Hello!
As the title suggests, I’ve been playing some ranked games and sometimes I notice that I get the same civilization twice.
Of course this could happen eventually but what really got me questioning if “random civ” was truly random was when I got Incas THREE (not 2!) THREE TIMES IN A ROW.
I wanted to ask if you guys are experiencing the same behaviour when choosing random civs to play games.
how many games did you play in total with random civ enabled? stuff like this is more likely to happen than people expect, human intuition is really bad at probabilities
Yesterday I played a 2v2 with a friend vs AI random civs. The first game we played vs chinese and goths. The second game was vs chinese and goths.
I figured it was just odd chances but if others are seeing this then maybe there is something.
Around 200 ranked games (all with random civ on). Also got repeated civs twice a bunch of times too.
in 200 games there is a ~10% chance of getting the same civ 3 (or more) times in a row on at least one occasion. this is with 42 civs, if you don’t have any DLC (35 civs) then it’s ~15%
i wrote a little console app that calculates this probability because of a similar discussion a while ago
This is pretty much confirmation bias. And OP is looking for more people with confirmation bias.
Just to backup @TwerDefender you can model this as a markov chain and verify that this number is correct using wolfram alpha or something similar. Check the top-right number in the result matrix.
always good to see someone double checking, so easy to have a tiny mistake somewhere
After putting the numbers into perspective, guess I had a considerable chance at getting the same civ 3 times in a row. Thanks!
Also, I didn’t take into account the amount of times I’ve played singleplayer with the random civ selected.
I’ve committed a mistake taking an isolated scenario without taking into account the amount of matches I’ve played. But I’m willing to keep testing the civ random option (this time with proper data).
What do you (and others) think the best way to test the random civ option behaviour?
I was thinking on the following:
i think you need a bigger sample size. Each civ has a (non-independent) ~1% chance of not occuring at all, so chances of only occuring once or twice are pretty high. So I think while the methodology is good, it will be hard to gain any meaningful results from this.
On average each civ will occur about 5 times, but a few only occuring twice wouldn’t mean anything. So unless something drastic, like one civ occuring 15 times, happens it won’t be super insightful I think
Ofc other people see patterns too, our brain is very good at seeing patterns…do you have an idea how many matches are played daily? If everyone who sees repeating civs opens a forum thread, there will be a lot to read - and yet nothing out of the ordinairy.