Building All the Ancient Age Wonders

[CKS]

Ha! In each of my games, I've had cities that rioted and units that I sent the wrong way because I thought I was moving a different unit. That is sloppy. The poor micromanagement is just laziness, though.

I think this shows that it isn't as hard as I thought it would be, and that the AI stinks on archepelago maps at all levels.

 

[Buttercup]

It's funny you should say that, as my recent result suggests there's more going on behind the scenes than we might not know about.

I had another stab at the Huge, French, Regent with the following criteria:

Raging Barbs, Archipelago 80%, Normal, Temperate, 3 Billion Years, Default Rules, Normal AI Aggression with the following eleven civs: Aztec, India, Japan, America, Arabia, Rome, England, Carthage, Iroquois, Greece, Korea.

This time, after a few laughable starts I immediately rejected, it spawned me next to a lake with three cows and two bonus grasslands and only one move to the coast. I'm not going to reject that.

The first try was fantastic, I was absolutely steaming along, I didn't even bother going for Code of Laws, just stayed in Despotism the whole time and took Monarchy after Philosophy (and still didn't revolt). I got nearly all of the Wonders with consummate ease, just the last leg to go: Iron Working, Mathematics and Construction for the final one, The Great Wall. With just 13 turns left to complete The Great Wall I got a message telling me that the Carthaginians had completed The Great Wall. Waaaat. This was in 250AD, which seems remarkably early for an AI to get TGW.

So I had a second try from my 4000BC save and this time did the usual route of one first tier tech them going straight for Writing through to Republic, only completing the Colossus on the way. I completed the Pyramids shortly after revolting to a Republic, which triggers the Golden Age for the French and I picked up the Oracle really easily. I then got an SGL for Polytheism and grabbed the Temple of Artemis instantly(although I don't think I really needed it?). I then proceeded to grab all the other Wonders, again with consummate ease before doing a run for Construction.

… But what of The Great Wall, by the time I'd got to Construction I was already thinking I'd lose it again, there really wasn't that big a difference in date to the previous method. But no message came and I constructed TGW in 360AD. Oh wow, I finally did it with the French, so just The Great Library to get for a final date of 590AD.

Which got me thinking...

Does the AI target Wonders in the early eras?? In the first game I grabbed all the Religious Wonders so incredibly early that perhaps that triggered the Carthaginians to then abandon that tree very early and completely beeline Construction whereas in the second game I came to those Religious Wonders much later...

Or perhaps it's something else entirely. Maybe the Carths were not at war in one game but were in another. Lots of different possibilities, but the upshot is that just because the AI appear rubbish in one run through, it doesn't mean that they will be if you tried to repeat the process on the same map. Is it luck? Or does going for Republic also have yet another hidden benefit that we were all unaware of?

Because, by date, the first, non-Republic, example was actually more efficient.

Edit: I had zero contact with any other civs and chose to learn all my own techs for these two runs, so tech trading via me didn't influence anything.

 

[Buttercup]

I've still been beavering away on the same French 4000BC save for a while now and I have to say, I've found some results that seem rather fascinating and initially unexplainable. There really is something weird going on here.

Here's the data:

Version 1:

Not building any Wonders and just waiting to see when the AI builds them all, various playstyles used throughout each try but always remaining in Despotism [four run-throughs]:

Colossus gets built between 710BC and 800BC (and always by the Aztec)
Oracle is 70BC to 50AD (various but America twice)
Pyramids is 50AD to 300AD (various)
Great Lighthouse is 190AD to 610AD (various)
Mausoleum is 390AD to 670AD (various)
Temple of Artemis is 320AD to 470AD (various)
Hanging Gardens is 410AD to 620AD (and always the Americans)
The Great Wall is 390AD to 660AD (either English or Carthaginians, two each)
The Great Library is 460AD to 640AD (either English or Carthaginians, one each, only completed twice in the time frame)

And there's no evidence in the results to suggest early completion hampers later completion, everything appears remarkably random within it's own timeframe. There is certainly no evidence here that 250AD is an expected completion date for The Great Wall, or any of the Wonders after the first four.

