March Madness: Civ4 AI Edition

[wetbandit]

I was thinking of drawing the matchups one at a time, in order to concentrate on each game as it happened and leave some mystery about who would end up in each one. However, I can just as easily draw them all at once and put a full bracket together if that's what people would prefer. Give me some more feedback on what you'd like to see.

This should be even more fun as the tournament progresses. I advocate for the release of a full bracket. That way, everyone will have time to vote over the next few days on every matchup. Then you can go on and resolve each tournament game at your speedy lurker-friendly pace (), without having to wait for a voting period in between each game.

Thanks for putting this together.

[Fintourist]

An alternative suggestion would be to take all the third place leaders, and run them in a winner takes all reprechage event to decide the best losers. It might be more interesting.

This is a good plan B/alternative, if Sullla has time for it and if nobody really "deserves" those wild cards.

[Sullla]

Answering questions here... Commodore, I don't recall specifically if Suryavarman built hospitals with his Medicine rush, but I'd tend to think that he did. The AI usually constructs every building possible in every city if they're not at war. Needless to say, it wasn't a very good plan.

We can figure out what to do about the wildcard leaders when we get there. My hope is that we'll "know them when we see them" from a game that ends with someone just barely missing out on the second spot. That said, it's also possible that we'll end up with no deserving civs, and maybe we will throw them all into a potluck game for the extra spots. (We also might end up with some games where only 2 civs survive to the end! You never know, Domination is far from impossible under these conditions.)

Opinion seems fairly split on drawing a full bracket versus going one game at a time, so let's stick with my original idea for now. We can always draw the remaining teams at once if we change our mind, we can't do it the other way around. Let's go ahead and announce the field for Game Two. This group is REALLY interesting.

[Sullla]

Hatshepsut

Hatshepsut reminds me of Suryavarman, in the sense that it's a very tough name to spell that I nonetheless have down by heart given her popularity for our Multiplayer events. Hatty is Spiritual and Creative for traits, and brings along the strong Egyptian package of the War Chariot and Obelisk. When using Restricted leaders, this is a very powerful combination. Hatshepsut the AI has a strong focus on culture, religion, and building wonders. Her low unit emphasis (only rated 2/10) can get her into trouble with the more aggressive AIs. Hatshepsut is below average in aggression rating at 3.7 out of 10, and makes for an excellent neighbor in Single Player games. She's a tough competitor if left unmolested by foreign attack.


Justinian

Here's the wildcard player for this particular round. Justinian has Spiritual and Imperialistic traits, and as the only Byzantine leader is renowned for bringing the mighty Cataphract as his unique unit. (He also has the unimpressive Hippodrome unique building.) We generally don't favor the Byzantines anymore due to their awful starting techs of Mysticism and The Wheel, but this doesn't matter for the Deity AIs, who all start with Agriculture, Wheel, Hunting, and Archery. Justinian AI has religious and military flavors, and heavily emphasizes building units (8/10 rating). He's well above average on the aggression rating at 7.6 out of 10, which wouldn't have stood out in the first game, but which sets him apart in this one. Justinian also has Theocracy as his favorite civic, which means he will never go to Free Religion. In short, Justinian is likely to found his own religion, and then hate everyone else who doesn't have it for the rest of the game. Expect him to dominate the game or get swiftly crushed depending on how diplomacy shakes out.


Mansa Musa

Mansa Musa is one of the game's economic powerhouses, with Spiritual and Financial traits. We're having a bit of a run on Spiritual in this game, apparently. His Malinese civ brings an excellent unit in the Skirmisher, and a decent building in the Mint. I once dubbed Mansa Musa as having the "winner" AI in the same sense as Tokugawa has the "loser" AI, since he favors trading and peaceful development. Mansa AI has gold and religious flavors, the latter of which you might not expect. He actually founds his own religion pretty frequently, despite not starting with Mysticism tech. Mansa Musa has a very low aggression rating of 1.6 out of 10; basically, this guy wants to be left alone to tech in peace. Mansa will be reigned in slightly by turning off tech trading (where he swaps techs like crazy), but he's still one to watch.


Mehmed

Mehmed is another excellent leader who has long been popular in our events. His trait pairing of Expansive and Organized gets him out to a fast start, and then keeps the surge going with cheap lighthouses, couthouses, and eventual factories. Mehmed's Ottoman civ brings along the Janissary unique unit and Hamman unique building, which is above average for a restricted leader pairing. As an AI, Mehmed is not what you would expect at all: he's actually very militaristic! Mehmed AI has military and culture flavors, along with the highest possible build unit emphasis (10/10) and an aggression rating of 7.8 out of 10. Don't be fooled by those builder traits, in other words. Mehmed will pump out a lot of units, and he's the most likely leader in this particular game to declare war. This could be a fox loose in the henyard for this match.


