AI city management still breaking down in V38,5

[Tajo]

Depending on empire size (larger maps allows larger empires) global unhappiness and global unhealthiness may go out of control.
There is no way to reduce pollution with units (source of global unhealthiness)
As for crime criminals may spawn faster than police can catch them (source of global unhappiness)

Disease is effectively obsoleted with some techs in Nanotech or Transhuman era.

Can you upload save?

Yeah but after having carefully sold the crime inducing buildings and putting few police units in the city, the crime still shot up by +503 and education went down -11. Just atrocious, abandoning city and building a new would be much more reasonable but there are nice wonders in there so this is damn annoying. Map isn't even a big one.

Save attached.

There is no way to reduce pollution with units (source of global unhealthiness)

Sure there are, park rangers and such.

As for crime criminals may spawn faster than police can catch them (source of global unhappiness)

Yeah but the AI must be able to respond to this. By the way, there were no criminals in the city and still isn't.

 

[raxo2222]

Hmm you are in Modern era.

Maybe AIs are producing criminals?

I see a lot of bandit criminals - it looks like AI has hard time arresting them.

 

[Tajo]

Hmm you are in Modern era.

Maybe AIs are producing criminals?

I see a lot of bandit criminals - it looks like AI has hard time arresting them.

There wasn't and isn't any in city that just flipped over to me. Instead the crime and especially the green face problem were out of control as crime shot up over +500 in a single turn despite heavy combing of crime inducing buildings. A turn or so later another city flipped from the same civ and it was totally fine after adding few police units but the map tells that majority of the German cities have suffered quite a while from excessive red and green faces.

So yeah, much work still there but starving the city smaller seems to help nicely in crime problem. There is also something else strange that I noticed in Germany AI. It seems to have gotten stuck somehow into war mentality as it says "-1 the war is going badly for us" even when the Germans haven't been in war for ages.

 

[Tajo]

Any progress on this game breaker?

 

[Yudishtira]

Does AI have lithium?

Indirectly its needed to make police mechs.

By the way how does AI fares with resource/building supply chains?
They get longer in later eras.

Two good questions here. Any answers?

The Mech Assembly Plant AND the Police Precinct both have very convoluted prereqs. Could it be that the AI knows to build the Mech Assembly Plant - when it can - but not (for example) the Robotics Lab, without which the Mech Assembly Plant will never be an option?

 

[raxo2222]

Two good questions here. Any answers?

The Mech Assembly Plant AND the Police Precinct both have very convoluted prereqs. Could it be that the AI knows to build the assembly plant - when it can - but not (for example) the Robotics Lab, without which the mech assembly plant will never be an option?

Yeah, sometimes AI doesn't want to build building prereq.
There should be some sort of incentive for AI, if given building is prereq for something.

 

[Dancing Hoskuld]

Yeah, sometimes AI doesn't want to build building prereq.
There should be some sort of incentive for AI, if given building is prereq for something.

If you made all of the buildings in the chain require each of the precursors the AI will be more likely to build them. eg if A leads to B leads to C and leads to D have D require A and B and C; C require A and B and B require A. The AI will now see A as much more important than if you had D requires C; C requires B and B requires A.

 

[raxo2222]

If you made all of the buildings in the chain require each of the precursors the AI will be more likely to build them. eg if A leads to B leads to C and leads to D have D require A and B and C; C require A and B and B require A. The AI will now see A as much more important than if you had D requires C; C requires B and B requires A.

Assembler buildings (Information era and later productivity boosters) would make some buildings EXTREMELY valuable - they have extensive demand tree.
Highest tier one requires lower tiers.
What if some building requires A OR B, A has 2 prereqs - A1, A2 and B has two other prereqs - A1, B1, and one of those prereqs is common?

@pepper2000 or @Toffer90
Is it possible to generate such tree of demand for buildings?

Laboratory, Factory, University and such other buildings would get promptly built anywhere
I think longest demand trees start at Library in Classical era - it is needed for University, that is needed by Laboratory, that is needed for some hi tech laboratories that is needed by assemblers, that are needed by Materializer - one of buildings at end of tech tree.

