How are random civs/leaders selected?

[Mercenary82]

Does anyone know if when starting a game, all leaders are chosen at random first? Or does the game choose civilizations first, then randomly assign leaders? The reason I'm asking is it seems like leaders who are the only leader for their civ seem to show up the most.

 

[Bushface]

Surely the key word is "random", although I believe (from results shown in the Combat Log) that the RNG in the game is streaky and not truly random. You'd have to start very many games to get a reasonably good idea as to whether the one-leader civs really do come up more often than the others, but remember there are 14 civs with one leader, as compared to 8 with two and 2 with three.
Come to think of it, in all the games I've played I have never yet met Ramesses or Stalin, and Shaka only once - which I remember because the poor soul was stuck on a waterless narrow peninsula cut off by mountains: worst start I've seen.

 

[LucyDuke]

I believe (from results shown in the Combat Log) that the RNG in the game is streaky and not truly random.

Keep in mind that truly random series are streaky. A great way to distinguish between a real and faked random series is that the real one has streaks.

 

[Gaius Octavius]

Come to think of it, in all the games I've played I have never yet met Ramesses or Stalin, and Shaka only once.

That's funny, I meet these guys all the time (not Stalin so much, but I get sick of Shaka.)

Just chalk it up to the RNG. Sometimes you meet certain ones more often, sometimes not. It does get a bit annoying after a while, though.

 

[noto2]

I'm sorry to necro this but I just started a thread on the exact same topic and did a forum search. I believe that OP is correct. I have found that the single leader civs pop up very frequently in the so called "random" leader selection, and I have a sample size that is not too small, 240 games. Zara, Sury, Wang make up almost 25% of my games, other common leaders are Pacal, Isabella, Charlemagne, Toku. As for civs with multiple leaders... out of my last 240 games I got Churchill twice, not Vicky or Lizzie, I got Napoleon once, I got Frederick once. Compare that to 26 games for Sury alone.

I really don't think it's random. I can't tell you how many times I've gone weeks without playing Civ, and then sat down to have a weekend of Civ gaming, and played more than one Sury game in a weekend. Sury, Zara, and Wang just seem to my leader in soooo many of my games.

 

[Manco Capac]

You necrophile!

============

Why several leaders appear more often than others? Because the rand.generator chooses amongst the available civs in the first step then chooses the leader with another random roll. But given several civilizations are restricted to few leaders up to only one, that two step RNG will lead you to skewed appearances of those recurrent leaders like Izzy, Wang Kon, Zara, etc.

 

[The Diocletian]

Sury, Zara, and Wang just seem to my leader in soooo many of my games.

This. It just gets plain annoying to see these leaders game after game.

Maybe I'm just having selective memory.

 

[noto2]

Tachy - I completely agree, that must be what the RNG is doing, therefore under representing the civs with multiple leader.

 

[gps]

therefore under representing the civs with multiple leader.

Not the Civ, only the leaders. If the random selection were based on the leaders only, the probability to have England, US or Russia in the game would be three times as high as the one leader nations.

 

[J@ckRB]

If 'Random' RNG find the actual civ first, then England should have (mathmatically) the same chance as being picked as, say Spain.
However, England have 3 leaders (Churchill, Lizzy + Viccy) whereas Spain has 1 leader (Izzy).

So, while the civilizations should be represented equally, 1 civ leaders are over-represented whereas 2/3 civ leaders are under-represented as a consequence.

Irritating, but that's how the 'RNG' is set up to be. If it chose the leader, and then found the civilization from the leader, then that (mathmatically) would represent all the leaders equally, but still would over-represent civilizations instead.

Question is, would you rather have over + under-represented leaders or civilizations?

 

[noto2]

Well at this point I'd rather it be based on leaders. I start about 2 games per week and I have not played as Elizabeth in years, nor Victoria, nor Bismark... I've played as Sury, Zara, Hannibal, and Wang, every single month.

 

[JujuLautre]

I did a quick sampling (with some approximation, considering only one leader per civ), and for 240 games, the top 3 leaders you see most will on average appear for a total of 36 times, that is 15% of all games, with the top leader appearing on average 13.2 times. Consider this is on average, so although we have a relatively high samples (240 games), I wouldn't bee surprised to see 3 most present leaders for close to 25% of the games.


Actually, the chance of having "Sury, Wang or Zara" for close to 25% of games is indeed quite small, but I wonder if someone tried computing an estimation of the percentage of N game sequences for which some human would see a pattern. I wouldn't be surprised if the number is high.

 

[plasmacannon]

It seemed to me that if I chose a certain leader, that their traits were usually represented in at least one other leader in the game. As if the game was trying to counter any advantage I wanted to gain from that leader that game.
So, if I pick Alex [AGG/PHI], sure enough another AGG (or in many cases PRO) would start right beside me ruining my AGG rush.
Try it sometime. Choose an EXP leader, and sure enough there will be an EXP leader in your game. In some cases IMP.
Want to choose IND and grab wonders? Start a game then count how many IND leaders miraculously showed up that game.

 

[noto2]

Anyway, it seems we've figured out what is going on - that the game rolls random civs and not random leaders, thus balancing out the civs generated but not the leaders, hence the leaders of solo leader civs show up relatively more often.

 

[gps]

Choose an EXP leader, and sure enough there will be an EXP leader in your game.

But that's to be expected. With eight traits available in an eight players game (with two slots each) the probability for any of the traits to be in the game ist about 2 to 1.

 

[Lennier]

And gets worse if you're playing huge (11-civ) maps.