[BTS] Immortal Wonderdates

[Major Tom]

I have compiled some statistics from my immortal games. This shows when the AI gets certain wonders and Lib, when I have not. It is standard settings (huts on!), Fractal and random opponents (6+human). The games are played with the Better BUG AI-mod, but not sure if that better AI alter priorities towards those wonders (think not).

Item| MedianYear| MedianTurn| Std.dev (turns)| Earliest| Latest| Count
Stonehenge| 2280BC | 43 | 7 | 32 | 60 | 21
The Great Wall| 1800BC | 55 | 10 | 35 | 68 | 15
The Oracle| 1080BC | 73 | 12 | 51 | 84 | 20
ToArtemis| 775BC | 84 | 12 | 66 | 108 | 23
TG Lighthouse | 800BC | 83 | 10 | 70 | 107 | 17
The Pyramids | 675BC | 88 | 11 | 72 | 114 | 17
TH Gardens | 25AD | 116 | 10 | 100 | 132 | 19
MoM| 250AD | 124 | 12 | 98 | 135 | 18
TG Library | 400AD | 131 | 12 | 107 | 155 | 19
Liberalism | 1200AD | 180 | 20 | 150 | 214 | 12
End date | 1928AD | 348 | 22 | 314 | 392 | 14

 

[Pangaea]

Thanks for doing this, very interesting and useful

From what I recall from normal BtS Immortal games, these dates look recognisable. And of course how early some of them go, depends on which AIs are present and if they have easy access to marble/stone (like capital BFC). See Stonehenge isn't included, but then it usually goes pretty soon.

 

[Major Tom]

Glad you liked it. Yes, some wonders are not listed as I do not aim for them myself. But perhaps one day I will expand the list. And yes things can vary a lot depending on AI present. Also on pangea type of maps things go usually faster. With two separate continents where everybody hates each other, teching and wonders are slow.

 

[Major Tom]

Do you think it is fair to assume that the first Great Artist born in the game comes from Music? Planning Music in next update, but not sure it is reasonable to use that GP-message.

 

[Major Tom]

Updated with Stonehenge, Mausoleum of Maussollos and the year game was won by AI. (either I didn't bother, missed it or abandoned the game an let Autoplay finish it).

 

[Kallikrates]

Somewhat re-assuring to see that Turn 67 (1320 BC) was really VERY early for the pyramids when I ragequit a few hours ago. Otherwise the numbers roughly agree with my experience. Some seem rather late but this might be because I usually make them high priority and try to build them myself (e.g. the Great library still in the BCs). For the ones I build reasonably frequently my rule of thumb was to get the Lighthouse and the Oracle before 1000 BC (better around 1200). And I was fairly positive that I'd be able to chop the mids by 1000 BC which seemed early. As it apparently is. It's probably Ramesses on the other side of the world or another IND bastard who happened to have stone nearby.
The lib date also seems late (but again if one makes sure to get it oneself one lacks experience when it would go otherwise).

 

[Major Tom]

I have updated the table with The Hanging Gardens. Planning some other stone wonders soon. Who would like Chichen Itza?

 

[Swordnboard]

Not particularly relevant for anything in the table up to the mids, but all of the later dates are greatly affected by how much one trades with the AI. Especially aesthetics wonders + great library, which can take the AI forever to reach on their own but we frequently trade around like candy.

@Major Tom are these isolated setups where you don't trade anything to the AI, or "normal" games you've played to an extent?

 

[Major Tom]

Not particularly relevant for anything in the table up to the mids, but all of the later dates are greatly affected by how much one trades with the AI. Especially aesthetics wonders + great library, which can take the AI forever to reach on their own but we frequently trade around like candy.

@Major Tom are these isolated setups where you don't trade anything to the AI, or "normal" games you've played to an extent?

It is the latter, random fractal maps so could be anything (iso, semiiso) but usually good trading opportunities.
Anyway, have a hard time to see your point. Techtrading human-AI is the name of the game. The average date should reflect what to be expected. Ofc, if you don't trade away aes and not build The great Library yourself that date should be longer. At IMM I guess AI can tech and build more than Pyramids by their own means before you have alpha and relevant techs to trade away. Hanging Gardens and MoM springs to mind.
Edit: BTW, I feel many games on IMM, AI-AI trades and bribes are quite frequent.
Edit2: Ok, I can agree that in iso things go slower. Wrote in third post that on pangea maps thing may go faster.

 

[Swordnboard]

Thanks! I agree with you that games with trades, etc. are more in line with "real" conditions. I just wanted to confirm that is what you were doing, because sometimes I like to collect data on early stuff especially by walling myself off somewhere and hitting enter a bunch of times

 

[Major Tom]

No worries

 

[Major Tom]

Thanks! I agree with you that games with trades, etc. are more in line with "real" conditions. I just wanted to confirm that is what you were doing, because sometimes I like to collect data on early stuff especially by walling myself off somewhere and hitting enter a bunch of times

If I find time I can sort out how many of each maptype it is (iso, semi, continents and pangealike (all fractal tho), and also plot in number of IND leaders in each game and check differences. Multiple regression formula should suit the task.

 

[Powerfaker]

@Major Tom
Really nice job, collecting these numbers.
Can you tell from your data if dates of the early wonders set a trend for dates of later wonders? What about Artemis, that's often nr 4 to go?

 

[Major Tom]

Sure, a trend that you mention should be easy as data are in a spreadsheet with appropriate rows and columns. Could also investigate trend about causing it.

Might look at it tomorrow.

 

[Powerfaker]

Not sure if you got my point => if for example in a certain game SH and GW are built earlier than the median year, is it more likely that Oracle or Mids etc get built earlier than median as well?

(Maybe tendency is a better word than trend? I'm not English native speaker)

 

[Major Tom]

Yes, got it.

 

[Major Tom]

@Major Tom
Really nice job, collecting these numbers.
Can you tell from your data if dates of the early wonders set a trend for dates of later wonders? What about Artemis, that's often nr 4 to go?

Updated table with Artemis.
I looked briefly at trends. It is only a weak correlation between wonders (but it is there). F.ex. if Oracle goes early, Artemis is early only 1 in 5 times (correlationcoefficient +0.27, measured by deviation of the finishingturns to the mean.)
When looking at some saves one funny thing is that if an AI is isolated he seems to prioritise wonders (especially with access to stone or marble). Take a look at some of the saves from BotM 160 where Ghengis Kahn (!) was in iso and started spamming wonders. http://gotm.civfanatics.net/results/index.php?month=70160&game=70160 Here is a screen from my save:

Spoiler F9 :

 

[Powerfaker]

Thanks for digging into this. I get the feeling there isn't a trustworthy correlation. Settings, or something like Iso you mentioned, may be more influential.
A funny date I got a couple of months ago was Pyramids ± 1500 AD (!!) on a tiny terra map with Willem and Isa, part of the Rise of the English challenge series (didn't submit this one).

 

[Major Tom]

Sure, the table cannot be used on other map sizes than standard (6 AI). The average dates has typically one AI with stone, one with marble and/or one IND leader.