J
[120607 22:51] <fenris> do i remember correctly if there was something wrong with the random-generator picking civs?
[120607 22:51] <fenris> ...and that you fixed it?
[120607 23:01] <fenris> found it
[120607 23:01] <fenris> http://www.xomicron.com/tracker/view.php?id=8
[120607 23:01] <fenris> id 8 :D
[120607 23:02] <fenris> Can you give me the details on what was wrong? And how that would maifest itself especially for multiplayer civ selection....?
[120607 23:02] <fenris> ref:
[120607 23:02] <fenris> viewtopic.php?f=68&t=75471
[120608 07:44] <dude> it's random
[120608 07:44] <dude> the 1.0c random number generator tended to repeat sequences slightly more often
[120608 07:44] <dude> and was limited to 32768 values
[120608 07:45] <dude> in 1.1 the limit is 4 billion values
[120608 07:45] <dude> and the period until the sequence repeats is far longer
[120608 07:45] <dude> but in both cases
[120608 07:45] <dude> it's just
[120608 07:45] <dude> there's no pattern to it
[120608 07:46] <dude> other than how it rejects some civs due to another player already selecting it
[120608 07:46] <dude> so it has to try again
[120608 07:46] <dude> and again
[120608 07:46] <dude> until it gets a unique one
[120608 07:46] <dude> which reduces the total civ count
[120608 07:46] <dude> to select from
[120608 07:46] <dude> and stuff :D
random in reality is a pseudo-concept, in perfection it's random, however using a human or a computer to select things at random always allows room for bias - due to in the former's actions, in the latter's programming.
The fact that in 1.0c it's called random, but in 1.1 its somehow "more random", is a prime example.
Alot of people have pointed out theyve seen patterns, it is a fallacy, I'd call it the "infallable recall of observation" fallacy. I remember more times I've looked at the clock and it's been 11:11 or 3:21, because it was an event that took my interest so I remember it better. Like wise I remember the times when I got saracens 4 times in a row, and huns 5 times n a row, because they stand out in my memory. I however am sure there are far more times when I got 5 different civs in a row, I just don't recall them so well because they are far more common - and thus mundane.
You can flip a coin 100 times in a row and get heads 100 times, it is still random (as much as it can be as determined by our actions and the environment and the laws of physics)
People have already shown its possible to make someone pocket in 3v3 by color picking vs ip's. Why wouldnt it be possible the 'random civ option' isnt that random at all?
It would be damn funny to see people fighting for red color in game :lol:
Time for some fun with math!
If something is really pretty random, there is a 95% chance that any variations from the mean will fall within two standard deviations. Anything outside of this is suspect as "not really random enough"
The mean of one civilization coming up is 2.777777......
The standard deviation is calculated as NVP ^ .5, where N is the number of trials, P is the probability that an event will occur, and V is the probability it will not occur.
50 X 1/18 X 17/18 ~=2.6234
The square root of that is about 1.619
So two standard deviations would be about 3.24
Therefore, if it were supposed to be random, we would be suspicious if a civilization turned up less than 0 times, or more than 6 times.
Mongols turned up 6 times for you, but not more. No civilization came up fewer than 0 times.
Unless a larger sample denotes otherwise, there is no reason based on the information provided to be suspicious of foul play. It is reasonable to assume a random distribution.