|
So I heard Tasteless and Artosis agree that different colors seemed to get certain races a different amount of the time. The obvious initial reaction to that is to say that it's some kind of cognitive bias, but I wanted to check anyways.
The first thing I did was boot up Brood War and start a game on an 8 player map. I set all 8 players to random, revealed the map, and wrote down what race corresponded to each color (Red, Blue, Cyan, Purple, Orange, Brown, White, and Yellow). I did this 96 times. Around 25 times in, I started to notice that every single game, the race distributions would be 3, 3, and 2; so there was allowed to be a maximum of 3 of any race and a minimum of 2. Naturally, this will make the results significantly less random (I assume). However, here are my results:
I had to learn a lot about statistics to figure this out. Anyways, the chi-squared for Blue (the color Flash consistently picked in the Ro4) gave a 96% chance that it is not a 1/3-each distribution of races, and that Protoss is significantly more likely than the other two races. White, Orange, and Red were also possible hits, but it's unsure, and the others seem to be pretty straightforward.
If I was to re-do the experiment, I would first try to find out the rules for the cap of how many races there have to be in any given scenario, find out the maximum number of players I can have to ensure randomness, and do it again. The way I did this, it's possible that some other factors were playing into certain colors being skewed, like the order that the colors pick their race or something.
What do you guys think? Is this something, or nothing?
|
It's interesting that blue shows a very high likelihood of bias. Nice find.
In 8 player games where all players go random, the race distribution will indeed always be 3-3-2.
Also, when both players go random in 1v1, it's very unlikely to be a mirror match. I'm a bit hazy on the details but I think it's a 1/24 chance, and the actual formula to get to that is obscure and bizarre. It was posted on here sometime back with various scenarios and I believe the specific race a player is also affects the distribution of the 'other two' (non-mirror) races of the random player as well.
For improvements on your testing, I'd recommend trying to reproduce what happens in Flash random games as closely as possible - meaning 1v1 game, with the exact game mode, number, race etc. of players and observers. It's possible/likely that some of these variables are intertwined, so higher likelihood of protoss in an 8 player game may not necessarily hold for games with less players.
I personally think Flash is onto something here and he's not just picking blue because he likes the color.
|
|
|
On November 06 2020 21:51 lurkgoon wrote:
It's interesting that blue shows a very high likelihood of bias. Nice find.
In 8 player games where all players go random, the race distribution will indeed always be 3-3-2.
Also, when both players go random in 1v1, it's very unlikely to be a mirror match. I'm a bit hazy on the details but I think it's a 1/24 chance, and the actual formula to get to that is obscure and bizarre. It was posted on here sometime back with various scenarios and I believe the specific race a player is also affects the distribution of the 'other two' (non-mirror) races of the random player as well.
For improvements on your testing, I'd recommend trying to reproduce what happens in Flash random games as closely as possible - meaning 1v1 game, with the exact game mode, number, race etc. of players and observers. It's possible/likely that some of these variables are intertwined, so higher likelihood of protoss in an 8 player game may not necessarily hold for games with less players.
I personally think Flash is onto something here and he's not just picking blue because he likes the color.
Eesh, I hope observers wouldn't be involved. That would be a huge pain lol.
On November 06 2020 22:12 lurkgoon wrote:tec27 posted an awesome explanation of how random works in this game depending on selections, though no mention of color in there. I'm not even sure if it was something he was looking for (?) or it is part of a different section of code. Who knows. Post #18 in this thread: https://tl.net/forum/brood-war/413197-few-mirrors-when-both-players-random
Thanks, that's really helpful actually.
|
|
|
So I'm gonna retry this, 1v1, gonna try out some random colors. Obviously I'll start with Blue and we'll go from there.
Looking back, I'm not even sure the first experiment was even any good at all. It just auto-loaded the 8 colors I listed, but that's clearly different from the 20 choices that you get if you start up a LAN game. I wonder if they're not even related, like in the same section of the programming.
I guess we'll see what happens. I'll try to update in the next day or two.
|
Ross something to take into consideration is certain colors not being selectable on certain types of maps. Like orange on a desert or white on a snow map or green on a grass map. I don’t know how that works but I think you aren’t supposed to be allowed to pick the color?
|
Alright, so I tried exactly as they did it:
RvZ
Random chooses the color Blue, Zerg chooses the color Random
Top vs Bottom, which is the game mode they chose
50 games on Fighting Spirit
I got:
13 games of Terran
22 games of Zerg
15 games of Protoss
According to a Chi-squared distribution with 2 degrees of freedom, that has a 60% likelihood of happening with random chance.
My brain is absolutely shot. I have not much to think about this anymore, because quite frankly I can't think anymore.
I'm glad I did this, and I had fun doing it, but I would not recommend anybody else do it, unless you are as autistic as I am.
|
such a random thread 
|
On November 07 2020 00:43 pebble444 wrote:such a random thread
*ZING*
|
I would suspect that one's color does not impact the randomization of race. If it did then Tec_27 would find mentions of some "selected color" variable when looking through the code in RandomizeRaces() and
GetRandomRace(...).
Unless this was added later in remastered, the plot thickens..
|
great work from a crazy guy. I respect your commitment
|
He was sitting in the blue chair for the match I think he was picking blue to match for the broadcast
|
|
|
|
|
|
EPT![[image loading]](https://pbs.twimg.com/media/EmIeR0LW0AAq8iN?format=jpg&name=large)
