When does the AI construct wonders???

[MensSana]

Wonders does sometimes give you a number of serious benefits, but the drawback of wonders is that you must be a wonders only constructor to build it and get that benefit.
Sometimes it is not even worth start to build a wonder because there must be an AI already on the verge of finishing it.
But when do the AI build a particular wonder???
I know this can variate from game to game, but would like to get some information on average when one can expect the wonder to be built elsewhere.

The main parameters must be

• Difficulty level
• Version of civilization IV (1.61 or warloard 2.08)

Factors like what resources the AI has, what AI civilizations we are facing and their innate abilities does variate from game to game, so it will be difficult to calculate based on those parameters.

But since we all played a number of games, it would be easy to calculate some statistics on when the actual wonders are being built, and the probability that we are the first to build some wonders giving us an edge on when to pursue a specific wonder

 

[VoiceOfUnreason]

But since we all played a number of games, it would be easy to calculate some statistics on when the actual wonders are being built, and the probability that we are the first to build some wonders giving us an edge on when to pursue a specific wonder

I'll offer you a suggestion on how you could work this out for yourself....

There are a number of examples where many people play the same game (GOTM? SGOTM. Realms Beyond Adventures/Epics. Challenges) where you can download a save from the end(ish) of the game. You've got access to a Gyathaar's log extraction utility, which will collect for you all of those dates, and which leader built the wonder. Perl is a free download. Mix utilities in an oven safe dish and bake on the lower rack for 90 minutes at 425; when crust is golden remove from the oven and let it cool. Serves thousands.

 

[MensSana]

Thats a good suggestion
I thought maybe people could post there findings but accessing the logs would simplify matters and also look at all wonders on a larger scale.
But the suggestion with perl ....Naahhh. I do not like the programming language a lot, and I'll stick to C++ since it is much faster to go through a lot of log files.
If people could directly post the logs in a text file, after using Gyathaar's log extraction utility, together with infor regarding version and difficulty, I would be pleased.

Of course I will post my findings in this thread after going through a few log files.

 

[Markus5]

I've always thought that the AI was an opportunistic wonder-builder. For example, to build the Hanging Gardens, you need an aquaduct. I don't think the AI plans ahead: "If I build an aquaduct then I can build the Hanging Gardens." But, if it builds an aquaduct because it needs it (or will need it, or just because its programmed to build it at some point) then the opportunity to build the Hanging Gardens is there. I can't prove this. I've never looked at the code. Its just a feeling.

 

[ratrangerm]

The things I've noticed about the AI and the Wonders it favors:

* The AI tends not to build The Hanging Gardens as much as other Wonders.
* The AI seems to chase Angkor Wat and Spiral Minaret.
* An AI civ that builds Pyramids tends to have stone more often than not.
* Aggressive AI civs seldom build Great Library because they don't pursue Literature that early.

 

[flamingzaroc121]

in the XML there is something about the leaders that tell how likely they are to build a wonder. it is basically the "ooh ahh" shiny factor. for monty there is NO shiny factor. some other leader (who's name escapes me) every wonder has a shiny factor and they will almost always try to build wonders

 

[MensSana]

Yes, some wonders are more popular for the AI than others, and
some AIs tend to build a lot of wonders, whereas others totally do not care about them.
These are things we know, but would it not be wonderful to have some exact statistics whereas when one can expect a wonder to be built by the AIs??

In that case at a certain timepoint we know, for example, its no use to go for the pyramids right now because there is a 80% probability that it will be constructed within the next 10 turns.

Would that not be a good list to have handy???
Also as flamingzaroc121 pointed out, the wonderbuilding is dependent on which civilization the AIs are managing, so it must be useful to know when a specific wonder will be constructed by the AI depending on which AI civilization your are in contest with.

 

[cabert]

Yes, some wonders are more popular for the AI than others, and
some AIs tend to build a lot of wonders, whereas others totally do not care about them.
These are things we know, but would it not be wonderful to have some exact statistics whereas when one can expect a wonder to be built by the AIs??

In that case at a certain timepoint we know, for example, its no use to go for the pyramids right now because there is a 80% probability that it will be constructed within the next 10 turns.

Would that not be a good list to have handy???
Also as flamingzaroc121 pointed out, the wonderbuilding is dependent on which civilization the AIs are managing, so it must be useful to know when a specific wonder will be constructed by the AI depending on which AI civilization your are in contest with.

statistics cannot be exact in the sense that it's still only a probability
I suggest you to make it the other way round :
If you want a given wonder, just do everything you can to have it asap.
It takes prechopping, prewhipping, fast teching, not trading away the tech leading to it... But don't rely on statistics.
I can't tell you how often I could build stonehenge in the AD years.
Or how often an AI used a GE to rush a wonder i was chasing.
The statistics won't help you!

The best info you can have is :
- the propensy to build wonders for a given AI (it's in the xml)
- the availability of the wonder (tech trade screen!) for those AIs
- the GE wonders already built (a 1 turn build is hard to see coming, but the GE can be foreseen)
- the availability of the bonus resource (trade screen)

If you use those informations well, you should be able to know whether a given AI could beat you to a wonder or not.

 

[Vulcans]

i agree statistics are meaningless.
one key factor to consider is the AI research pace, keep an eye on what technologies the AI have. some games the tec rate is fast and some the tec rate is slower, naturally depending on how much tec you trade, the map, difficulty level, wars, etc.
now naturally monty wont be that much into the wonders, while Lois, Hatshepsut, Asoka etc will likely chase them as soon as they can. So look at who is on the map with you.
the other factor is their city size and production capability. Look at their city size, see how many mines etc they have, and so you can get an estimate of their likely output. Include any relevant buildings for the time in the later ages to increase production. Check which resources they have that could speed things up, or if they are organized. So now you know when they had the tec available to start it, how much production it requires, and how much production they have in their best production city. Then you can make a estimate, time + required production/estimated production. Assume worst case scenario that they start the wonder as soon as it