowed the Roman navy to easily cover the flanks of the invasion force.

Nothing says war like the forces of a third party staging on your border.

[Gray Fox]

Markshot posted: "My general approach is to avoid huge standing armies and instead invest in development."

Okay, I'll bite. A military needs money, manpower, metal and equipment. None of that is needed to "develop" a region. Overstocking these leads to more admin burden and waste. I often do not max out my military, so that I can use those resources for diplomatic transactions. Is that what you mean?

[MarkShot]

No, it is only money which when at peace can be put to better use reinvested in policies (like bumping trade); yes, you pay up front, to bump, but the overall ROI can be very nice.

But yes, you can use those other resources to trade just as money. Sometimes trade deals and diplomacy is the shorter path from A to B than force of arms.

[Pocus]

Actually that's not completely true. It might be not working as well as I intended too, but I believe that's because how the AI thinks.

To answer directly the question, yes the AI factors comparative economy in the 3 resources when checking the chances for war. But contrary to a human mind, where a strong argument or counter-argument can be a definitive YES or NO for a given choice, then the AI works mostly by accumulation, i.e even if Rome economy is stronger, it will just be a few % of de-incentive-ation (please find me the correct word ) .
And then that's all wrapped up with a probability roll. You can rationalize that by saying it is a blunder from the enemy ruler, a bad roll.

[MarkShot]

Pocus,

The word you want in English, I believe is disincentive.

Until you license Googles Deep Mind, I think the AI is good enough. Since this is the underlying problem with all game AI's that don't blatantly cheat ... the AI neither learns nor reasons.

Like in my current game, I set up a brilliant war ... in so many games ... it is like that ... try that with an experienced human gamer and you will get very painfully pre-empted.

[MarkShot]

I remember the days, you used to say "IA", and "she". As soon as you said "she" and not "it", I felt very intimidated!

[dunnsa]

I remember the days, you used to say "IA", and "she". As soon as you said "she" and not "it", I felt very intimidated!

Athena!