AI cheating on size of cities just before they are conquered?

[buiz]

You all know the situation. In the initial phases of the game, you have just built up a sufficient army to start making life miserable for your closest neighbours. You have the necessary 5 horsemen stacked 2 tiles from your neighbour's city, but you wait until the city is size 2 (if it is located a strategically important place), so that you actually take over the city instead of destroying it which is what would happen with a size 1 city.

But here comes the strange part: I have now tried a number of times that I patiently wait for the city to become size 2, then it becomes my turn and I move in with my horsemen and THEN the city changes to a size 1 (before my horseman actually attack) -- during my turn!

Some would argue that it is because I am too long away from the city to see/register its developments, but this can't be it because I am just two tiles away (just outside his borders).

Is the AI really cheating in the sense that they can change city size during my turn in order to avoid that I take over a city in an important place? Have anybody else experienced this or is it just me who should be laying off the caffeine ?

[trickey]

I have never experianced this. I muts admit, i have never really payed attention... ill havr to check it out.

[Catt]

There's a long thread on this over at CFC (search for threads by poster "cracker"). I don't know if this has been explained or not, but it's been debated for a couple of weeks (it seems) now.

Catt

[Barchan]

I think that, unlike CivII where the size of the city determined if it was destroyed or not, in CivIII it's whether or not a city has produced any culture points that determines if it's captured or destroyed. In CivIII, even a size two city is destroyed if it hasn't built up any culture points. I've seen it happen many times....

[buiz]

Interesting, glad to see that I am not the only one to have noticed this. Haven't tried destroying a city of size 2 yet, though, but did actually take a couple of cities today at size 1 without destroying them.

This suggests that there is indeed more to it than just size, but I do not necessarily believe it is culture since the borders of all these cities are mostly just one tile (meaning 10 culture beakers not yet produced, meaning probably no temple/library has yet been built).

Can anybody help with the scientific explanation on why some size-1 cities are not destroyed?

And secondly, can anybody explain why the AI seems to be able to change its city size during YOUR turn, so as to avoid that the city is conquered? By definition it must be a bug?

[Theseus]

I don't think I've ever seen this... you sure it's not just AI poprushing?

[Aeson]

If a city's cultural borders have expanded, it won't be destroyed even if it's size 1. Try to follow the AI's capitol around, letting each new one expand it's borders before taking it. That way you can keep each city.

The "scared to death" bug is really interesting, you can investigate the city beforehand and confirm it's size, or even look at the saves. The population loss is during the player's turn.

[JohnM2433]

Unless you actually see the city size change during your turn and while you are right next to it (so it is not shrouded by the fog of war), I would hesitate to say that this is a bug. The city could have built a worker, or it could have experienced starvation.

I think I've seen size 1 cities both destroyed and not destroyed when captured. Question: does a citiy lose a population point only when one of its defenders is defeated in an attack (or as a result of bombardment), or also if it is captured because it is undefended? If only a defending city loses population, then maybe a size one city loses its last population unit and is destroyed if it is being defended, but it is OK if it is undefended, since there is no battle and no civilians need die as a consequence. I don't really keep track of these things, so I'm not sure.

Cities really shouldn't be destroyed unless they are razed or abandoned. Even if the population gets killed off, why should all the buildings be destroyed too? Maybe they should remain as unproductive "ghost towns" until a settler or a worker is moved in.

[buiz]

Aeson, I read through the "scared to death" thread and it seemed to conclude that it is NOT a bug as you say, but rather a result of the actual turn sequence in civ3:

quote
It is NOT as we all expect:

human science turn, human production turn, human move turn
AI1 science turn, AI1 production turn, AI1 move turn,
AI2 science turn, AI2 production turn. AI2 move turn....

but
human science turn, AI1 science turn, AI2 science turn
human production turn, AI1 production turn, AI2 production turn,
human move turn, AI1 move turn...
unquote

This means that if the AI is poprushing a unit in his city, the effect is seen AFTER the human production turn, which is by the player perceived as during "the human turn".

In such case we are looking at flawed fundamentals in civ3 (don't care what anybody says, the AI should not be able to make changes during human's turn), and not a bug?

Anyone asked the question to the Firaxis guys?

[buiz]

Wait a minute, something does not add up here. The frame-by-frame example in the thread clearly shows that it is only AFTER the human's first attack that the city changes to size 1.

The human attacks once, the city remains at 2. Just before the second attack, the city changes to 1.

This means that the change happens during the human's MOVE turn, which doesn't correspond to the turn sequence theory.

Guess we are back to the bug theory, then? Has anyone got any feedback from Firaxis on this one?

[dawidge]

Related to this was a behavior I witnessed in 1.16. I'm not sure if it is still in the game, but the AI seems to be able to draft *during* the human move turn. Since it was in 1.16, I had examined Egypt's defenses with the "multi" cheat (which confirmed what I had checked out with individual city espionage). I began an attack and noticed that instead of two veteran and one regular Infantry, they also had 2 conscripts. I sometimes wich *I* had the opportunity to draft between a declaration of war and actual attack.

[steven8r]

I cannot vouch for the City Size event, but if the pop decrease happens between attacks--During the Human Move turn--isn't that just the effects of the attack--lowering the population?

As far as Size 1 cities being destroyed or not, I think it has to do w/ any improvements, not just cultural ones. If a size 1 city has no improvements, i.e. no buildings, then it is destroyed. If the city has at least 1 improvement (and it doesn't get destroyed in the attack) then you can capture it.

During the early stages of the game, it appears as if the AI goes more for expantion than for improvement. So it could be that the small cities haven't produced anything except Settlers and/or Units.

[buiz]

Alright, so we established that the Scared to Deatch... thing.. exists. Over at CFC they seem to agree that it is a bug, but is it really so, or do we need to look again at the turn order concept?

[Zero-Tau]

This bug got me thinking. If I've understood everything right, then during the human turn, the AI somehow uses the "Rush" feature and gets a spearman completed instantly. And since this is an early war, the AI is under Despotism, so rushing causes the city to shrink, which is what we see as the "Scared to Death" bug.

Now, rushing causes population loss under Despotism or Communism, but causes money loss under Monarchy, Republic or Democracy. So what I was thinking was: isn't it possible that the human declares war on a civ under Monarchy, Republic or Democracy, and then the AI rushes out of turn, causing extra defenders to appear in poorly defended cities, but this effect is not visible to the human, since we can't see the AI's treasury shrink.

No wonder the AI always seems prepared to defend every city.

Now, about the origin of the bug: I think that to speed up the game, Firaxis made it possible for the AI to think during the human turn. So when one spearman is killed, the AI decides that the city isn't defended well enough and decides to rush a spearman. Due to some screwed-up coding, the AI is then able to rush during the human turn. I think that either this is the explanation, or this is a built-in AI cheat. (This is pure guesswork, of course, but it seems like a reasonable explanation.)

Zero-Tau