Spoonwood's Selection Strategies

[Spoonwood]

If you check the opponents in many HoF games for say a diplomatic or spaceship victory, you'll find that the opponents have the scientific trait. A lot of 20k games have a bunch of scientific opponents, though there also exist some 20k games where the opponents seem like weak researchers. Some of those principles might usefully get applied to some other games... say pangea games at a level where you struggle with the tech pace or where you want to decrease the tech pace.

One principle in selecting your opponents might lie in maximizing your tribe's benefits with respect to your opponents. Specifically you might want to maximize the advantages of your starting technologies to hope you can trade for more ancient technologies with less resources. So, as an example, let's say that you want to play with Rome on a standard map with 7 opponents:

1. You have Alphabet and Warrior Code as your starting techs. So, you don't want to have any Militaristic, Commercial, or Seafaring tribes to start with. An exception though is Japan. This way, the AIs will either have to research your starting techs for themselves or get them from you.

2. Say you also want to eliminate the scientific trait also to hopefully give you more time to build up an economy. Then your opponents could be Egypt, Japan, Maya, Inca, America, the Celts, and Arabia.

A potential downside might be that the AIs will have less commerce than if you had selected more Seafaring and Commercial opponents. For such a list, Japan also has a unique position in that they make for the only tribe that starts with a unique tech.

As another example, let's say you pick Germany as your tribe. Then, you don't want anyone scientific or militaristic... except possibly for Japan. But, let's say you also want to maximize the number of tribes with Alphabet in hopes of trading for it early on.

For your opponents then you could have The Hittites, The Iroquois, England, France, Carthage, India, and Spain.

If you use that principle of always having your opponents not have your starting techs, then every time you play a different tribe you can have different opponents without selecting random opponents. Tribes with Alphabet and Pottery have the least number of possible opponents when using that principle. Japan has the maximum number of possible opponents.

 

[justanick]

It is well known that religios civs have a anarchy duration of 2 turns. This value does overwrite the maximum anarchy duration for AI which at Sid is zero. Choosing religios oppenents at high difficulty level may therefore help. At Sid maximum anarchy is zero, at Deity and Demigod it is 1 turn, at Emperor it is 2 turns.

 

[Lanzelot]

Religious is also a trait that is least useful for the AI: Temples are a waste in most cases anyway, so it's no big deal, if the AI gets a discount on these useless things...

Another criterion for selecting opponents is also: do you want strong trading partners and fast research? Then chose agricultural opponents. The AI handles this trait quite well. Do you want to slow down the tech pace and keep the AIs weak? Then avoid the agricultural ones.

 

[Spoonwood]

Another criterion for selecting opponents is also: do you want strong trading partners and fast research? Then chose agricultural opponents. The AI handles this trait quite well. Do you want to slow down the tech pace and keep the AIs weak? Then avoid the agricultural ones.

I would have thought that commercial makes for a better trait for the AI's research at upper levels, since they already have so much of a food discount to begin with. They also end up with commerce earlier if they have the industrious trait. And depending on map type, the seafaring trait also gives them more commerce.

Also, doesn't the AIs tendency to sometimes water tiles partially make up for them not having food due to overrun with the agricultural trait?

I'm not saying you're wrong Lanzelot... I'm just saying that my intuition isn't the same here. I don't have any playtests, say by using random seed with different opponents to test this. Nor do I recall any sort of tests along these lines. If you Lanzelot or anyone else does any playtests, you might want to turn on mass regicide and use your king units as scouts and limit trading/make sure you do trades at the same time to try to control for other possible variables with respect to the tech pace than traits. Also, you can use the same seed number with all the other settings the same to ensure that you have the same map. Though, I don't know if your starting position would be the same, and AI-AI trading will differ with different opponents.

 

[justanick]

Religious is also a trait that is least useful for the AI: Temples are a waste in most cases anyway, so it's no big deal, if the AI gets a discount on these useless things...

But as AI is unable to properly use luxus rate those temples and cathedrals may be more useful for them as for humans.

I would have thought that commercial makes for a better trait for the AI's research at upper levels,

I think so, too. Also commercial civ tend to have more gtp they are willing to trade away, especially if they priotize build wealth as England AI does. Those trades could in turn of course reduce their research to normal levels.

 

[Lanzelot]

I don't have any play tests to support my theory, it's just my (perhaps subjective) memory of many past games: whenever there was a runaway/strong AI, it most often seemed to be an agricultural one. Perhaps coincidence, perhaps I just "forgot" the cases where it was a non-agricultural AI...

But my expectation would be, that agricultural is a useful trait on any difficulty level: on Sid, the average town without food bonus grows every 4 turns. If the map is right and the AI founds many towns on fresh water, that would be only 3 turns for an agricultural AI. In the early phase this is a big bonus that should give them an edge over their competitors?!

But you are right that being AGR is not essential on let's say Deity and Sid: on these levels almost any AI will be strong enough to become a useful trading partner for the human player...

On the medium levels (Monarch - Demigod), however, I think there is some truth in my observation. In order to keep up with the development of the human player, the AI either needs a very good start position or needs to be agricultural. (On the lower levels, everything is hopeless anyway... )

 

[Lanzelot]

But as AI is unable to properly use luxus rate those temples and cathedrals may be more useful for them as for humans.

That's true. Which reminds me of another useful "trick" to ne used, if you want to create strong trading/research partners: gift your friends your spare luxury resources! (When the AI runs into happiness problems, it uses clowns instead of the lux slider, which of course cripples their growth and their commerce. By gifting them luxuries, they can put the clowns back on the fields, continuing growth and strengthening their economy.)