Generally early buildings, that never obsolete would see highest increase of value by AI, as those are base of extensive building chains.
There are resource requirements too.
If factory is worthless and resource it produces is needed by building, which AI sees it as needed for 10 other buildings, then AI would build such factory?

NASA doesn't have complicated prereqs - just rocket fuel (and by proxy its factory or traded) and aviation base in three cities.
Despite that it never gets built by AI.

 

[KaTiON_PT]

@raxo2222 would it work if the NASA project is removed and the earlier space units be made available with technologies?

 

[raxo2222]

@raxo2222 would it work if the NASA project is removed and the earlier space units be made available with technologies?

Well Pepper did that intentionally: it makes Launch Effect, which in turn you can build space units anywhere, that is ones buildable on Earth.
Space units are already unlocked with tech too.

 

[Dancing Hoskuld]

The problem is that the AI only looks at what is best this turn. This means that any building that opens many buildings or units will be considered a better choice than one that just opens one. The AI could be improved to make it look further to see what the buildings that open allow but that would make turn times slower. Humans on the other hand can see that they want D and therefore need to build A, B and C as soon as possible.

I have suggested better ways of doing this. For example in the real world you only need one tyre factory in your nation, except perhaps in the time of war, but we allow them in every city and expect the AI to know that it is bad to build more than one.

In my view the Factory building needs to go. We have split it into multitudes of specialist factories so why do we have it? Also the specialist factories need to be limited in some way. Maybe only allow one or two per nation. However that needs to be done in a way that does not increase the steam roller effect of large verses small nations. Similarly for the Manufacturing plants.

 

[tmv]

We have split it into multitudes of specialist factories so why do we have it?

The production is not split, I think. The factory is one of the most important producing buildings, especially after a few important buildings in that regard (like the Tannery) are obsoleted. And with the exponential curve you need every you can get.

 

[raxo2222]

The problem is that the AI only looks at what is best this turn. This means that any building that opens many buildings or units will be considered a better choice than one that just opens one. The AI could be improved to make it look further to see what the buildings that open allow but that would make turn times slower. Humans on the other hand can see that they want D and therefore need to build A, B and C as soon as possible.

I have suggested better ways of doing this. For example in the real world you only need one tyre factory in your nation, except perhaps in the time of war, but we allow them in every city and expect the AI to know that it is bad to build more than one.

In my view the Factory building needs to go. We have split it into multitudes of specialist factories so why do we have it? Also the specialist factories need to be limited in some way. Maybe only allow one or two per nation. However that needs to be done in a way that does not increase the steam roller effect of large verses small nations. Similarly for the Manufacturing plants.

Those megafactories, that produce multiple resources?
They are supposed to replace factories.

The production is not split, I think. The factory is one of the most important producing buildings, especially after a few important buildings in that regard (like the Tannery) are obsoleted. And with the exponential curve you need every you can get.

Hes talking about resource producing factories, one for each resource.

 

[Dancing Hoskuld]

And with the exponential curve you need every you can get.

Which is of course the other problem. Buildings from from start have had OP production so that the AI will choose to build them. This lead to needing to increase the costs of everything to balance it out which lead to the ever escalating costs and returns.

When the AI was improved we did not go back and reduce all the costs and returns on the buildings "because they were balanced" but they could have been reballanced at a lower level without detriment to the AI and we may have avoided the integer overflow problem and exponential curve.

 

[Thunderbrd]

Which is of course the other problem. Buildings from from start have had OP production so that the AI will choose to build them. This lead to needing to increase the costs of everything to balance it out which lead to the ever escalating costs and returns.

When the AI was improved we did not go back and reduce all the costs and returns on the buildings "because they were balanced" but they could have been reballanced at a lower level without detriment to the AI and we may have avoided the integer overflow problem and exponential curve.

We haven't been seeing overflows in production. The exponential curve is normal to a game of improving functions through technology over time. Local values don't tend to overflow - it's national level stuff that needs to be protected against that and production is never tallied nationally unless it's for the charts (which I think happens in python and thus avoids any overflow issues). Yield producing sources have not yet been updated very well... still so many buildings even into mid game eras are only giving +1 when they should be giving much more than that because it should improve as time goes on. This is why I wanted to make buildings morph and advance at tech achievement points. It's actually a very important thing to do if we're going to solve many of these issues - but I don't think we are tbh.