March Madness: Civ4 AI Edition

[Brian Shanahan]

* Thanks to the brave/bored souls who submitted predictions for all twelve of our games: Ceiliazul, Bobchillingworth, von Adlercreutz, slowcheetah, Dantski, Commodore, David Corperial, Yell0w, Hashoosh, Twinkletoes, Sian, and Brian Shanahan. That's a lot of ballots to get in over two months of time. I hope everyone had fun.

Next time we do this, I'll be making twice the picks, one set for myself and the second a rng picked set. That way we'll have a proper benchmark for randomness.

Brian Shanahan's random number generator entries led to a series of zero scoring results, and the lowest per-game score of the competition at 6.5 points.

[Dantski]

Thanks for running this Sullla, I always like seeing some stats at the end too.

[Katon]

Brian Shanahan's random number generator entries led to a series of zero scoring results, and the lowest per-game score of the competition at 6.5 points. mischief

And yet it actually did OK by the best-five scoring. Go figure.

As everyone else has said, thanks for running this, for the effort you put into keeping track of everyone's predictions, and especially for the very entertaining writeups.

[Tyrmith]

Thanks Sullla that was a ton of fun, glad I was able to catch up at the end of the series.

[Ellimist]

The following turns were relatively quiet. Suryavarman and Suleiman continued their ineffectual war, exchanging units in largely pointless fashion, no territory changing hands. Lots and lots of swords and axes and chariots dying along the border, like something out of a medieval tragedy. Justinian built the Buddhist shrine and Elizabeth built the Hindu shrine. (Note: the AIs always produce tons of Great People on Deity because they run so many specialists all the time. Thus nearly every holy city always gets its shrine in these games. This is another reason why Philosophical seems to be fairly pointless as a trait for the Deity AIs.)

I think you may have that backwards. The AI tendency to use so many specialists is probably the reason they do so well with PHI. They're going to use the specialists regardless, but PHI means that they get more of a benefit from it.

[Ceiliazul]

that's a cool chart, but how did you differentiate between 52 and, say, 32? Score at time of death?

[Ellimist]

that's a cool chart, but how did you differentiate between 52 and, say, 32? Score at time of death?

He ranked them by the turn they were eliminated on.

http://www.garath.net/Sullla/Civ4/survivor14.html

[von Adlercreutz]

Thanks for doing this!

[Ituralde]

It was a lot of fun to read this and being able to bet on the outcome made it all the more interesting! Thank you for doing the fantastic writeups and also for tracking all the Scores. I'm a sucker for statistics so all the different ways of rating were really interesting. Also the analysis of the different traits and leaders was interesting.

Can't believe I got into the Top Ten with my Championship votes!

So... same thing next year?

[Brian Shanahan]

I think you may have that backwards. The AI tendency to use so many specialists is probably the reason they do so well with PHI. They're going to use the specialists regardless, but PHI means that they get more of a benefit from it.

The real value of PHI is getting the early specialists out quicker, and then using them properly. By the end of the game you're not actually going to be that far ahead with PHI than without (as the numbers get very high after about seven GPs popping), so the AIs with their inability to properly use their GP pops strategically (e.g. settling GAs in stupid locations, not emphasising GSs for beelining purposes) won't actually get that much extra use out of PHI.