AI Beelines

[mark13]

Double post. What am I doing wrong here?

[This message has been edited by mark13 (edited September 14, 2000).]

[mark13]

SII,

An interesting idea, and one which I have tested in 2 seperate games. In the first, secrets to the Needlejet chassis lay in Centauri Meditation (yeah, right )
and my results were as follows:

- In the first game, Yang did not TOUCH Doctrine: Air Power until at least 50 years after I had it (I went to it as I normally would). Instead, it went down toward Ind. Auto, before striding down to Superstring Theory, not going anywhere near Centauri Meditation. It then acquired Centauri Empathy (from another faction), and then went straight to Centauri Meditation, immediately BEFORE Doctrine: Air Power.

- In the second game, the Foil chassis could be acquired upon the discovery of Centauri Empathy (another logical transfer). Yang was trapped on a little island (big enough for about 5 bases at most).
The research pattern went as follows:

1) Applied Physics (obvious military applications)
2) Social Psych (prereq. for CE, obvious build apps)
3) Centauri Ecology (immediate prereq. for CE)
4) Information Networks (prereq. for Nonlinear Maths, BIG conquer tech)
5) Biogenetics (prereq. for CE)
6) Secrets of the Human Brain (build/conquer value 0 BUT immediate prereq. to CE)
7) Centauri Empathy

Draw what conclusion you will from that, but it gives strong evidence on the part of AI beelining. But, from the first game, there is clearly a limit as to how far the AI will go in order to get a key technology, or a key chassis, in this case.

Regarding the faction files, there is absolutely nothing in the faction files themselves, and there doesn't seem to be anything coded.

Also, any critics who want to criticise SMAC for being a "copy" of Civ II, should read each of the above posts. The fact is that the SMAC AI is VASTLY superior to that of Civ II (read "Dumbest AI contest" thread in the Civ II Strategy forum).

Once again, well done Firaxis!



[This message has been edited by mark13 (edited September 14, 2000).]

[Adam_Smith]

quote: Originally posted by Adam_Smith on 09-05-2000 01:22 PM This thread is about particular beelines that each AI makes... you might want to know what techs you might steal ahead of time. Also, if you are playing against a probe-low tech team, and you know they plan to steal tech, then you might want to know what tech they will, or will not be stealing. Anymore you guys can think of?

I've already experimented plenty. It's called playing the game. mark13 that's not a criticism on what you are doing. I think that's great. What it is is a statement that I already know from experience that AI's beeline towards particular techs in general. If you don't know, for example, that Yang has a tendency to beeline to D: AP then you simply lack experience. This thread really was designed for experienced players to discuss different beelines that alternative AIs tend to research. Unfortunately, no such players have contributed. I guess this place is really dead.

[mark13]

quote: Originally posted by Adam_Smith on 09-14-2000 10:35 PM I already know from experience that AI's beeline towards particular techs in general. If you don't know, for example, that Yang has a tendency to beeline to D: AP then you simply lack experience.

That is the main problem, you have already outlined most of the beelines themselves. All I really tried to do was expand on these theories, to establish why the AI does what it does, in order to better predict the AIs actions in the future. I thought that maybe this would give more accurate an explanation as to why the AI chooses these beelines.

Secondly, I apologise now for any diversion from the subject in hand on my part. The point is, AS, that you had already uotlined what the AI goes to, and I just tried to establish why.

[Fitz]

Mark13, don't worry about Adam_Smith, he seems to have a marked aversion to experiments for some reason. I applaud your efforts to find out the why of AI beelines. If you happen to come up with any other reasoning, please let us know.

And A_S, I would be interested to know if you have discovered anyother beelines past D:A. I tend to play games much furthur than that, and if you've noticed the AI beelining for a particular level 8 or 9 tech, I would very much like to know.

[mark13]

Ooooooh......
A tough challenge you set, Fitz, given my limited experience in this part of the game. However, there may be limited findings in this part of the game, as the tech tree severely narrows at about Level 7. However, the basic beelines are outlined thus:

Morgan - Industrial Nanorobotics - it has a build value of EIGHT
Yang - Unified Field Theory
Santiago - also Unified Field Theory
Deirdre - all the "Centauri" techs (Psi, Genetics etc.)
Miriam - when she gets that far, similar to Yang.
Lal - not sure, tends to go down Explore and Discover fields, I'd guess at Secrets of Alpha Centauri.
Zakharov - tends to go for reactor upgrades (e.g. Fusion Power, Quantum Power, Singularity Mechanics)

Again I would stress that the tech tree narrows severely from about Level 7 on, and as I concluded earlier, some tech advances depend partly on circumstance.

