45 coaches online • Server time: 22:56
Forum Chat
Log in
Recent Forum Topics goto Post Finishing the 60 Gam...goto Post Borg Invasiongoto Post GIF Guide
SearchSearch 
Random numbers, luck and playing the odds...
 Goto page Previous  1, 2, 3, 4, 5, 6  Next
FUMBBL Forum Index » FUMBBL » General Chat
Post new topic   Reply to topic
View previous topic Log in to check your private messages View next topic
koadah



Joined: Mar 30, 2005
Post   Posted: Jan 04, 2006 - 11:30 Reply with quote Back to top
Buur wrote:
Pie.... Pie pie pie..... Want pie now..... Pie ... PIE! PIE! PPPPPPPIIIIIIIIEEEEEEE!!!!!!!!
-Buur


Yet more nonsense!

What is the probability that just saying "pie, pie, pie" will actually get you a kind of pie that you like?

How many kinds of pie are there? How many do you like? What kind of RNG are you using?

Will you have beer with the pie? How much beer do you need to drink before any pie will taste good?

_________________
Image
O[L]C 2016 Swiss! - 19th June! ---- All Star Bowl XII - Teams of Stars - Sign up NOW!
Vicimus



Joined: Nov 16, 2005
Post   Posted: Jan 04, 2006 - 11:32 Reply with quote Back to top
It's a psuedo-pie and thus will cheat you. If you eat it you will consistently roll double 1s.
Rynkky



Joined: Aug 03, 2004
Post   Posted: Jan 04, 2006 - 11:35 Reply with quote Back to top
Vicimus wrote:
No, it proves my maths was flawed. Nothing wrong with the Java RNG, just my brain.

The numbers where all right, I was just configuring the percentages incorrectly. I mean, do you really think the people at Java are so stupid they can't even create a "consistently inconsistent" random number between 1 and 6? ... gimme a break.

Anyway, the argument is closed. Look now.


Heh, you obviously missed the irony in my post. Wink
koadah



Joined: Mar 30, 2005
Post   Posted: Jan 04, 2006 - 11:36 Reply with quote Back to top
Vicimus wrote:
I have no idea. JavaScript was invented by the same people who made Java, although they are two different things really. JavaScript is interpreted by whatever browser you are using. The best way to really test the Java RNG would be to look at that page through a java browser. But that's just getting silly. I can guarantee now there will not be any discernable difference in the results. Any random number generator from any programming language is going to be absolutely fine and unfaultable for our purposes. Hell, even for online gambling purposes.


Surely javascript was invented by Netscape and java was invented by Sun.
I have never heard anything to sugest that they are linked in anyway.


EDIT:
Vicimus wrote:

The best way to really test the Java RNG would be to look at that page through a java browser.


No. It would be to code the test in java and not javascript.
You may get the same result, you may not.

I suppose it is a fair bet, but how do we know that different JVMs will always produce the same set of random numbers. Anyone tried IBM?
I assume that the client would throw a sunc error if there was a problem.

_________________
Image
O[L]C 2016 Swiss! - 19th June! ---- All Star Bowl XII - Teams of Stars - Sign up NOW!
Vicimus



Joined: Nov 16, 2005
Post   Posted: Jan 04, 2006 - 11:49 Reply with quote Back to top
Koadah,
JavaScript was definitely first used in Netscape ... but I thought they were in cahoots with Sun to topple the Microsoft empire. Who then came up with "JScript" (that interprets JavaScript with a MS flavour). In-fact, I'm 95% sure that's right.

Rynkky,
Sorry mate. I think I did, I just was unsure. People have been saying all sorts of strange things and I am having trouble keeping up.
Vicimus



Joined: Nov 16, 2005
Post   Posted: Jan 04, 2006 - 11:55 Reply with quote Back to top
koadah wrote:

No. It would be to code the test in java and not javascript.
JavaScript is an interpretted language. When you call the Math.Random function it says to the browser "choose a random number". If the browser was written in Java it would obviously choose that random number using the Java Virtual Machine and the in-built java random number generator. What I do not know is if there is some stipulation that says every browser that interprets JavaScript has to do so using the same random number generating algorithm.

Whatever, I'm 100% sure there would be no statistical difference. I actually (like I said in my other post) wrote this in Java first. I translated it to JavaScript only so it was accessable to everyone else on The Internet without bothering with an Applet.
fen



Joined: Sep 10, 2005
Post   Posted: Jan 04, 2006 - 11:58 Reply with quote Back to top
Quote:
1% - the probability of a zombie picking up the ball that is inside a tackle zone and then dodging out.


Is that right? Cause by my figuring it's 5/6 to pick up the ball then 4/6 to dodge out. Not including a team reroll. Which doesn't feel like it works out at 1%.

I'm not up for doing the math but as an estimate I thought it was nearer 16%. Curious because I've managed this with AG2 players a lot more than 1% of the time.
koadah



Joined: Mar 30, 2005
Post   Posted: Jan 04, 2006 - 12:02 Reply with quote Back to top
Vicimus wrote:
Koadah,
JavaScript was definitely first used in Netscape ... but I thought they were in cahoots with Sun to topple the Microsoft empire. Who then came up with "JScript" (that interprets JavaScript with a MS flavour). In-fact, I'm 95% sure that's right.


Netscape were the first company to licence java that does not mean that any java code went into javascript.

Sure everyone tried to gang up on microsoft. In those days MS thought that MSN would be their own rival internet! (continue reading when you have stopped laughing).

So, you may be right. But you may be testing completely different RNGs. If you are using IE then you definitely are.

