AI Wonder Building

[Blaupanzer]

Ever wonder why the AI has several cities building the same wonder at the same time? (You can see this by checking intelligence thru you foreign advisor and clicking on cities.) I'ts because they add the total production together in order to build the wonder. For example, if Bombay has 350 shields to a 600 shield wonder, Punjab has 150, and Delhi has 50, the wonder will be completed in Bombay. This seeming cheat is apparently an offset to the player's caravan strategy. The Civ II team must have developed this strategy to overcome all the Civ I complaints about the AIs being able to incrementally buy wonders in that game. Means that the player can stop that wonder in Bombay by taking Punjab on the turn after you are warned that 'The Indians are about to complete the Manhatten Project" or whatever.

[cpemma]

If it builds SS parts the same way it would explain how it's building it so fast.

edit - just seen Smash's reply below - less shields needed for AI SS

[Winston]

Yes, the AI uses cumulative building for Wonders to keep up with the human player. Another good reason why they're called Wonders I gather.

On the other hand, the AI doesn't seem to make use of rushbuilding of Wonders - plus they haven't got the advantage we humans have of being warned when one is about to be completed.

Pretty fair deal overall I think.

[debeest]

Not that I have any evidence or reason to believe one way or the other -- does anyone have any actual EVIDENCE for the oft-made claim that the AI pools its shield production across cities? Or for the similar claim that it simply cheats itself free shields? I know that it generally needs fewer shields to build things, but that's a consistent, predictable rule-bound "cheat." How has anyone ever demonstrated that it cheats on shields in other ways?

[Smash]

yes.If you empty an ai city thru attacks.They are able to "rush" a unit in that city.
Temples, I know for sure appear in cities without being built.Had it in a mp game recently where I lost connection and the ai ran my civ for 1 turn.Fortunately my wonder was not abandoned but there was a temple in the city.I did not build it.2 cities had half built barracks.Exactly half.20 sheilds.Even with the smaller sheild boxes of emperor level,this was not possible as the cities only produced 3 and 4 sheilds at that point.

[debeest]

Again, not that I have reason to think you're wrong, but I'm trying to keep an open mind. If someone wipes out YOUR last unit in a city, YOU can rush-build a unit there, can't you? (Although, it won't appear until the other civs have had their turns; but if that's happening, could it be an issue of WHEN the unit appears, rather than how the cost is paid? That would be fully consistent with the AI's magical ability to produce units on the same turn when it learns the tech for them.)

Also, 20 shields in a box could be produced by (1) disbanding a 40-shield unit there, or (2) rush-buying a 20-shield unit and then switching to barracks-building (though I can't see why the AI would need to do that). Neither seems like a good explanation, but they do seem like possible non-cheat explanations. Moreover, assuming that the AI does cheat that way, why would it HALF-build something?

The temple magically appearing in a city where you had been building a Wonder and you still have the appropriate shields toward that Wonder -- now that one is odd. On the other hand, it was a glitch situation all around. I guess I'd have to consider this to be evidence, but not fully convincing.

[shade]

maybe this one will help convince you:
an egyptian city(production of 4 shields)starts building a wonder(one of the later ones=> 600 shields)1 turn later 1/3 of the box is filled(and the Egyptians have no caravans at that time,nor enough money)2turns later=> 2/3 is finished
(I rushbuild the wonder phew)
=> not really explainable with 4shield production,smaller box or even disbanding units(because I am sure there weren't enough of them around and the city was far away from the rest of their empire)

Shade

[Blaupanzer]

The example I gave was real. Initially, I was unable to stop the wonder, MP, which appeared in Punjab. On reload for test only, taking Punjab delayed the wonder by several turns.

[debeest]

The events you guys describe are certainly very suspicious. Your belief about AI shield-cheating is widely held, and I wouldn't be the least bit surprised if you're entirely right. I hope I'm not giving the impression that I think you're wrong; I don't. It's just that I've been reading this forum long enough to have seen quite a few definitive, empirical, experimental statements that later turned out to be less than perfectly accurate for all occasions. It's a very complicated game, and sometimes there are factors that we have trouble accounting for. In my own limited investigations into AI cities, I haven't seen anything that revealed this kind of cheating; once it happens to me personally, I'm sure I'll be much more willing to accept everyone else's anecdotal evidence....

[Marquis de Sodaq]

I, too, have seen odd shield accumulation by the AI. I'd send in a spy to snoop around, then destroy whatever was being produced --> 0 shields remain. The very next turn, there are far more shields than the city produces. At first I just thought units were sneaking in and disbanding, but once more spies arrived on the scene, it was clear that no units were entering.

[debeest]

I'll have to start spending more time spying on enemy cities rather than just co-opting them....

[Marquis de Sodaq]

spies

Snooping and destroying improvements or production is a good way to weaken an AI economy. I usually do it when I don't expect to be able to keep the city if I capture it. For instance, my reinforcements are still 12 turns away and the city in question is in a nest of AI cities. Or if they have enough gold to bribe it back once I grab it.

One way the AI might do this is - at least when starting over from zero shields accumulated - is the rush buy an improvement, then switch production to a unit. Why, I don't know, but with the AI it's never easy to understand...