[mark13]

SII,

I hadn't thought of that factor, that maybe the slots themselves might have an effect. On further analysis:

- University on massive (256x256 planet) on scenario editor, in the slot of the Hive. Technological advances were as follows:

Ind. Base (from pod), Planetary Networks, Polymorphic Software, Biogenetics, Applied Physics, Centauri Ecology, Social Psych, Optical Computers, Doc. Flex (on 50-70% planet), Secrets of Human Brain, Doc. Loy. (as bonus from Sec. Brain), Ethical Calculus, Intellectual Integrity, Cyberethics, Advanced Subat. Theory.......

I didn't go any further, beacuse these research traits do not, in any way, represent those of the Hive. Steering well clear of D:AP, it looks to be heading straight for Fusion Power.

A second game monitoring Hive tech progress in the slot of the University yielded similar results (following usual Hive threads):

App. Phy, Inf. Net, Social Psych, Nonlinear Maths etc.

[Simpson II]

Adam:

Of course all experienced players know the AI bee-lines. We're trying to work out the rules so that poor saps like you can work it out, rather than coming around asking everyone what they are. (just kidding.. )

mark13,

I know there is nothing in the faction files to change the AI behaviour other than the build, explore etc. flags, but Tim Train indicated that they programmed faction-specific behaviours in addition to these. If so, the only real way to do that would be to associate the behaviour with a particular faction 'slot' in alpha.txt. So perhaps swapping e.g. the HIVE, HIVE line with the UNIV, UNIV one would produce different behaviours?

I'm afraid I've very little idea about late-game beelines. Lal likes pre-sentient algorithms, and someone like advanced spacelight, but by this time all of the AIs are usually pacted and swapping techs like mad, so I don't know who. Also I usually transcend at about that point, so I've no idea what the AI does later on.

[Adam_Smith]

Wow! Cool.

Thanx for contributing guys.

I'm not sure about what techs the AIs shoot for past level 6. The reason is that I never paid that much attention. By that time the game is usually decided and I start another one or I'm not dependent on what the AI is researching. There are some exceptions to that though. Sometimes some of Aki Zeta's and Zak's research is usefull. I think I remember them going for discover techs. I've never really been in a research race down the line with the AI (that would be fun). I take what I need from them and then I never need them again, because I pull ahead.

I think that the AIs (in general) try for Orbital Spaceflight and Advanced Space flight before other techs of similar level. I'm thinking in particular to the two PBEM games that I'm playing that have advanced to that level. In one of them Datajack has been researching Orbital Spaceflight for some time (probably I will end up giving it to her). I will let everyone know if she goes one to Organic Superlube and then to Advanced Spaceflight after that.

The other game I'm in is with Yang AI. He's actually winning, because my opponent and I are fighting to the death. I just built a nuke for him, and should be able to nuke a size 16 base. On the otherhand there is no real chance for me to win. I'm only able to stay alive as Yang's ally. Anyway, Yang has researched Orbital Spaceflight. I stoled Organic Superlube from Pirates, and traded that to Yang for Orbital Spaceflight. Now Yang is trying for Advanced Spaceflight. He doesn't have Unified Field Theory or even Mogpole Magnets yet.

Does anyone know what techs the following factions tend to strive for:

Believers
Peacekeepers
Either of the Aliens
The Planet Cult
The Angels
The Drones
The Pirates (I think they go similar to the Spartans)

[Basil]

In single-player, both aliens regularly get D:AP before I do. Considering as I usually have several more techs by this time than they do (I play blind research), I suspect that they're beelining.

[Qualicide]

its obvious the cult goes similar to the gaians, the caretakers might too because of the +1 planet etc

the usurpers probably go up to D:AP much like yang, as the 2 are very similar

thats about all i can think of, roze is probably like lal, in other words, we have no idea

------------------
http://thangorodrim.angband.org/angband.html (alright meybe i didnt make it)

[Qualicide]

oops

------------------
http://thangorodrim.angband.org/angband.html (alright meybe i didnt make it)
[This message has been edited by Qualicide (edited September 26, 2000).]

[DilithiumDad]

AI research choice is even more sophisticated than that --it also takes into account SP's and which SP's have been built.

Pirates beeline for Adaptive Doctrine and Cult beelines for Centauri Genetics because of the free abilities and facilities (same with Roze and Presentient).

------------------
Creator of the Ultimate Builder Map, based on the Huge Map of Planet