(CONFIRMED) City Governor works suboptimal tiles

I just noticed this in the city of Leptis:



The governor has chosen to work the two lumbermilled tundra forests for a total yield of 2F/4H/0C. If those two citizens had instead been assigned to work the iron mine and one of the coast tiles the total yield would have been 2F/5H/3C.

I understand that there are times when the governor wants to work low food tiles to slow growth, but it's never advantageous to work tiles that produce less hammers or commerce.

There was no emphasis selected in this case.

Save game:

http://forums.civfanatics.com/uploads/105859/BadGovernor01.CivBeyondSwordSave

I have experienced this as well...it seems to happen whenever new population growth would result in unhappiness.

Definitely a change from Warlords, as the cities would otherwise work toward growing population, regardless of the happiness outcome (we would have to adjust manually ourselves). Personally, I preferred the old way better, as it's easier to manage unhappy citizens than it is for the city to simply become stagnant and work subpar tiles.

But in this case the governor's choice doesn't effect growth rate. It's 2F in both cases. Just less hammers and commerce using the governor's choice.

ah, I see what you're saying; sorry I misunderstood you. Yes, that's kind of pointless, isn't it?

I'll have to pay attention in my games and see if that happens regularly or if this is just an odd fluke. Either way, we're going to be forced to keep closer tabs of our cities with BtS.

Yes, it happened to me when a city's growth would cause unhealthiness. I settled a city on a massive flood plain, and to grow to population 2 would cause 1 unhealthiness. To avoid this (apparently), the governer chose to work a desert/mountain tile yielding only 1H/1C instead of working a flood plain 3F/1C.

I have run into this error, too, even with an emphasis selected. It may have to do with an AI algorithm that selects the next best tile sequentially rather than considering the best combination holistically. The reason the AI would be written that way is that it is O(n) versus O(n!) to compute and normally results in pretty good optimizations.

One way to improve the algorithm would be to try an alternate assignment for tiles that get left out of the assignment even though they could produce more of a resource than the tiles that were assigned. The code would be a little messy, but it would optimize the solution.

The way it would work is after making assignments, the three unassigned tiles with highest food, hammer, and commerce production would be considered. If any of those tiles were no better than what is already assigned (e.g. the next tile only produced as much as the last tile assigned) then they could be ignored. But in the case of an unassigned iron mine, gold mine, fishery, etc., the AI would work out a possible improvement using this tile.

That would solve the situation in the OP and would not make the algorithm much more time-consuming. It would require writing a routine to reshuffle tile assignments, but that is not a large problem.

I've also noticed this, and I agree it's linked to the / & / caps ... the AI appears to force "stagnate growth" when growth will result in unhappy citizens or unhealthiness.

I've noticed a similar City Governor bug, which I just posted here:

City Governor force-works new improvements .