AI Leaders Randomly Selected or...?

[SirWill90]

Granted I tend to play as the same several leaders lately, i.e. Frederick, Victoria, and Peter, but are the AI opponents truly random, or are certain AI leaders more likely to oppose a human playing a certain leader?

I ask because it seems that in every single game I am opposed by Montezuma and Mbzemba or whatever, of Kongo. I find them both more annoying than the other civs (Monty is always angry bc he has fewer luxuries, Kongo guy same but bc I haven't sent missionaries) and maybe the game senses this bc I'll be damned if they aren't there every single game.

Brazil's leader is a close third although he is sometimes not present.

The other leaders seem truly random, but what is it with these three?

I know I can customize who is in the game but that takes some of the fun out of it. Too bad there isn't an option to exclude certain leaders.

Anyway... random or by design?

 

[Uberfrog]

It's random. The fact that you find Monty and Mvemba most annoying means you are more likely to remember when you play against them. Just confirmation bias, really. They might really have been in every game, but that's a possibility when the leader selection is random.

 

[Josephias]

This is more of a suggestion, but i think it fits in this thread - one thing I think would be worth to implement is a way, in advanced game config. options to ajdust random options: this is, that when I set a "adjusted random" leader it is random but from within a group of leaders I previously selected.

This way you could do very different and interesting things like an all-female leaders game, but without pre-selecting each player, or maybe a "balanced" TSL Earth game by forcing each one of your rivals to be from one contintent, but without knowing which European leader you'll find, which Middle East leadr you'll find, or which Asian leader you'll find, etc...

Or in the case of the OP, he could "ban" Monty and Nzinga from its games, while keeping the selection of leaders random.


BTW (this is completely of topic, but nevertheless... now we are talking of game setup options - ¿is there a possibility to change leader name hidden somewere?).

 

[SirWill90]

I seem to remember in early (perhaps also later?) builds of Civ IV there was a way to reset random seed or something, that had to be done periodically. One held down certain keys as one started the game, I seem to remember. Anything like that in VI?

 

[Rosty K]

Well, 'random' means 'random', not 'making sure it's completely different every time', so that explains it. For example I didn't get Pericles as an opponent in my first 15 or so games... And I play with 12 civs.

 

[SirWill90]

Well, 'random' means 'random', not 'making sure it's completely different every time', so that explains it. For example I didn't get Pericles as an opponent in my first 15 or so games... And I play with 12 civs.

Thanks for the vocabulary lesson. For me, when I consistently get the same two supposedly random AI opponents for many consecutive games, it begs the question, How random is random here?

 

[Uberfrog]

Thanks for the vocabulary lesson. For me, when I consistently get the same two supposedly random AI opponents for many consecutive games, it begs the question, How random is random here?

You and many other people.

These kinds of threads come up a lot, and not just about Civ either, you see it with people suspecting their iTunes Shuffle isn't really random either.

The human brain is designed to look for patterns. Randomness is by definition the absence of a pattern. It is confusing for our brains to picture true randomness so when most people hear that something is supposed to be random, they expect the closest approximation: that the selection will be different every time. But that's not what random means.

On the scale of the number of games you've played, the chance of getting one or two leaders almost every time is finite and not that unusual.

You'd need to play hundreds (if not thousands) of games and record all the leaders you see before you could start to judge whether you're seeing anyone significantly more frequently than anyone else.

 

[Rosty K]

Thanks for the vocabulary lesson. For me, when I consistently get the same two supposedly random AI opponents for many consecutive games, it begs the question, How random is random here?

It wasn't a vocabulary lesson. It was a probability theory lesson actually

I'll rephrase (and it's also quite well put in the post above):

Random number generator pulls numbers from a distribution. The only requirement it follows is that for a sufficiently large number of games you start (that's at least a few thousand) each opponent will pop up approximately equal number of times. There's no requirement to the order in which they do it, so certain combinations can and will appear consecutively. And here applies the remark about the human brain and patterns

Having said that, there may be other things affecting who appears on the map (don't cite me on this, I have no idea). E.g. what are the chances of finding Norway/England on a watery map, or getting a watery map when playing as Norway/England on a 'shuffle'?

 

[antimony]

Just for fun, I ran some numbers into a statistical software. Say you're playing 10 games on a standard map, each time the game picks 7 random opponents out of 18 (no DLC). So your 'average' opponent appears in about 4 games.

- You have about 8/9 chances of seeing all the opponent leaders (1/9 of missing 1, negligible chance of missing 2).
- Your most common opponent will be present in 6 to 8 games out of 10 (96% of simulations are in that range).
- You'll get between 1 and 4 opponents coming back in at least 6 games (most likely 2 or 3, 20% chance of 4).

Well, 'random' means 'random', not 'making sure it's completely different every time', so that explains it. For example I didn't get Pericles as an opponent in my first 15 or so games... And I play with 12 civs.

This seems extremely improbable if Pericles is as likely as any other leader.

It makes me wonder if the system first picks opponent Civs, and in the case of Greece, picks one out of two leaders. Has anyone ever seen both Pericles and Gorgo in the same random game?

 

[Rosty K]

As far as I understand, Pericles and Gorgo can't appear together as random leaders, so essentially our statistical considerations should be about civilizations, not leaders.

 

[Uberfrog]

In the very first game I played, I was Pericles, and Gorgo was an AI. Unless they changed something after the first patch, I think it's quite possible to have two Greeces in one game.

 

[kaspergm]

In the very first game I played, I was Pericles, and Gorgo was an AI. Unless they changed something after the first patch, I think it's quite possible to have two Greeces in one game.

They did change that in a patch, however. Now you can choose in advanced settings whether it's allowed, iinm.

