AI research priorities - a problem?

Theseus · May 2, 2002, 11:42

I've noticed on the strategy threads that post-1.21 everyone is much happier with the AI's approach to war, specifically in terms of mass of forces and the use of certain units.

One of the things that has jumped out is that with the attack flag now switched on for some defenders, the AI often attacks with huge stacks of riflemen, infantry, etc.

I'm not sure if there's a problem with its use of mounted units, as I have seen major attacks with Knights, WEs, etc.

BUT...

Last night I got into a HUGE war with India and Russia, playing as Japan.

I was waaaaay behind in techs... at points I was fighting WEs with Horsemen, and defending with Musketmen when the AIs had Riflemen.

So I beelined for Military Tradition. When I got it, and after some tech extortion, I realized that while I was the only civ with this tech, and everybody else was up to Industrialization!!

Now, this makes no sense to me. It's been a war-torn game, and I can;t understand why the AI civs would have skipped this tech.

Especially Russia!!

=DrJambo= · May 2, 2002, 11:58

now you'll find the AI bee-lining for techs that are essential for era advancement only. It'll often not bother with chivalry, printing press-free artistry, military tradition or any other tech not required for era advancement.

In an attempt to slow the AI's pace down a little and encourage the AI to go for useful techs like military tradition i made many of these techs a 'requirement' for era advancement.

Boris Godunov · May 2, 2002, 12:30

Quote:

Originally posted by =DrJambo=
now you'll find the AI bee-lining for techs that are essential for era advancement only. It'll often not bother with chivalry, printing press-free artistry, military tradition or any other tech not required for era advancement.

In an attempt to slow the AI's pace down a little and encourage the AI to go for useful techs like military tradition i made many of these techs a 'requirement' for era advancement.

Do you think it would help or hurt to have ALL techs under an era required for advancement to the next? I've been toying with this idea, and just wanted to ask anyone who had done it hwo it affected the game.

Theseus · May 2, 2002, 15:11

That explains the various comments on the lack of Knights and Cavalry.

Big mistake for the military AI. I might have to go into the editor finally.

Boris Godunov · May 2, 2002, 15:43

In the game I just played, the AI was crawling with cavalry. I don't recall seeing many knights, but at that point I was in a peaceful stage. What I still see constantly is it being modern times and the AI will have modern armors marching alongside scores of bowmen and, especially, Civ's special units (even if those units are waaaaay obsolete).

Lucilla · May 2, 2002, 15:45

IMO at least Democracy should be a mandatory advance.

In my current game I'm about to discover Mass Production and I'm second in the tech race. Nobody discovered Democracy or even Printing Press so far, so I can't buy it from anybody.

If I start researching it myself, I'll fall behind in the advances, so I have to get on without it.

Concernig the other optional techs, I think it's ok to leave them as they are, but I would suggest that the AI should go after Military Tradition for Russia, otherwise the benefit of that special unit would be lost.

Fitz · May 2, 2002, 17:46

Cool, now I can research those techs and use them to extort stuff out of the AI. Or not trade them and keep a lock on those technologies.

Jamski · May 2, 2002, 17:55

Surely this is good, I got Bach's no problem last 5ish games - the AI didn't even have music theory, and, one turn before its ready... sell, sell, SELL ! But don't do this with millitary tradition unless you have ALL the salpeter

-Jam

Lucilla · May 2, 2002, 18:08

Quote:

Originally posted by War of Art
Surely this is good, I got Bach's no problem last 5ish games - the AI didn't even have music theory, and, one turn before its ready... sell, sell, SELL ! But don't do this with millitary tradition unless you have ALL the salpeter

-Jam

The longer I think about it, that strategy seems to be quite useful. I'll see what I can make of it...

Theseus · May 3, 2002, 00:58

Well, if we know the AI's research paths, there are obviously exploits.

I guess I'm talking to Soren... the use of Knight and Cav-level units seems a little off, and it might be due to research prioritization.

BTW, the Russians finally started attacking with Cossacks in my current game, but it was too late, as I had already upgraded to Infantry. Also, probably due to lack of funds, they were showing up one by one as opposed to en masse.