Version 2:

In this version I again run through the scenario four times, only this time I'm attempting to build all of the Wonders using a whole variety of methods: noted as one attempt not bothering to go Republic, one attempt going for building them all in my Capital plus Republic, then two more Republic attempts without the Capital stipulation:

Attempt 1 - not bothering to go Republic - all Wonders built by 660AD, The Great Wall was built long after 250AD.
Attempt 2 - going Republic but trying to build all the Wonders in the Capital - Lost the Mausoleum at 250AD.
Attempt 3 - going Republic but no stipulation as the where to build them - all Wonders built by 590AD, The Great Wall was built long after 250AD.
Attempt 4 - as attempt 3 - all Wonders Built by 510AD, The Great Wall was built long after 250AD.

And I find it fascinating that I again lose a Wonder at 250AD, and even more fascinating that it's not even the same one, but it's still one that has no trend to finish that early by any normal metrics.

Version 3:

In this version I'm following my usual Wonder building path with the exception of also building The Oracle, that is: I build Colossus, then side-track to The Oracle, then get the Republic sling, then build The Pyramids, get the Golden Age, then build The Great Lighthouse, then build the Mausoleum, with the only break in the run to spend a turn or two building a Temple or a Library (and the initial Warriors, Worker and Settler obviously), then build The Great Library just because it's either that or Wealth and because at least it prevents an AI from getting it. So in this version I'm leaving the Temple of Artemis, the Hanging Gardens and The Great Wall for the AI to get whenever it wants. After just two attempts a very stark pattern has emerged:

Attempt 1 - 570AD Artemis is built, 580AD Hanging Gardens built, 810AD Great Wall built.
Attempt 2 - 700AD Hanging Gardens is built, 710AD Temple of Artemis is built, 780AD the Great Wall is built.

And this really shows how something weird is happening here. In both instances the Great Wall is not only last, but last by a huge margin, that the Temple of Artemis is also extremely late by it's own standards as is the Hanging Gardens.

And it's only by examining all of these variations that one realises that there is definitely something else going on besides regular random variation. Both version 2 and version 3 are examples of all of the early Wonders being completed early, there is virtually no room for the AI to pre-build anything in version 2, and yet it was in version 2 that I once again hit the 250AD problem. In version 3 the AI has nothing stopping it from beelining the three Wonders and yet all of the dates for these three Wonders are vastly later than if I either hadn't built any Wonders or if I had tried to build them all, it makes no sense at all.

Most importantly, it doesn't seem to make any difference if you grab the three early Wonders early, there will always be a 250AD out for the AI for some reason as a high chance result. But only if you try to build all the Wonders.

I can't explain the reasoning for this though, I have no idea how to make this relevant to a power gamer or general player, I just know there's something in this.

 

[Lanzelot]

Are you playing with SGLs? I think that the AI, if it gets one, will use it to rush a wonder. So I think unless you manage to get all the techs first, it may be better/safer to play without them.

 

[CKS]

In Buttercup's situation, you may be right about SGLs, but in general I think they help the player more than the AI. I don't think the AI saves SGLs, so if they can't rush a wonder right away, I think they'll use it for whatever they can right now. Also, if you stack the deck (with civ and map selections) to make it easier to build them all and play a fast research game, the chances of the AI getting an SGL are not very good.

Buttercup is playing a non-scientific tribe against Korea and Greece, which are both scientific and commercial - the tribes most likely to pop SGLs. Under these conditions, if you can't play a fast research game, an SGL is pretty likely to cause issues.

 

[Buttercup]

Yeah, I'm pretty sure SGLs aren't the thing here. I'm not counting it out, it's certainly a possibility, but as CKS states, it's a big outlier.

I got bored of reloading the start so today I had a stab at completing a 20k. I used the version 3 method and was again able to build all the Wonders in my capital, building both Artemis, Gardens and Wall all at very relatively late dates.

Unfortunately, I was too efficient & got to Education before remembering I hadn't built the Library yet & then I repeated the mistake by getting to Railroads before I'd started Templars. Lol.

But I can now cross off the notion that it's some kind of 20k level-scaling thing. I'll probably have a few more investigative sessions, but there seems to be so many permutations that I'm not sure if I'll ever nail it. Even repeating an event once can still be random weirdness from two completely different reasons and if I can't replicate to 250AD thing then I'm not going to commit to anything, but if I do it'll likely spur me on to keep trying stuff out.