 

[Hep Roc Heretic]

It's random. The fact that you find Monty and Mvemba most annoying means you are more likely to remember when you play against them. Just confirmation bias, really. They might really have been in every game, but that's a possibility when the leader selection is random.

In seven games, I have already played against everyone at least once except for Gorgo. Yeah, the games with Monty and Mvemba stand out the most because they are annoying!

I used to choose "shuffle" for map type to start a game, but it always seemed to create a Pangaea map, which is my least favorite. This last game, I chose fractal, and what did I get? Yeah, another Pangaea-looking map.

 

[Josephias]

In seven games, I have already played against everyone at least once except for Gorgo. Yeah, the games with Monty and Mvemba stand out the most because they are annoying!

I used to choose "shuffle" for map type to start a game, but it always seemed to create a Pangaea map, which is my least favorite. This last game, I chose fractal, and what did I get? Yeah, another Pangaea-looking map.

And continuing the theory talk, you're describing an example of perception Bias (which can be the case as well for the OP)

If you least like Pangea, then any other maps are fine, you'll remember the most the Pangea cases, because they stand out in your mind as remarkable (bad) experiences. Same for finding Monty or Mbemba "another" time. (you do count how many times you saw these two, however don't care on how many times did Cleo or Quin appear).

 

[DizzKneeLand33]

It's not just confirmation bias, though. The game is "sticky" with some of the RNG, and this has been true throughout the Civ series. Not a big deal, but to assume that numbers are truly random is as bad as saying they are not random at all. This goes for other games as well (omg the CD version of Axis & Allies comes to mind -- the debates were endless).

But, it is also true that it is almost as non-random to have every civ exactly evenly distributed throughout your Civ as it is to have the same civs over and over.

Let's just say that as I've been doing some testing (playing 50-100 turns and restarting) a lot that it is amazing how many times the *same* civs show up in similar locations if I remain the same civ. So, it's not 100% random is my guess. Start bias plays some kind of role in all of this, and if it plays *any* role then it is not 100% "random."

 

[Josephias]

Let's just say that as I've been doing some testing (playing 50-100 turns and restarting) a lot that it is amazing how many times the *same* civs show up in similar locations if I remain the same civ. So, it's not 100% random is my guess. Start bias plays some kind of role in all of this, and if it plays *any* role then it is not 100% "random."

This might have some logic, both start bias and agenda types having a play. If the game "random" algorithm is made in a way it balances start location bias and agenda types (e.g. as much aggresive as peaceful players), you may trully find to be getting more times leaders with opposite agenda / start bias , and less times leaders with similar agenda / start bias.

Other thing you mention as well, the rng seed sems to "move" slowly. I start normally random, then skip the leader if I'm not feeling to play it, but often when I skip the first it turns appearing again two or three times until I decide to start the game (and It's not that I take 20 restarts before I decide I want to start playing).

 

[Hep Roc Heretic]

And continuing the theory talk, you're describing an example of perception Bias (which can be the case as well for the OP)

If you least like Pangea, then any other maps are fine, you'll remember the most the Pangea cases, because they stand out in your mind as remarkable (bad) experiences. Same for finding Monty or Mbemba "another" time. (you do count how many times you saw these two, however don't care on how many times did Cleo or Quin appear).

I played against Cleo in my first game and haven't seen her since (was it something I said?).

 

[Ogro]

A solutions would be to use a QRGB random generator to create a random set for your game.

1- You create a list like: 1-Arabia, 2-France, 3-Brazil,...,etc.
2-You go in a site like this: https://www.random.org/integer-sets/
3-Configure it to generate 1 set with the unique random integer quantity equal to the numbers of civ you want in your game.
4-Configure the range of each number to be within the range of the list in step 1
5-Click get sets to get the numbers then manually add the civs in your game based on your list.

It is an extra trouble but if the random aspect bothers you could be worth while.

 

[Uberfrog]

It's not just confirmation bias, though. The game is "sticky" with some of the RNG, and this has been true throughout the Civ series. Not a big deal, but to assume that numbers are truly random is as bad as saying they are not random at all. This goes for other games as well (omg the CD version of Axis & Allies comes to mind -- the debates were endless).

But, it is also true that it is almost as non-random to have every civ exactly evenly distributed throughout your Civ as it is to have the same civs over and over.

Let's just say that as I've been doing some testing (playing 50-100 turns and restarting) a lot that it is amazing how many times the *same* civs show up in similar locations if I remain the same civ. So, it's not 100% random is my guess. Start bias plays some kind of role in all of this, and if it plays *any* role then it is not 100% "random."

I'm willing to believe there could be irregularities with the RNG. There is no such thing as a "truly random" number generator after all.

Start bias may introduce a non-random factor, but I was under the impression that the field of civs in the game was chosen before the map was generated (as it would be if you chose every opponent manually). Start bias would, I imagine, only come into play as to how the civs are distributed on the map, not who is selected in the first place. So if you play the same civ, start bias will affect who you see as your immediate neighbour, sure. But I don't believe it affects whether that leader was going to appear in the game to begin with.

I really don't think Firaxis have complicated the leader selection algorithm by pitting different agendas against each other. It is much easier to have a pseudo-RNG, and that's what everyone is expecting, so why fiddle with it?

I understand people's frustrations but we should assume the leader selection is random, until we see some non-anecdotal evidence that it is not. As @antimony's experiment before showed, the number of available leaders (22 with DLC) vs the number of player slots (10-12 for a standard to large map) is limited enough that you really would expect to see multiple leaders cropping up again and again even in a perfectly random system.

 

[antimony]

As a small note, if you want to do testing in game, it's easier to run Debug Mode and type "Reveal All", so you can see all the Civs on turn 1:
http://gameplay.tips/guides/253-civilization-vi.html