Franklin Roosevelt

Roosevelt brings along Industrious and Organized traits, another pair geared more towards economic development. He is unfortunately chained to the American civilization, which means the Navy SEAL and Mall unique items. Still, this game may go long enough for those to come into play at some point. Roosevelt AI has production and gold flavors. Despite his traits, he's not programmed to emphasize wonders that much (only 4/10), which is somewhat of a surprise. Roosevelt AI is quite peaceful overall, with a low focus on training units and an aggression rating of only 2.6 out of 10. He's even easy to get along with when sharing a different religion, as Roosevelt places little emphasis on faith. This guy is fairly uninteresting, I don't quite know what to make of him.


Willem van Oranje

You thought Mansa Musa and Mehmed were going to be the top economic leaders for this game, eh? Well as it turns out, Willem got randomly drawn into this game too! Willem has Creative and Financial traits, making him a near-universal ban for our Multiplayer events. He's held back somewhat by being paired with the Netherlands as a civ, who should have limited success on a Pangaea map. Their East Indiaman ship and Dike unique building are far stronger on water maps. Willem AI has gold and science flavors for his programming, and moderate emphasis on building units and wonders. In an interesting quirk, Willem cares almost nothing for religion at all, with a very small penalty for different faiths and equally small bonus for shared faith. He also has a strangely high aggression rating at 6.7 out of 10, which was not what I expected at all. So while Willem is programmed to emphasize gold and science, he's also programmed to declare war at a fairly high rate. Very unusual mix.


Zara Yacob

Zara is our last leader for this game, and yes, we drew the final two leaders in alphabetical order. Random numbers make for seemingly unusual results. Zara has Creative and Organized traits, bringing along the Oromo Warrior and Stele as unique stuff. I have a soft spot for Zara since he led us to victory in the Apolyton Demogame a number of years back. As an AI, Zara has growth and religious flavors. He sits in the middle of the table on most of the important numbers, like unit and wonder focus, and has a medium aggression rating of 5.6 out of 10. Zara AI also places a huge emphasis on religion, with a massive shared faith bonus (as much as +9 diplo points!) and a large penalty for differing faiths. Like Justinian, his favorite civic is also Theocracy. Zara AI therefore mostly follows a middle of the road path, but he carries a serious grudge over religion.

Overall, this group has a lot of economic leaders and strong builder traits. At the same time though, there's a bunch of leaders who are passionate about their religious faith, and eager to do nasty things to those who don't share their beliefs. I expect this game to break down along religious fault lines at some point, and those differences to drive conflict between the civs. Only two leaders can advance out of this group, barring a possible wildcard, which means we're going to lose someone strong here. Who will manage to survive and advance?

And since people requested it, our starting screenshot (spoilered for size, not extra info):

[Sullla]

Here's a standard form for predictions. Let's try to pick these categories:

Winning Leader
Second Place Leader
First One To Die
Total Leaders Alive at Game's End
Victory Type and Victory Date (in Turn number)

If there's enough interest, we can award points and keep a running score tally. I'll leave two days for people to make their picks, and post the game results on Wednesday evening (my time). We can tweak the format from there. Hope to see some entries, how's this all going to play out?

[Commodore]

Mansa first because Mansa.
Hatty, I like her odds.

Justinian to die first, because he's an angry man.
Five leaders alive.

Space win, turn...278.

[Mardoc]

Winning Leader: Definitely Mansa Musa. Strong trait, and nice tracts of land too.
Second Place Leader: Zara. Not in Mansa's way, not part of the mid-continent battlezone.
First One To Die : Gotta go with Washington for this role. Too many militarist neighbors.
Total Leaders Alive at Game's End: 4, with one down to a stub
Victory Type and Victory Date (in Turn number): Domination, 350

[Bobchillingworth]

Cool


Let's see...


Winning Leader: Mansa Musa, best land, good traits and uniques.

Second Place: Mehmed, the AI does well with his passive boosts (both traits and uniques) and he has a lot of land to settle. Too many forests, though.

First to Die: Zara, I'm guessing he winds up on the wrong end of a religious war. He also has a somewhat cramped start and too much jungle.

Total Leaders Alive: 5, the east is destined to be a bloodbath, but it's far enough away from likely runaway Mooseman that I suspect most AI will be merely crippled instead of outright destroyed.

Victory Type: Diplomatic, T257, the last game demonstrated fairly well that the AI has almost no sense of self-preservation when it comes to voting, and it's a lot faster to build the UN than all the spaceship parts.

[von Adlercreutz]

Mansa first, he seems to get a lot of land peacefully here (depends on where Dutch and Ottomans expand but still), which is exactly what he needs.
Mehmed second, though he has a forest start if he can get Roosy or Zara early enough he should be strong. Also has elephants. And I root for warmongers...
Roosevelt first to die, he is between two civs with high unit prob and has low prob himself, enough said.
4 leaders alive.
Culture win T325 (How close the guessed turn number needs to be in order to be counted as "right" guess?)

[Brian Shanahan]

Winning Leader: Mansa
Second Place Leader: Willem
First One To Die Roosy
Total Leaders Alive at Game's End 4
Victory Type and Victory Date (in Turn number) Ask me after the game.