Two Ideas

[Duncan Idaho]

Originally posted by Xin Yu
Now seems there is a filter for size 4, so the guess would be that the capital will move to the oldest or closest or most culture improvement or most accrued culture city with size >=4. Have to test it to figure out.

I am guessing , only after two experiments on rather large end-game situations, that the following applies. Cities are sorted on the following, in order, and the top one is chosen:
1. City sphere of influence
2. City population
3. ?
...
n. Date founded
Frozen palace city is excluded.

The sphere of influence thing might explain the odd results DrSpike got with the size 3/5/6 cities, since at that phase of the game the cities are expanding those by large strides.

Then again, take heed that this was sheer speculation on an impoverished sample of limited size

Maybe we could start by compiling a list of candidate factors for the selection of palace relocation.

[DrSpike]

well I'll chip in another couple of observations.

1) I'm not sure the sphere of influence thing explains what happened to me; both cities had similar culture ratings, and culture currently produced per turn. Also, why would reducing the pop through rushing settlers change the sphere of influence?

2) I had another size 5 city that was completely ignored (but not if I rushed the target city down to 5 so they were equal). So two size 5 cities had to be reduced to 3 to prevent the capital going to them, despite the fact that the target city had a pop of 6. The other size 5 city was fine, and was still 5 when I successfully bounced the palace.

Food for thought.

[Duncan Idaho]

Originally posted by DrSpike
1) I'm not sure the sphere of influence thing explains what happened to me; both cities had similar culture ratings, and culture currently produced per turn. Also, why would reducing the pop through rushing settlers change the sphere of influence?

It doesn't. Sphere of influence is 1, then grows to 2 when town has culture>=10, then grows to 3 when culture>=100, and so on. Maybe the target city had 97 culture and a few turns later it had 101, which increased its cultural radius (=sphere of influence). Just a shot in the dark.

Originally posted by DrSpike
2) I had another size 5 city that was completely ignored (but not if I rushed the target city down to 5 so they were equal). So two size 5 cities had to be reduced to 3 to prevent the capital going to them, despite the fact that the target city had a pop of 6. The other size 5 city was fine, and was still 5 when I successfully bounced the palace.

The stuff I wrote about sorting was meant to indicate that sphere of influence would supersede the city size; if sphere of influence is bigger, size doesn't matter.

[DrSpike]

I see. An interesting theory, but the ignored size 5 city had the most culture and highest sphere of influence of all the cities outside of the original capital. Plus there was no change in the sphere of influence of the target city during the tests.

[Xin Yu]

Maybe the AI is smart enough to pick a large city in the center of your civ? Was the size 5 city at the edge of your civ?

[DrSpike]

I had two groups of cities separated by inhospitable terrain. And yes, the two cities it was trying to bounce to were in the middlish. The 5 that was ignored was slightly less central (but was a more developed city), and the one I was trying to bounce to was in the other group. It's clearly not just pop; maybe pop and placement explains what happened, but I'm not sure how the AI would decide what a central city is.

[Aeson]

Maybe it has something to do with production/commerce? The highest population city wouldn't always be the most productive. I've been able to move my palace every time, but always to a size 3 city while all the others were size 2 or smaller.

[DrSpike]

There are some good suggestions here, and when I get time I'll pore over the savegames and see if any of the theories fit the data.

[Marquis de Sodaq]

Originally posted by Xin Yu
Maybe the AI is smart enough to pick a large city in the center of your civ? Was the size 5 city at the edge of your civ?

I've noticed, watching the AI wage war, that it tends to march on the capitol city as the primary goal. If other cities are in the way, it sacks them, but the target is the capitol. This could be used with your strategy.

[Blaupanzer]

Build Forbidden Palace, near to the original capitol but more central. Then, make sure the target city for the Palace is at least two more in population than any other city. Then disband your capitol (which you have been using as a settler/worker farm. If in the ancient period, this works like a charm. I've done it recently (post 1.17f) in three games where the capitol was at one end of a continent or on a coast.

In the Middle Ages, the jump cannot be guaranteed solely by population. Then, using a leader makes more sense, if no wonders are available. (The Middle Ages start with several Wonder-building opportunities, then they peter out.)

For yucks, I tried to disband a city with two Wonders in it that my opponent was about to retake. I got the "too dangerous response with no option to proceed. This makes me suspect that cities with Wonders cannot be disbanded, although an example of one is obviously anecdotal.

[DrSpike]

I too have used this strat a lot. I agree the effect of disbanding the capital becomes harder to predict the later in the game it is - there are clearly other factors involved that aren't important for the early game.

[Xin Yu]

Could be trade. Some cities have road connecting to other civ's capitals, thus are favourable.