"Random" Civs don't seem so random...

Just an observation. When you play the game on the default settings; the Civs are seemingly random. The locations are supposedly random too.

But I've started a few games where Rome was around the exact same location every time.

Is this a bug or just his start bias or something?

Details.

Screenshots.

What do you mean by same location?

Probably start bias because areas of the map scripts putting terrain types in certain areas.


No, warmongers being next to you is not an auto-reroll. Shaka/Japan/Germany is not a OP neighbour...

I know that the game loves picking the same few civs out of a pool of 100+ ones with mods for me.

I seem to get Ethiopia, Poland and Polynesia every single game on a large continents map.

I went through a really long phase where Monty was always my next door neighbour, but I think he eventually got fed up of me kicking his ass!
Seriously, I think it IS random, but I also think the RNG gets stuck from time to time...so you might call it random-ish!

First, it's pseudo-random, so depending on how their algorithm is set up, they could get much more rediculous clustering than should really ever happen in a true random output. This clustering would translate to certain civs being selected more often.

Second, I would assume it partially depends on who you're selecting as your leader as well; since with location bias your civs bias could be very near another civs bias, making you seem them more often.

Or maybe the designers put in some troll % variable, where if the RNG rolls well on your game as Venice, you'll see Austria, Mongolia, Greece, Siam, and the Huns. I actually am kinda leaning towards this being in there, since every time I pick a diplo heavy civ I get at least 2-3 of the people from that group.

You can use this randomizer and manually input the civs into advanced setup. : http://georgeskleres.com/civ5/

I think the pseudo-random number generator is probably pretty good (since they've been used in lots of applications in lots of programming languages for an awfully long time). As a result, I think that this is simply an issue of perception, and an underestimation of how often you would (statistically speaking) expect to see any given civilization in a game.

Assume that you have the complete set of 43 civilizations and play standard size games with eight players (you and seven AI). Then the probability of seeing civilization X in a game is (42-choose-6)/(42-choose-7), which is about 19.4% of the time, or roughly one in five. (The probability is higher if you own fewer civs.) So there is already about a 4% chance (0.2 * 0.2 = 0.04) that any given civilization will appear twice in a row.

But it is much more likely that two successive games will have at least one AI in common. For, the probability that none of the seven AI in game 1 are included in game 2 is (35-choose-7)/(42-choose-7), which is about 25% of the time. So, 75% of the time, two successive games will include at least one AI in both of them. And at least 30% of the time, there will be two or more of the same AIs included in two successive games.

Sometimes random events aren't as rare as we think they are. (And the basic idea behind this computation is the same as the one that shows that in any group of 30 people, there is a better than even chance that two of them have the same birthday.)

Nice analysis krc. It really is just confirmation bias making random civs not seem so random. Four games in a row where none of the civs repeated would feel truly random, but in fact would be almost a statistical miracle. That said, I wish there was an option to pull in civs round-robin style, or to pick your opponents based on map features or what civ the player picked. I have gotten in the habit of hand-picking opponents to provide a challenge. It takes away an element of surprise and discovery, but the games feel more consistent. (But that could be confirmation bias too!)