Report Questionable Behavior

[Wodan]

Re: the 10% not being worth it

On that logic, why build Libraries? It's only 25%. If it takes on average 6 turns to research a technology (very dependent on game speed), then it will take you 24 turns to get 1 technology ahead of your opponents. And at that moment they will start benefitting from the research bonus that civlizations get for technologies that have already been developed by other civilizations.

And, it's even worse than that. You have to spend production capital to make the Library in the first place. Plus, you have to build them in every city to get the bonus civ-wide.

Wodan

.... Serving a vital need of the civ community.

 

[Uncle_Joe]

10% science might not seem like much, but it definately adds up over 100s of turns. Even if it only puts you 2 techs ahead in the whole rest of the game, 2 techs is 2 techs. Add in the potential for MORE happiness from FR AND the lower upkeep and its certainly not a no-brainer to remain OR for the bulk of the game.

But I think the point is that the situations vary and so does the 'better' Civic. There really isnt a blanked 'better' choice between OR and FR. I'd just like to see that reflected in the game a little more.

 

[Wodan]

How about a on topic post.

Here's a screenshot of my current game. It shows how the BetterAI is still not quite there, in regard to city placement. I've reported this before. I don't know if the BetterAI team intends to tweak this some more before releasing 1.0, but I hope so.

Anyway, take a look at where Stalin put StPete. A total of 6 tiles of overlap, and he gave up fresh water. All he gained was a shot at an extra spice. (The spice is disputed with Thebes.)

In addition, he makes impossible to work several tiles. (not resource tiles, normal tiles)

My guess is that the resources are still driving a huge modifier. The AI builds cities to get resources. Everything else is secondary. (I don't think that's the best way to have them do it.)

Wodan

 

[jkp1187]

How about a on topic post.

Here's a screenshot of my current game. It shows how the BetterAI is still not quite there, in regard to city placement. I've reported this before. I don't know if the BetterAI team intends to tweak this some more before releasing 1.0, but I hope so.

Anyway, take a look at where Stalin put StPete. A total of 6 tiles of overlap, and he gave up fresh water. All he gained was a shot at an extra spice. (The spice is disputed with Thebes.)

In addition, he makes impossible to work several tiles. (not resource tiles, normal tiles)

My guess is that the resources are still driving a huge modifier. The AI builds cities to get resources. Everything else is secondary. (I don't think that's the best way to have them do it.)

Wodan

I don't know, this one isn't necessarily what I'd do, but it's arguable. I mean, eliminate Thebes and there's no problem, St Pete gets access to extra spice AND the clams after border expansion. Plus, it is 3 squares from Moscow for easy reinforcement. I mean, this IS Stalin, after all....

 

[Wodan]

I don't know, this one isn't necessarily what I'd do, but it's arguable. I mean, eliminate Thebes and there's no problem, St Pete gets access to extra spice AND the clams after border expansion. Plus, it is 3 squares from Moscow for easy reinforcement. I mean, this IS Stalin, after all....

Yeah, and he already has 2 spice in Moscow.

Besides, why "eliminate" Thebes when you can conquer it?

Overlap is as much as concern when it is internal to your empire as when it is with another civ.

Bottom line, we can't attribute too much intelligence to the AI. Contingency planning on "elimination" of a rival city vs "I already have this resource" are probably too much to ask.

All I'm pointing out is that I strongly suspect the algorithm is placing too much emphasis of resources within the fat cross.

Wodan

 

[jkp1187]

Yeah, and he already has 2 spice in Moscow.

Besides, why "eliminate" Thebes when you can conquer it?

Overlap is as much as concern when it is internal to your empire as when it is with another civ.

Bottom line, we can't attribute too much intelligence to the AI. Contingency planning on "elimination" of a rival city vs "I already have this resource" are probably too much to ask.

All I'm pointing out is that I strongly suspect the algorithm is placing too much emphasis of resources within the fat cross.

Wodan

Didn't notice the spices in Moscow. That makes it less arguable.

 

[Roland Johansen]

Would you rather take the risk that the people reviewing your report take a look, say to themselves, "This doesn't prove anything," and simply delete it and move on? It's it much better to have me pick it apart first.

Wodan

.... Serving a vital need of the civ community.

You were right. I was a bit tired when I posted it. I should have added the numbers in the first place. The numbers + the graph would have been better.

Re: the 10% not being worth it

On that logic, why build Libraries? It's only 25%. If it takes on average 6 turns to research a technology (very dependent on game speed), then it will take you 24 turns to get 1 technology ahead of your opponents. And at that moment they will start benefitting from the research bonus that civlizations get for technologies that have already been developed by other civilizations.

And, it's even worse than that. You have to spend production capital to make the Library in the first place. Plus, you have to build them in every city to get the bonus civ-wide.

Wodan

.... Serving a vital need of the civ community.

I was arguing that it was not very likely that the science output of free religion would cause a dramatic science lead in the late game. The bonus is just too small. Nor Me is right that the relative bonus from this civic is closer to 5% because lots of buildings which increase science output have already been build this late in the game and the 10% science bonus is only applied to the base science output.

