Storxx
Joined: Aug 02, 2003
|
  Posted:
Jul 11, 2003 - 09:45 |
|
Hi,
we discussed in the channle, that some of us have experienced,
that if u reroll a result to fast, the same result is comming up.
So a double Skull rerolled and again the same, shouldn't come up so often.
Could it be that there is delay till the random mashcine comes up with new results ?
Bye
Vagabond, coach of the ever-dieing Whirl-Wind-Walkers |
|
|
BadMrMojo
Joined: Aug 02, 2003
|
  Posted:
Jul 11, 2003 - 16:11 |
|
I really think that it's just plain old bad luck. Of course, there is only one thing I am superstitious about in the whole wide world. And that's dice.
I always make it a point to click somewhere else (in the dialogue box, for example) before rerolling anything. Why? I have no clue, but I'd swear I get better results. Is this realistic, scientific data? A documented bug?
Hell no. It's really just luck. |
_________________ Ta-Ouch! of BloodBowl
Condensed Guide for Newbies |
|
Christer
Joined: Aug 02, 2003
|
The time between the accesses to the randomizer is completely irrelevant. Java uses a very established random number generator formula, which you can find in Donald Knuth's book "The Art of Computer Programming, volume 2".
Essentially, it updates a big number called a seed for every access. This seed is scaled down to whatever number range you ask for (1-6 most often in javabbowl) in a way that makes each number equally likely to occur.
The probability of rolling the exact same dice you rolled last time is 1/36. Given the number of rolls, it's likely to happen at least a few times during each game (counting blocks, dodges and whatever rolls you do).
As for the time aspect, let us play with the idea of a random number algorithm that is dependant on the time.
Let's assume your computer runs at 1GHz which isn't very unlikely..
Now, being a java program, let's assume that one instruction (tiny little mathematical operation) takes 1000 times longer than a pure assember isntruction. This is very very likely to be far exaggerated. But even so, you'd have 1,000,000 operations per second. The amount of time it takes for you to react to a reroll dialog or press the reroll counter could be say half a second at best. That would mean the computer would manage 500,000 instructions in this time, during which it would have pleeeenty of time to update any internal variables.
Ok, so what if the random number was dependant on the system timer? Normally, these system timers have a resolution of 1ms, so it changes 1000 times per second. During that same half a second "reaction time" the timer would change 500 times. It's physically impossible for you to react within 1ms, even if the client would display any dialog or result within 1ms of asking for the random number (Which doesn't happen. Displaying dialogs or drawing images takes longer than that).
-- Christer |
|
|
Christer
Joined: Aug 02, 2003
|
Oh. And for the double skull thing.. Let's see if we can't do some math on that as well..
The probability of rolling four skulls in a row is 1/1296.
You have 32 turns in a game (16 turns per player). FUMBBL sees ~300 games per day which means 9600 turns.
So, the expected number of quad skulls is: 9600/1296, or ~7.4 assuming an average of one block per turn.
Let's say the average number of blocks per turn is 2 (low number, I'm sure) which gives you about 15 quad skulls _each day_.
I'm not at all surpised people are seeing them being mentioned on the channel. I'd be surprised if they didn't show up.
Note that the number goes up substantially if your opponent has block and you don't.. (1/81 instead of 1/1296).
-- Christer |
|
|
Chappy
Joined: Aug 02, 2003
|
  Posted:
Jul 11, 2003 - 17:42 |
|
Crhrister....you make my brain hurt........ |
|
|
locutus
Joined: Aug 02, 2003
|
  Posted:
Jul 11, 2003 - 19:16 |
|
Christer, what do you eat?. Can i get some?... |
|
|
Christer
Joined: Aug 02, 2003
|
You guys should see the strategy article I'm working on...
-- Christer |
|
|
psikobunny
Joined: Aug 02, 2003
|
  Posted:
Jul 11, 2003 - 20:49 |
|
Thanks for the dose of perspective Christer, I hope it will cut down on some of the whining a tiny bit.
I wish everyone could take a course or two on probability, statistics, or game theory before they are allowed to complain about how the number generator "sucks" or "is out to get them." Even a Psychology class about selective perception would do. |
|
|
Sinner
Joined: Aug 02, 2003
|
  Posted:
Jul 11, 2003 - 20:58 |
|
oh crap last game i rolled 3 double 6 in a row... please help me stats are for me... lol |
_________________ Sinner
Darkie's Dreams - successfully cherrypicking any race, any coach, any rating, any number of DP since 20/09/2003 ... and still winning! |
|
slyspygb
Joined: Aug 02, 2003
|
  Posted:
Jul 11, 2003 - 21:26 |
|
I like the theory that you are working on christer but have you taken into a/c the variable of the little green dice that no-one like to roll as it brings nothing but death to characters........in your own team. |
|
|
|