HazieDaVampire · May 3, 2002, 11:38

I've been swarmed by Knights on easy! You should of seen my face!

HazieDaVampire · May 3, 2002, 15:47

Ever since you's have said that i've been denying the AI Shivilry!

Flanker · May 3, 2002, 16:13

Quote:

IMO at least Democracy should be a mandatory advance.

In my current game I'm about to discover Mass Production and I'm second in the tech race. Nobody discovered Democracy or even Printing Press so far, so I can't buy it from anybody.

If I start researching it myself, I'll fall behind in the advances, so I have to get on without it.

I don´t really agree. Because that´s one of the few occations when you actually have to think twice before you decide what to research. This adds a little level to the gameplay. So I think it´s good to have these non-eras-techs, they are usually very good to research, and while the AI doesn´t do that, you could get a lead over him by doing so. Just because the AI is in the industrial age while you´re still in the medieval, this could still mean you have more techs than the AI.

By the way, I´ve noticed that the AI seldom takes the research path to electricity, but rather develops steel and refining. It´s good for me, because then I´ll be the first one to build Hoover Dam. Anyone else noticed this?

Lucilla · May 3, 2002, 17:44

Quote:

Originally posted by Flanker

I don´t really agree. Because that´s one of the few occations when you actually have to think twice before you decide what to research. This adds a little level to the gameplay. So I think it´s good to have these non-eras-techs, they are usually very good to research, and while the AI doesn´t do that, you could get a lead over him by doing so. Just because the AI is in the industrial age while you´re still in the medieval, this could still mean you have more techs than the AI.

By the way, I´ve noticed that the AI seldom takes the research path to electricity, but rather develops steel and refining. It´s good for me, because then I´ll be the first one to build Hoover Dam. Anyone else noticed this?

Well, after the discussion in this thread I actually went for democracy and had then no problems selling printing press and democracy for other techs to the AI. So the strategy actually works out.

And yes, I noticed the preference of the AI for certain techs as well. In the Middle Ages I usually get invention and gunpowder before the AI does, whilst he goes for Education. But that might also be, because I usually own the Great Lib, so that he seeks to neutralise this advantage.
In the Industrial Ages I go for steam power and industrialization first and then go for electronics. Once I built the Hoover Dam, I'm often able to buy refining from the AI.

lockstep · May 3, 2002, 17:57

Quote:

Originally posted by Flanker
By the way, I´ve noticed that the AI seldom takes the research path to electricity, but rather develops steel and refining. It´s good for me, because then I´ll be the first one to build Hoover Dam. Anyone else noticed this?

Yeah, I noticed it too. At monarch, the AI's will sometimes beat me to the Sistine Chapel, but almost never to the Hoover Dam, even if they have entered the Industrial Age several turns before me.

Grrr · May 3, 2002, 18:07

The AI tech needs to be improved a little, but I think it is quite good.

Dominae · May 3, 2002, 18:22

Knowledge of the AIs tech priorities, whatever they are, will always be sort of an "exploit"; the AI acts in a regular/repeatable pattern (here, concerning techs, but in many other things as well), and it's just too easy for the human to adapt.

One of the best ways to avoid this problem is to have different civs (or different civ "personalities) have different research priorities (a la SMAC). This way (and assuming there are always some good techs alon each "path), the AI's research behaviour would be a lot more difficult to exploit.

Changing the values in the editor will only make the AI research techs according to your "tech-estimation algorithm". If you know that the AI always rushes for Military Tradition, it would be easy to avoid it and trade for it once it depreciates fully, spending your beakers on other less-popular techs. You'll also be guaranteed a lot of Cavalry wars every time you play, which strikes as kind of boring (the peaceful Builder civ would no longer exist, if building masses of Cavalry is hard-coded).

Just some thoughts...

Dominae

player1 · May 3, 2002, 18:28

I always go for Monarchy.

Why?

Since it's probably last thing that AI would research.
I usually get all post witiring, Currecny and Construction techs by trade.
For that time I research Polytheism and Monarchy.
And get Hanging Gardens fist.

Grrr · May 3, 2002, 19:13

I agree with Player1. I still think the AI should be more inclined to reasearch useless techs.