AI - human trade revealed

Hi. I have now replaced this article with a rewritten version that got its own new thread: AI - human trade revisited

This is grea TNO. The tech selling on higher levels in particular. I know that the research cost decreases the larger the number of AI already holding the tech.

Using the same 1000 cost of tech how much would I get by selling it each of the 7 AI compared to their respective costs to develop it. I understand that you get a 45% additional bonus for the first AI (on Deity).

What about the second AI and the following? Does the bonus increase for the last AI players or it does not change in %.

What if I buy a tech worth 1000 for 1917 and sell it to two other guys. How much I would get?

Thanks,

Yndy, you're not making it easy for me

I must come back to you after I have checked theory with practice...

straight to the FAQ!

thanx a huge bunch!

I also posted on the other site, but if this is correct (which I assume it is considering TNO:s very good record), it would mean the human is in fact handicapped on Regent, which is supposed to be an equal match between the human and the AI.

I would really like to see some further tests and cross-checkings.

Hurricane. Here you'll get the raw numbers, so feel free to check them (and do further testing if you want to):

This is a scenario with one AI civ only. The monarchy price in paranthesis is when the Hanging Gardens is already built.

AI seller (minimum gold the AI accepts)
Chieftan: Currency 342, Monarchy 694 (684), Republic 794
Warlord: Currency 345, Monarchy 700(690), Republic 801
Regent: Currency 355, Monarchy 721(711), Republic 825
Monarch: Currency 402, Monarchy 815(803), Republic 934
Emperor: Currency 479, Monarchy 968(956), Republic 1112
Deity: Currency 654, Monarchy 1321(1308), Republic 1522

Human seller (maximum gold the AI will pay)
Chieftan: Curr 342, Mon 694, Rep 794, wine 300
Warlord: Curr 339, Mon 688, Rep 787, wine 298
Regent: Curr 329, Mon 668, Rep 764, wine 289
Monarch: Curr 359, Mon 726, Rep 833, wine 283
Emperor: Curr 383, Mon 773, Rep 887, wine 268
Deity: Curr 495, Mon 999, Rep 1152, wine 261

I also did a few tests where I changed the AI-AI trade bonus and got exactly the same numbers as above, and a few tests where I changed the AI cost factor and found the results completely consistent with my formula.
Ex: A deity game with AI cost factor increased from 0.6 to 1 would change the demand for republic from 1152 (794 / 0.6 * 1.15) to 913 (794 / 1.0 * 1.15).

Very interesting. Curious how did you develop the tests.

== PF

Great job!!!

This helps out a lot!

Originally posted by planetfall
Very interesting. Curious how did you develop the tests.

Click to expand...

The test was developed in the same way that the other tests I've done: Create a small scenario that isolates exactly what I want to test.

In this case I made two small adjacent islands, each with one city (one egyptian and one roman) and one warrior of each side on the opponent's island for immediate contact.

I gave both civs all techs up to currency, while only one of them got currency, monarchy and republic. I also gave both civs 10.000 gold so that tech buying wouldn't be a money issue.

I then started the scneario on chieftan, checked the best price the human could get for all techs (without actually buying/selling any), reloaded and repeated on warlord. I think you'll get the idea from there.

To test the luxury trade, I increased the size of the Egyptian island to host 30 towns, all connected by roads, and one witha harbour. On the roman island I added roads, a harbour and two wines. I then checked what Egypt would pay for the wine on all six difficulty levels.

Very impressive Egyptians in fact were chosen well for testing because of their average aggresiveness (3 IIRC). More peaceful civilizations are actually better trade partners, always offering more lucrative deals (France and India were always my best customers and source of goods - I tend to trade with more aggresive civs like Persia only to raise their attitude). But Egypt example gives indeed better general idea how this trade thing works. Really great job!

TheNiceOne,

Thanks for the explanation now I now how to test things I am curious about.

== PF