The bonus of organized religion is also not really 25% anymore in the late game in your old cities which already have large production bonusses. And it only applies when you're constructing buildings. It's quite hard for the AI to compare these civics.

The AI needs to do the following things:
1) put a relative value on 1 science point compared to 1 hammer.
2) calculate the increase in science per turn
3) estimate the amount of production going towards building production and calculate the increase in hammers per turn.

I don't know if the AI has access to all the numbers necessary to do the calculations. I don't know how easy it is to do this comparison. It's clear that the choise is not easy as we humans don't even agree what is the best option.

By the way, I agree with your city placement criticism. The AI should have placed that city one tile to the east. I also think that the AI picked that location because of the value it places on resources.
It is of course very difficult, maybe even impossible, to write a city valuation method which will result in the 'right' city placement. When humans place a city, they look at all the land around the city to see how it will fit in a continent wide city placement pattern. The AI uses a very local valuation method. And if this valuation method doesn't value resources highly enough, then it might not go for the cities with many resources early enough.

I would prefer a two step city placement algoritm. A first step where the AI divides the land into a city placement pattern so that very few tiles are lost and cities have low overlap. And a second step where the AI actually chooses which city in this pattern is the most attractive one to build first. The first step in the algorithm wouldn't value resources very highly but mainly look at city overlap and tile loss (tiles inbetween cities which will never be used). The second step would value resources very highly to determine which city is the most attractive one to build first.
I don't know how feasible such an algorithm would be. Especially the first step might be difficult because it requires a significant amount of calculations to determine an 'ideal' city placement pattern. Every city placement is dependent on the city placement of other cities and the number of combinations of city placement grow very large very quickly.

 

[RFHolloway]

Could you weight down the values of second and subsequent resources?

e.g. first spice worth 100, second one worth 80, third and subsequent ones worth 40. That might help.

 

[Roland Johansen]

Could you weight down the values of second and subsequent resources?

e.g. first spice worth 100, second one worth 80, third and subsequent ones worth 40. That might help.

In general, the second instance of the same resource is worth less. But on the other hand, if it allows me to trade for something that I don't have myself it can be worth exactly the same as the first instance of the resource.

 

[jkp1187]

I saw the AI not building the Internet when it could. So I had a look at the source and it seems your not the only people to have trouble with > and <. As far as I can tell, the AI will only build it when they've met 2 or fewer players.

That's bizarre. Can you point me to the place in the code where this is located? Is it just a simple fix of replacing a "<" with a ">" (or vice-versa)? (And this should probably be reported in the Bug Reports forum....)

 

[Nor Me]

I don't know if the AI has access to all the numbers necessary to do the calculations. I don't know how easy it is to do this comparison.

In this particular case, it wouldn't be too hard to calculate this. It should be simple to calculate the value of these civics for each city. Some civics and civic combinations would be more difficult.

Blake posted the heuristics he used for the civics a few months ago. They only depended on easy things like the number of cities. They are an improvement on vanilla civ. Having said that, I still think that Free Speech > everything might be better than the current Better AI legal civic choices.

It wouldn't be too hard to come up with better choices based on more direct calculations without slowing things down noticeably. But the effort involved means that I doubt it's going to be a priority however much we complain about it.

I would prefer a two step city placement algoritm. A first step where the AI divides the land into a city placement pattern so that very few tiles are lost and cities have low overlap. And a second step where the AI actually chooses which city in this pattern is the most attractive one to build first.

Look at the competition Stalin appears to be facing in Wodan's screenshot! This is going to be a bad idea for him. A human might think that they can count on conquering Thebes but that's a little optimistic for an AI. This kind of thing happens often enough that planning for the future might not be worth it.

That's bizarre. Can you point me to the place in the code where this is located? Is it just a simple fix of replacing a "<" with a ">" (or vice-versa)? (And this should probably be reported in the Bug Reports forum....)

It's in CvCityAI::AI_projectValue() in CvCityAI.cpp, one of the lines with getTechShare(). I've reported it, just not on this site. When I changed it, Cyrus built the internet so it looks to be genuine.

 

[Khamul]

Is it just my imagination, or is there some changes in capitulation? Does the Better AI capitulate more easily to other AI than before?

My last game (Prince level) I and Augustus were fighting against Gengish. I captured one city, and Augustus captured one city. Gengish had still about 10 cities, Gengish managed almost completly destroy my stack and there was not much roman groud forces left - and Gengish capitulates to Augustus. Gengish still had 95% of land and 45% of population. So the war was over. Augustus got a good vassal, and I got one city, surrounded by mongol culture.

This is a second time something like this happens: AI capitulates easily to other AI, without suffering any real losses.

 

[jkp1187]

I have not seen this. The AI will capitulate to any player when they're at war and an extreme imbalance of power exists (and they're through the initial stage of "REFUSES TO TALK". I once had Ragnar capitulate without setting foot in his territory. I've also had civs capitulate to me only after taking one or two cities from them (both before and after BetterAI.) Depends on the situation, I think....

