A.I. Moves capital - why?

[Ace]

I have played several games (at Diety in MGE) recently, and the A.I. civs never moved their capitals. Then, this morning, I started a new game, and before 1500 BC, 3 of the A.I. civs moved their capitals? In the past, I have seen this happan, but have never heard a reason for this happening. Does anyone have an explaination of why the A.I. does this? And why it will skip several games, then do it several times in one game?

The only reason (trigger) that comes to mind is poor terrain. There must be something in the logic that causes this..but what

 

[starlifter]

My observations are similar to yours. I've seen the AI move the capitol to one city, and move it again a few turns later. When I find the AI, I can see no "reason" for the change. When you wipe the AI's capital out, it usually takes forever for it to make a new one... and seems to prefer courthouses before capital in late game. It sounds like the game algorithm has a threshhold that is sensitive to moving the capital, but I've never examined it closely. Of course, people can see and detect patterns with almost no effort, and a digital computer program must devote enormous resources to comprehend the same data. I'm sure the AI program takes only a crude look, and the triggers are not what we (as humans) would expect or make a decision on. I wish it would not do that, and just use the shields to make a wonder or some units, to make it more realistic.

 

[Ace]

Originally posted by starlifter
.... When you wipe the AI's capital out, it usually takes forever for it to make a new one... and seems to prefer courthouses before capital in late game. It sounds like the game algorithm has a threshhold that is sensitive to moving the capital, but I've never examined it closely....

Except if the AI has a SS, than it instantly builds a new palace if it has the coins( I have seen an AI build 4 successive palaces in one turn, but, unfortunately for it, it ran out of money on the fifth try). It just seems strange, especially going several games in a row without a capital being moved Then in my game this morning, 3 different AIs moved their capitals, all before 1500BC. One doing it could be "logical", but 3 of them....?

Ah well, it only effects the AI's construction, its just curious seeing that annoucement popping up.

:tank:

 

[Rowan]

Yeah,I've seen the AI move its capatil afew times and some times the AI country 'splits"{ie becomes two countrys}.
Has any one else noticed this?
[dance]

 

[Ren]

The AI moves capitals very frequently and in most games with me, which is fine with me! Especially before I find them, I haven't noticed this frequency in mid-late game, though maybe that's because I have more things keeping me occupied by then

 

[funxus]

As Ace mentioned, it only happens in some games, and all other civs change capital. They then switched back to the original capital. Maybe they don't have enough to produce and when there's nothing else to do, they build a palace. In this month's GOTM it happened, 17 I think. Maybe someone else noticed this?
I don't complain though

 

[Kev]

I have a feeling that random capital movement has something to do with not having any other improvements available. In essence, the AI sees it has enough defense in a given city and there are no improvements available with the given technology. When this happens, instead of building another unit or a settler/engineer, it begins to build a palace - viewed in this case as a regular city improvement. My guess is that even if it discovers a tech that would give a new improvement, it continues with what it started.

This may explain quick-moving capitals over a few turns. It could be that two or more cities started palaces (that's possible, right?) and they finish them at different times.

Of course, this is different from capital moving due to military action - capital escaping if you will.

 

[Marlos]

Let's reverse the question momentarily. There is a certain logic to moving your capital late in the game -- to better center (or "cover") the outposts of your civ. Perhaps the AI is moving for a better defensive position from its front lines?

Still, it does seem wasteful, given all of the other things that need to be done.

 

[Ace]

True, but late in the game you should be in a democracy, or fundy, and the capital location has no effect in those two types of government.

Besides, the events I was questioning took place prior to 1500BC, hardly "late game".


:tank:

 

[duke o' york]

Originally posted by Kev
I have a feeling that random capital movement has something to do with not having any other improvements available. In essence, the AI sees it has enough defense in a given city and there are no improvements available with the given technology. When this happens, instead of building another unit or a settler/engineer, it begins to build a palace - viewed in this case as a regular city improvement. My guess is that even if it discovers a tech that would give a new improvement, it continues with what it started.
This may explain quick-moving capitals over a few turns. It could be that two or more cities started palaces (that's possible, right?) and they finish them at different times.

This is also the reason that when the Roman AI builds the Hanging Gardens you get the message "Rome (Roman) has built the Hanging Gardens" but then also "The French have changed wonders from the Hanging Gardens to the Lighthouse" and "The Romans have changed wonders from the Hanging Gardens to the Great Wall". The AI will build whatever is left in the queue, and if this means that they have 4 cities all building the same wonder then so be it. Of course the AI doesn't realise the potential fo continuing to build a wonder after is has been finished elsewhere so this makes for a lot of wasted shields . If I was stuck for things to build then I'd either churn out settlers by rearranging the tiles for lots of food and shields or just make caravans and then pile them out side the proposed wonder city.

