How random is Random AI civ selection?

[Badtz Maru]

That statement is wrong.

I may be remembering something wrong, but there was grouping of civilizations and I was pretty sure it was based on traits, though maybe it was starting techs. It's been a LONG time since I played Civ3...maybe it wasn't matched because of similarity, but because of differences, so that civs that had same traits (or techs) were not next to each other often.

Does anyone else know what I'm talking about? I can't remember the details but I am certain there was civilization grouping in Civ3.

**FOUND IT**
From the Civ3 FAQ thread:

What are 'culturally linked starting positions'?
It means that civs in the same game will tend to start near civs that share the same culture grouping as themselves. (I.e., the Germans start near the French, and so on)
this is straight from Dan @ Firaxis, so you better check twice before you doubt it! Thanx Dan!

http://forums.civfanatics.com/showthread.php?t=23341

 

[TheMeInTeam]

THE BURDEN OF PROOF!!!!!! YOU ARE A LAWYER, ARENT YOU?

No, my training is in management and accounting. However, the latter should really be considered the same thing . Accounting involves little more than basic arithmetic and on occasion minor algebra...and a whole hell of a lot of applied law .

 

[Kullervo]

It may be that I am terribly mistaken, but I think that if you choose a particular civ but assign random opponents, there IS a pattern. Whenever I play as Hatshepsut (arguably the worst civ in the game for war) I tend to either start next to neighbors like Monty or Nappy or Toku, or there are insane teching leaders like Gandhi (who always claims the great library and the pyramids, which I want VERY bad) or Mansa. When I start as a spiritual civ, I tend to have opponents that either build wonders very fast or wage war quite a bit. I think... and I may be wrong... that there is something in the game engine that assigns opponents that have some sort of advantage over you to even things out.

Maybe it's just me, but when I play as a random civ, I end up as the Romans or the Americans 80% of the time.

 

[thekaje]

In my experience, anecdotal evidence is the best kind.

 

[Öjevind Lång]

I may be remembering something wrong, but there was grouping of civilizations and I was pretty sure it was based on traits, though maybe it was starting techs. It's been a LONG time since I played Civ3...maybe it wasn't matched because of similarity, but because of differences, so that civs that had same traits (or techs) were not next to each other often.

Does anyone else know what I'm talking about? I can't remember the details but I am certain there was civilization grouping in Civ3.

**FOUND IT**
From the Civ3 FAQ thread:



http://forums.civfanatics.com/showthread.php?t=23341

Yes, but that was an option one could choose before starting a game, wasn't it? I don't remember clearly, though; I haven't played Civ III for a very long time.

 

[sourdiesel]

You know, that is a really great story.

Your posts are the best.

 

[Öjevind Lång]

Your posts are the best.

I'm sorry if you two found my little reflection on being Sitting Bull, exterminating Washington and renaming his capital for myself so boring. I thought it was somewhat amusing. Goes to show, doesn't it?

 

[thekaje]

No, it was a great story. It was just a non sequitur.

 

[budweiser]

I do not believe it is totally random. You would get a lot more blow out games if it were. I believe they throw in foils and group leaders against you based on a tier system. Because Civ III did it, it's probable they did it here too. Depending on the civ I choose, I will often get some heavy weight competiton for neighbors.

The version you play may also be a factor.

 

[Continental Op]

I think (and I'm not arguing against the fact that there is a difference between random number generators) that there is no intentional pattern going on around here. Some of us will never see Winston Churchill, some us (like me) get him all the time. Some of us will never see Hatty or Shaka, while some of us will get them both in every game. Some of us will probably win a lottery, while most of us never will. I've even heard that somewhere in the world somebody won the main price in the lottery twice, so sometimes the lightning can strike twice in the same place.

It's the same thing as when you are claiming that the fight odds are biased towards the computer. They really are not. Just today I losed six times in a row with over 75% odds, got annoyed, and attacked the AI with lots of less odds. And I won three times in a row with less than 30% odds. That's how randomness works. I guess that's why they call it random.

 

[Kullervo]

That's how randomness works. I guess that's why they call it random.

Agreed. But I think the question here is: IS it all this random OR is there a pattern to civ selection (despite the fact that the settings are called random). As I posted before, I think it's not as random as it seems.

 

[thekaje]

What if they DIDN'T call it "random"? Wouldn't that be confusing? What if they called it "yellow"! I'd be completely at a loss!

 

[Kullervo]

Not quite certain what you mean...

 

[Badtz Maru]

I think (and I'm not arguing against the fact that there is a difference between random number generators) that there is no intentional pattern going on around here. Some of us will never see Winston Churchill, some us (like me) get him all the time. Some of us will never see Hatty or Shaka, while some of us will get them both in every game. Some of us will probably win a lottery, while most of us never will. I've even heard that somewhere in the world somebody won the main price in the lottery twice, so sometimes the lightning can strike twice in the same place.

It's the same thing as when you are claiming that the fight odds are biased towards the computer. They really are not. Just today I losed six times in a row with over 75% odds, got annoyed, and attacked the AI with lots of less odds. And I won three times in a row with less than 30% odds. That's how randomness works. I guess that's why they call it random.

I think there may be some unintentional patterns to explain people's varying experiences. It may be that, under certain circumstances, certain civilizations are more likely to be chosen as opponents, and because different people play with different settings they may get different non-random patterns. If a certain civilization is much more likely to appear if you are playing as a certain leader on a small map with 4 AI civs, it's not going to be reported very often, but some people may run into it all the time.

For instance, I could see how certain possible methods of "randomly" selecting civilizations could make some other civs more or less likely to appear if you usually select the same leader, or the same number of civilizations, or certain map types. There may also be parts of the opponent selection programming that were added intentionally during development, decided against in the favor of more randomness, and accidentally left in or re-introduced in patching. Maybe there was something like the Civ3 grouping of civilizations that was decided against, but not removed completely. Since there is still some randomness, and it's definitely not a game-breaking issue, it could be left unaddressed.

 

[Lone Wolf]

The government has paid Firaxis to select "random" civilizations non-randomly to form subconscious patterns in our brains. To be free from the governmental mind manipulation, always select the civs yourself.

 

[Continental Op]

The government has paid Firaxis to select "random" civilizations non-randomly to form subconscious patterns in our brains. To be free from the governmental mind manipulation, always select the civs yourself.

Exactly. I mean, I get it that there may be some unintentional patterns going on due to bad random number generation, but I don't get that the developers would intentionally create a pattern and then lie about it to the gamers. What would be the point in that?

 

[Kullervo]

Possibly they have a different idea of randomness? Or it's conditional randomness? No one can read ALL of the fine print in a document.

 

[LegionSteve]

Possibly they have a different idea of randomness? Or it's conditional randomness? No one can read ALL of the fine print in a document.

No, but a lot of civfanatics who understand the code have looked at many aspects of the code, and if there was something other than the RNG choosing the Civs they would have pointed it out. When you look at some of the other things that have been spotted and reported to Firaxis it seems pretty unlikely that something like this would go unnoticed.

 

[CHEESE!]

What if they DIDN'T call it "random"? Wouldn't that be confusing? What if they called it "yellow"! I'd be completely at a loss!

'How yellow is Yellow AI civ selection?'

 

[Badtz Maru]

No, but a lot of civfanatics who understand the code have looked at many aspects of the code, and if there was something other than the RNG choosing the Civs they would have pointed it out. When you look at some of the other things that have been spotted and reported to Firaxis it seems pretty unlikely that something like this would go unnoticed.

The source code hasn't been released, has it? Do you have any links that show the details of how the game selects civilizations?