Are some AI civs more random than others?

[jerVL/kg]

After a long extended break, I started my 5th game of Civ4 last night, and my first on Prince. I don't think I've met all the AI civs yet, but already I'm seeing many familiar faces. So far my games have been like this:

1. Napoleon, Chieftan: Hatty, Mansa, Montezuma, Asoka, Cyrus, Bismark, Washington, Kublai Khan.

2. Saladin, Warlord: Hatty, Mansa, Montezuma, Cyrus, Catherine, Louis XIV, Julius Caesar, Isabella, Qin Shi Huang, Roosevelt.

3. Huayna Capac, Noble: Hatty, Mansa, Asoka, Catherine, Louis XIV, Julius Caesar, Bismark, Alexander, Tokugawa, Genghis Khan.

4. Catherine, Noble: Hatty, Mansa, Montezuma, Isabella, Qin Shi Huang.

5. Catherine, Prince: Hatty, Mansa, Montezuma, Cyrus, Julius Caesar, Victoria, a German civ (dunno which one yet), plus 1 or 2 unknowns.

I can't help but notice that Hatty & Mansa have shown up in every game, Monty's been in 4 out of 5, Cyrus & Julius 3 of 5, and Catherine's been in 2/3 of the games where I didn't play Catherine. At the same time, a good number of civs have only shown up once or twice, and several I've yet to play against at all.

Is this just screwy random luck?? Or do Hatty/Mansa in fact show up far more often than the rest? Are the types of AI you get based on difficulty level?

I'm interested in hearing about the "usual suspects" in other people's games.

 

[Xanikk999]

Hati and Mansa are in my current game although hati was wiped out in ancient times by the AI.

Sometimes i wish mansa wasnt in my games. On prince and above he almost always has a tech lead over you. I have almost three times as many cities with a hell of lot of matured cottages ( All towns). And he still runs away with the tech lead. I dont get it. I had to pillage all his towns just to put him back in his place.

 

[pholkhero]

some civs DEFINITELY occur more than others...just as when you select random civ for your own civ, you get a few commonly recurring characters...it makes random settings somewhat lame....i was turned onto random.org in order to make things a little more random...just 'roll' up some numbers and away you go...

 

[jerVL/kg]

Are some civs more frequent on higher difficulty levels? I noticed that Toku & Alex didn't show up until I played on Noble, and England never showed up until Prince.

 

[voek]

i am guessing the same thing, but haven't got the same proof. More annoying, i.m.o. is the seemingly superior financial opponents (eventually). It would be nice if every civ would go for his own best victory type, given the situation it is. Now i think the financial civs just run for space race....

 

[Carcosa]

I have had Mansa Musa in every game of Civ4 I've played.

Every single game.

 

[Civious]

I've played 4 games. Catherine, Caesar and Monte have been in every one. Mansa was in 3 of the 4. I've never seen France, but am playing as Nap in my current game.

I saw England - Victoria in my very first game (settler).

 

[MrCynical]

I'm wondering if there's a slight quirk in the way it's designed that means it's more likely to pick a leader if they're the only one for that civ, than if they're one of two possible leaders. I also seem to see a lot of Mansa Musa and Montezuma. On the other hand in over 20 games I have never seen Kublai Khan or Asoka. Genghis and Ghandi have shown up a few times but the alternative leaders have been completely missing.

 

[katank]

Mansa, Monty, and Isabella seems to show up a lot in my games.

Some AI's do definitely seem to be in more than others.

 

[Carcosa]

I'm wondering if there's a slight quirk in the way it's designed that means it's more likely to pick a leader if they're the only one for that civ, than if they're one of two possible leaders.

I've never had Huayna Capac in a game, and I've had Kublai Khan show up more often than Genghis.

Considering all our different experiences, I'd say things are probably actually working quite randomly... except for Mansa Musa. That guy stalks me and it's creepy.

 

[voek]

Maybe there is a greater chance for a particular civ when playing civ X. For example, playing as Inca, gives you a greater chance for meeting England and Mongols (example). Just to balance the type of game (not only tech'ers or warmongerers).

 

[Rex Tyrannus]

I agree partly with voek. Not the "playing with civ X" part. More like playing with trait X. I think the game wants to keep a certain mix of traits.