Is it just my imagination, or is there some changes in capitulation? Does the Better AI capitulate more easily to other AI than before?

My last game (Prince level) I and Augustus were fighting against Gengish. I captured one city, and Augustus captured one city. Gengish had still about 10 cities, Gengish managed almost completly destroy my stack and there was not much roman groud forces left - and Gengish capitulates to Augustus. Gengish still had 95% of land and 45% of population. So the war was over. Augustus got a good vassal, and I got one city, surrounded by mongol culture.

This is a second time something like this happens: AI capitulates easily to other AI, without suffering any real losses.

 

[Yakk]

So I was fighting a War with the Russians -- and they pillaged their own roads and improvements (inside their cultural borders).

This might be them being extremely clever -- "Pillage it before you do" or "Pillage it before you capture the city" -- but it seemed very strange to me.

 

[jkp1187]

So I was fighting a War with the Russians -- and they pillaged their own roads and improvements (inside their cultural borders).

This might be them being extremely clever -- "Pillage it before you do" or "Pillage it before you capture the city" -- but it seemed very strange to me.

Whoa, that's interesting. Do you have "before" and "after" save files by any chance? I'd actually be interested in seeing that.

Scorched earth policy!

 

[ese-aSH]

So I was fighting a War with the Russians -- and they pillaged their own roads and improvements (inside their cultural borders).

This might be them being extremely clever -- "Pillage it before you do" or "Pillage it before you capture the city" -- but it seemed very strange to me.

so much realistic !!! the ai must have read the stories of the russian campaign of napoleon !

 

[ese-aSH]

How about a on topic post.

Here's a screenshot of my current game. It shows how the BetterAI is still not quite there, in regard to city placement. I've reported this before. I don't know if the BetterAI team intends to tweak this some more before releasing 1.0, but I hope so.

Anyway, take a look at where Stalin put StPete. A total of 6 tiles of overlap, and he gave up fresh water. All he gained was a shot at an extra spice. (The spice is disputed with Thebes.)

In addition, he makes impossible to work several tiles. (not resource tiles, normal tiles)

My guess is that the resources are still driving a huge modifier. The AI builds cities to get resources. Everything else is secondary. (I don't think that's the best way to have them do it.)

Wodan

hi wodan ! in a way i desagree with you : having 6 common tiles for 2 cities means that each cities may work 17 tiles (instead of 20). and thats not that bad !
such a situtation would obvously cause no pbs in the early games (because the cities are not big enough to work all tiles) and finally have reduced impact on late game (the cities only miss 3 standart tiles, which is not that great).
I change my city placement recently, allowing my cities to share some tiles : and that works ! I only keep exceptionnal spots clear (for main prod city or main science city).

 

[Wodan]

hi wodan ! in a way i desagree with you : having 6 common tiles for 2 cities means that each cities may work 17 tiles (instead of 20). and thats not that bad !
such a situtation would obvously cause no pbs in the early games (because the cities are not big enough to work all tiles) and finally have reduced impact on late game (the cities only miss 3 standart tiles, which is not that great).
I change my city placement recently, allowing my cities to share some tiles : and that works ! I only keep exceptionnal spots clear (for main prod city or main science city).

And what, exactly, benefit is gained for this overlap?

1) dynamic toggling of tiles worked (allowing 2 cities to share, for example, a high-output mine)
2) cities closer to capitol, resulting in lower city maintenance
3) more cities per total land area
4) other?

Note that none of the above are the case in my example.

Wodan

 

[Yakk]

The AIs attack tactics are questionable. A stack should know if it has a decent chance to take a city before it actually tries -- and I've had an attack stack that was cleary insufficient to kill more than 1 or 2 units attack a city over a river rather than move around it.

The AI also seemed prone to attacking the closest city. This made "super stacking" my defences against the AI rather easy.

Next, the AI doesn't view barbarian cities as an easy source of new city locations.

The AI also seems to be missing any anti-barbarian smarts. Barbarian attacks are highly predictable -- they tend to move their troops along known paths, coming from areas of high-fog. Quite often placing a handful of units along the barbarian approach paths (both defensive and offensive) seriously stems the tide. (I noticed this, because I modded up the rate of barbarian spawn, so areas with serious barbarian flow are much more common: the same effect of a barbarian flow can exist in standard civ, it just requires a larger foggy area).

Next, the AI doesn't seem to want to sally that much. I can often move a mediocre small stack near an enemy city, and the AI doesn't smash it -- even if it has a decent chance of being able to kill 2 units and only end up with at most 1 unit dead (and quite possibly none).

 

[ese-aSH]

And what, exactly, benefit is gained for this overlap?

1) dynamic toggling of tiles worked (allowing 2 cities to share, for example, a high-output mine)
2) cities closer to capitol, resulting in lower city maintenance
3) more cities per total land area
4) other?

Note that none of the above are the case in my example.

Wodan

well i guess 3) is the good answer where should russian have grabbed land south ? theyre surrounded ^^
plus they claim the second iron and thats a good point.
anyway i dont say i would have build a city there, i just mean this is not so stupid