 

[starlifter]

This may explain quick-moving capitals over a few turns. It could be that two or more cities started palaces (that's possible, right?) and they finish them at different times

I can tell you this much, from testing it on a limited basis two years ago.... there is a trigger that is determined city by city, on whether or not to build a Palace. When a city determines it should build one, it does not care about all the other cities, many of which each think THEY should build one, too.... and barring an attack on their city, or production change, etc.... when those cities are done, you'll see a ripple of new Palaces ("move capital") messages.

I'm not real clear on precisely what triggers the decision to move a capital, but one thing it does look at is proximity to the Human. It likes to build a capital close to the "threat", but then it is very likely to have another city counter-build one, and eventaully it seems to gravitaty back to the original capital.

 

[Kev]

Again, I submit that it has nothing to do with strategy. I feel that this happens when a city in question has built the improvements allowed by its current technology and there are no wonders available. If it has a nice quota of units, then it likely decides that "it's time to build an improvement" at which time it looks at the list of possibilities. If there is only a palace as a choice, well it starts one of those. Again, not to move its base for any strategic reason, but because it sees the palace as a buildable improvement and just starts it. Of course, it's usually a city that's early in the city list - they're more likely to have all of the other improvements completed. Could explain why it also ends up moving sometimes BACK to the original capital - that city did not have a palace as a choice for a long time, so it may have completed all of the other improvements within that time as well.

I further submit that if a city does start a palace and, for example, discovers seafaring, then in most cases it will continue the palace rather than switch to a harbor. Not sure if that's 100%.

Of course, I've not done any testing on this. I guess it's possible to start a trash game and wait until a moved capital message is given and then look in on the city in question to see if there were other improvements it may have chosen. Of course, new technology may have been discovered from the time it started, so who knows.

When looking for triggers (which must exist as I have trouble thinking the Civ2 AI actually considers capital placement in regard to maximum effect or military vulnerability. If that were the case then they'd actually build cities that INCLUDE specials rather then missing them by one square), I still think it's an internal one like I mentioned rather than a reactionary one to outside influences.

 

[funxus]

Again, I submit that it has nothing to do with strategy

Do you mean that this thread is misplaced in the Civ 2 - Strategy forum? It's not, it's in general discussion.

 

[Old n Slow]

I think that Kev has the right of it. I've often started & restarted the same game & played to about the end of the early Wonder building. In some cases, no capital switching. In others, a number of civs would switch capitals, often twice or more.

The strategy is pretty simple -- build a few units, build improvements. If the city is moderately maxed out in terms of choices to build (such as no additional improvements) and sufficient units so far, then why not start a palace (in the menu list of options) & switch over to something else (which the ai doesn't seem to do).

I think that the critical elements are -- moderately early in the game; limited choices of things to build (i.e. low tech levels); at least two cities; probably not at war. I've also noticed that at the lower levels (prince) the ai starrts wonder building circa three (for some civs) or four cities. It might be that the palace switching becomes less of a phenomena when the civ has 4+ cities & a Wonder that it could build.

Maybe a strategy could be to gift some non infrastructure techs (especially chemistry & physics) to help slow them down -- see what happens.

 

[Ace]

IIRC, the times that the A.I. moved its capitals, I was NOT in contact with any of the AI civs...could that be related to the trigger mechanism????

 

[MajorGeneral2]

Certainly the AI can focus more on domestic issues if it has met no one else. However, I think that the AI has just run out of things to do. You know, enough unmanned fortresses, aimlessly wandering ships, half-empty transports, lone catapults ...

 

[Kev]

Originally posted by Ace
IIRC, the times that the A.I. moved its capitals, I was NOT in contact with any of the AI civs...could that be related to the trigger mechanism????

In this case, I think it bolsters my own point. AI's that are relatively isolated are probably more likely to be capital movers - the reason being that they cannot trade as much for techs and are likely to "run out" of improvements to build.

funxus: I'm sorry, when I wrote that it has nothing to do with strategy, I was referring to the AI not switching capitals as part of strategy. I did not mean to say that this discussion should be elsewhere or anything.

 

[funxus]

Kev, I'm the one who should be sorry I misunderstood that line.

Does the AI try to keep up with the human research? If I'm slow on research, they're slow, and if I'm advancing fast they are a little faster? Or does this just have to do with the key-civ? If so, that might be a sign that you have bad research.