That being said, I keep my leader random, but seem to get the same group of Catherine, Saladin, Qin, Washington over and over again. Never do I end up as an aggressive leader, when I set it to random.

Also, I nearly always meet Hatepshut, HC, Mansa, and Ceasar. Kublah shows up often (thank God, since the alternative is his grandpa). Mao is a regular.

On the other side, I've only seen Frederick once.

Go figure.

 

[voek]

Agreed with the trait-part. I meant the same but my formulation was not specific.

 

[KrikkitTwo]

I'd guess there probably isn't any difference, given the sample sizes are so small (if a few 100 games were tabulated...then we might have something)

There is the issue of whether the Civs or the leaders are what is at equal frequency, because if it randomly selected Civs first and THEN randomly selected leaders, Hatty, Caesar, Monty, HC, Isabella, MM, etc, should be more common than Washington, Napoleon, Catherine, Genghis, etc.

 

[naterator]

at one point i swore mansa was following me, then it was isabella, right now it's ceasar, so i pray for ivory, but anyway, maybe it goes in streaks, some kind of glitch where if a leader was in the last game, he's more likely to be in the next... but i suspect it's random.
Krikkitone's question of the chicken and the egg is a great one, because if it chooses leaders only, then civs with 2 leaders would occur more, and if it chooses civ then leader, then leaders who don't "share" their civ would hold an edge.

 

[Fetch]

Judging by all the "I've had .... in my games" the appearance seems to be random. All the AI's names have come up, it seems. Remeber that randomness is, by definition, not an even distribution. Therefore, if you are seeing the same faces for several games in a row, that's a normal anomoly. If you play 1000 games, you've got enough data to determine if the AIs are random or not... 5 games isn't enough samples.

It would also be very easy to implement a random AI leader in the code for the game. Making certain leader show up more on certain difficulties or victory conditions, or whatever, would be a bit more involved, with no real benefit.

 

[Dominus Belli]

I tend to agree with Fetch's point about the relatively low sample size for each person, but I also am intrigued by the argument that games are generated with certain mixes of traits to keep gameplay balanced. Perhaps it could be a combination of both? For example, if I choose Cyrus or Saladin, AI civs could selected from warmonger leaders. Personally, I have frequently seen Tokugawa and Monty, and I tend to take leaders with traits more geared toward building rather than warring.

 

[voek]

Judging by all the "I've had .... in my games" the appearance seems to be random. All the AI's names have come up, it seems. Remeber that randomness is, by definition, not an even distribution. Therefore, if you are seeing the same faces for several games in a row, that's a normal anomoly. If you play 1000 games, you've got enough data to determine if the AIs are random or not... 5 games isn't enough samples.

It would also be very easy to implement a random AI leader in the code for the game. Making certain leader show up more on certain difficulties or victory conditions, or whatever, would be a bit more involved, with no real benefit.

You can't speak of the amount of samples needed if you don't know the population size, although 1000 would be quit a lot. Most arguments made in these forums are indications not significant proven statements... Maybe there is a real nutcase who wants to spent time testing 1000 games?

P.s. the argument of all the random leaders named in these forums doesnt make it random, because a lot of people have a few favourable leaders/civs.

 

[Fetch]

You can't speak of the amount of samples needed if you don't know the population size, although 1000 would be quit a lot. Most arguments made in these forums are indications not significant proven statements... Maybe there is a real nutcase who wants to spent time testing 1000 games?

P.s. the argument of all the random leaders named in these forums doesnt make it random, because a lot of people have a few favourable leaders/civs.

(Trying to avoid talking about statistics on a Civ IV board) With 1000 (read: "a lot of") samples you could see if there was relative parity between all the civ leaders-- playing 5 games doesn't give you any indication of that. I'm not talking true-blue stats. As for your PS, I'm not sure what you're getting at.

After rethinking what i said above, I can see why the designers would purposefully put a "mix" of AI traits on a map together... keeps it interesting.

 

[voek]

After rethinking what i said above, I can see why the designers would purposefully put a "mix" of AI traits on a map together... keeps it interesting.

I agree. I assume there must be some coding to be found concerning this 'randomness'...