| Forum Index > BW General |
|
Wuselmops
Germany346 Posts
 In the bw.de-forum someone stated that the random generator of BW isn't Laplace-like anymore when at least 2 players choose random. For example the probablilty of a mirror-matchup in RvR is 1/24 (it would be 1/3 with a Laplace-generator). Also one race isn't picked more than 3 times in a game. He said he disassembled the code to find that out. Can anyone confirm this or give more details? | ||
|
ShabZzoY!
Great Britain760 Posts
Theres a tread where it was discussed at length | ||
|
Wuselmops
Germany346 Posts
| ||
Liquid`Drone
Norway28789 Posts
random vs random = only 1/24 becomes a mirror matchup (I used to think 1/27, but 1/24 makes sense too) but picked vs random = mirror matchup 1/3 of the time. also if 4 people all go random (say a 2v2), it always becomes 1/1/2 (or, I guess you might get something different 1/24 of the time) (if I play 2v2 all random and both me and my team partner gets zz, I can always assume that my opponents are tp) so yeah he's right, but it's nothing new, although it might very well be new to a lot of players. =] | ||
|
Asta
Germany3491 Posts
2 random players: chance of a mirror match up at 1/24 3 random players: not 3 times the same race 4: no race 3 or 4 times, only one race two times 5: no race more than 2 times 6: no race more than 3 times, at most one race 3 times 7: no race more than 3 times, exactly one race 3 times 8: no race four times or more from my experience it might be true that there are such regulations (3v3 is almost always pretty balanced, you never get less than 3 different races) | ||
|
ShabZzoY!
Great Britain760 Posts
http://www.teamliquid.net/forum/viewmessage.php?topic_id=21905 | ||
|
Esp1noza
Russian Federation481 Posts
| ||
|
oHInsane
France727 Posts
| ||
|
Liquid`Drone
Norway28789 Posts
| ||
|
Patriot.dlk
Sweden5462 Posts
They also modifed things like this in Wc3 TFT.. It's harder to get crossmaped, meaning the startposses is not randomed equally there either=p Thank god for no TvT when going random  | ||
|
Wuselmops
Germany346 Posts
Have you also noticed that random-players are worse in mirror-matches? Now I have two reasons for this: 1. Random players play mirror-matches less often because they are unlikely when they play vs other random players. 2. Random players usually have this advantage: They can put themselves into their opponent's position because they often play the matchup vice-versa. Only in mirror-matches the one-race-only players can do this as well. | ||
|
SeBASTa
China1147 Posts
On February 21 2006 08:09 Liquid`Drone wrote: I don't think it's a bug, and personally a good part of why I love rvr is I don't have to worry about mirror matchups. =[ I'd hate if they changed this. the mirror matchups trend to be zvz personally i never met pvp and only 1 time tvt,but zvz is very common when i massed rvr games. | ||
|
Wuselmops
Germany346 Posts
| ||
|
IdrA
United States11541 Posts
On February 21 2006 08:16 Wuselmops wrote: its more this one, when you play random you have more to learn so unless you actually practice 3x as much as anyone else your not gonna play each race as much. but for example tvp you also learn from playing pvt so you dont need to practice tvp quite as much as if you didnt play pvt. mirror matchups the only way to learn it is to play it, so they dont get to play it from both sides. 2. Random players usually have this advantage: They can put themselves into their opponent's position because they often play the matchup vice-versa. Only in mirror-matches the one-race-only players can do this as well. | ||
|
prOxi.swAMi
Australia3091 Posts
if a player chooses random he should be able to expect exactly that: random! | ||
|
Wuselmops
Germany346 Posts
| ||
|
haduken
Australia8267 Posts
| ||
|
Wuselmops
Germany346 Posts
| ||
|
Sfydjklm
United States9218 Posts
| ||
|
nysikto
Norway337 Posts
| ||
|
Physician
United States4146 Posts
There is something we refer to as the, "Random bug" that i've known for years and years. It does not work on observer maps (it may, I forget and have a lot of christmas alcohol in me), or if any person in the game PICKS his race. In 1on1's, it is indeed random. And is always random. The chance of a mirror matchup however in RvR is extremely rare and almost never happens. But, it occasionally does. In 2on2's, if all four players random. And you are z/z your opponents are t/p (automatically). If all four random and you and your ally get z/p, one of your opponents is T. And the other may be z/t/p. Likewise if you are t/t they are z/p. Etc.. etc... In games with 8 players, for race wars. All players can random without ever having to fear that you will have a team of five protosses, three terrans etc... It is always 3 of one race, 3 of another, and 2 of the other. Unless one person picks and fucks it all up. So in any game including 1on1, it is indeed random. There are days when you will random zerg almost for an entire day, random terran, or random toss. But they are rare instances. I didn't read any of the other posts. So if anyone else pointed this out, good job. If not, what a bunch of fucking morons. Happy Holidays dt Germany. December 25 2004 16:35. I have disassembled Starcrafts "random" algorithm long ago, Nal_testie is right of course. It's quite complex and the person who coded it was probably drunk. The probability of a mirror 1vs1 matchup if both players go random is exactly 1/24, for instance. Only 2 useful things said in da original thread. | ||
|
Haemonculus
United States6980 Posts
On February 21 2006 07:35 Liquid`Drone wrote: well random vs random = only 1/24 becomes a mirror matchup (I used to think 1/27, but 1/24 makes sense too) but picked vs random = mirror matchup 1/3 of the time. Random vs Random SHOULD result in a mirror 1/3 of the time, as should picked vs random. Probability wise they're the same. How exactly does the generator decide to pick 1/24 instead? | ||
Insane
United States4991 Posts
On February 21 2006 07:35 Liquid`Drone wrote: well random vs random = only 1/24 becomes a mirror matchup (I used to think 1/27, but 1/24 makes sense too) but picked vs random = mirror matchup 1/3 of the time. It should be 1/3 for both I think. There are 9 possible matchups in RvR, and 3 of those are mirror matchups -- The race one person gets should not affected by the race the other person gets. Edit: Although apparently Blizzard did it oddly. | ||
|
Locked
United States4182 Posts
perfectly valid reason IMO, wouldn't it suck to have like 8-players random and 6 of them get terran, or if you're mass RvRing then you get mirrors 1/3 of the time? | ||
|
Haemonculus
United States6980 Posts
| ||
| ||
|
|
PiGosaur Cup
GSL
Rogue vs Percival
Zoun vs Solar
Replay Cast
GSL
Cure vs TriGGeR
ByuN vs Bunny
KCM Race Survival
Replay Cast
Replay Cast
Escore
OSC
Replay Cast
[ Show More ] |
|
|
EPT