I do agree that as far as the game is concerned it doesn't really matter.

_________________
Image
O[L]C 2016 Swiss! - 19th June! ---- All Star Bowl XII - Teams of Stars - Sign up NOW!
Topper



Joined: Aug 03, 2003
Post   Posted: Jan 04, 2006 - 12:05 Reply with quote Back to top
Actually I think it is 16, 67% without a RR.
The chance of picking up is 1/3, and the odds of dodging is 1/2 so it´s 1/3*1/2=1/6=16,67%
I think I´m right at least Smile
Vicimus



Joined: Nov 16, 2005
Post   Posted: Jan 04, 2006 - 12:05 Reply with quote Back to top
Sorry. Typo. You guys are obviously right about the zombie.

Regarding JavaScript and Java, who knows? Not me. But I don't think it alters the point at all. I can tell you first-hand that the results were no different running to code thru Java.
koadah



Joined: Mar 30, 2005
Post   Posted: Jan 04, 2006 - 12:15 Reply with quote Back to top
Vicimus wrote:
koadah wrote:

No. It would be to code the test in java and not javascript.
JavaScript is an interpretted language. When you call the Math.Random function it says to the browser "choose a random number". If the browser was written in Java it would obviously choose that random number using the Java Virtual Machine and the in-built java random number generator. What I do not know is if there is some stipulation that says every browser that interprets JavaScript has to do so using the same random number generating algorithm.

Whatever, I'm 100% sure there would be no statistical difference. I actually (like I said in my other post) wrote this in Java first. I translated it to JavaScript only so it was accessable to everyone else on The Internet without bothering with an Applet.


OK, I am just funning. Smile

The pure java browser idea is not 'best' because most of your users will not have a pure java browser so will need to find and install one. Most of them will have a java enabled browser that could run an applet. Smile

You still need to be sure that the same JVM is running the game, the pure java browser and the applet. Otherwise you may as well pick any C library and test that.

_________________
Image
O[L]C 2016 Swiss! - 19th June! ---- All Star Bowl XII - Teams of Stars - Sign up NOW!
Vicimus



Joined: Nov 16, 2005
Post   Posted: Jan 04, 2006 - 12:23 Reply with quote Back to top
Hehehe. You're just putting doubts in peoples heads now. I'd bet my right ball that the results would not show any statistcal difference no matter what language you used. Razz
Rynkky



Joined: Aug 03, 2004
Post   Posted: Jan 04, 2006 - 12:29 Reply with quote Back to top
Very simple randomn number tester in Java.

Compile

javac RandomNumberGenerator.java

Run

java RandomNumberGenerator 6 2000000

Results

Number 1 was rolled 333364 times (16.7% of all rolls)
Number 2 was rolled 332975 times (16.6% of all rolls)
Number 3 was rolled 333489 times (16.7% of all rolls)
Number 4 was rolled 333630 times (16.7% of all rolls)
Number 5 was rolled 333014 times (16.7% of all rolls)
Number 6 was rolled 333528 times (16.7% of all rolls)

Code

Code:
import java.util.Random;

public class RandomNumberTester
   {
   private int sides;
   private int numberOfRolls;
   private long[] rolls;
   private Random randomizer;

   public static void main( String[] args )
      {
      RandomNumberTester rnt = new RandomNumberTester(
         Integer.valueOf( args[ 0 ] ).intValue(),
         Integer.valueOf( args[ 1 ] ).intValue() );
      rnt.runTest();
      rnt.showResults();
      }

   private RandomNumberTester( int sides, int numberOfRolls )
      {
      this.sides = sides;
      this.numberOfRolls = numberOfRolls;
      rolls = new long[ sides + 1 ]; // slot 0 wasted
      randomizer = new Random();
      }

   private void runTest()
      {
      for( int i = numberOfRolls; i > 0; i-- )
         {
         rolls[ ( int )( randomizer.nextDouble() * sides ) + 1 ]++;
         }
      }

   private void showResults()
      {
      for( int i = 1; i <= sides; i++ )
         {
         StringBuffer sb = new StringBuffer();
         sb.append( "Number " );
         sb.append( i );
         sb.append( " was rolled " );
         sb.append( rolls[ i ] );
         sb.append( " times ("  );
         sb.append( roundToDecimals( ( ( double ) rolls[ i ] / numberOfRolls ) * 100, 1 ) );
         sb.append( "% of all rolls)" );
         System.out.println( sb.toString() );
         }
      }

   private double roundToDecimals( double value, int decimals )
      {
      double divisor = Math.pow( 10, decimals );
      value *= divisor;
      value = Math.round( value );
      return value / divisor;
      }
   }
qk



Joined: Oct 21, 2005
Post   Posted: Jan 04, 2006 - 12:33 Reply with quote Back to top
I think that people who doubt about the random generator wouldn't be happy with real dice anyway. They are far more unreliable!
macike



Joined: Jun 25, 2005
Post   Posted: Jan 04, 2006 - 13:22 Reply with quote Back to top
Vicimus wrote:
The only thing I ever tried to prove is that the Java Random Number generator is not faulty and will not cheat you in a game on Fumbbl. I am having a lot of trouble understanding your what is you're even trying to disagree with and your logic to why you disagree.

The little bit about probability statistics is just general interest. It's not an attempt to prove anything about psuedo random numbers. It's just a bunch of facts (assuming my math is correct).

Pseudorandomness is fine from the statistical (as they pass some statistical tests) point of view (when you collect the bunch of results and examine/analyze them) - macro view.

However 'bad' or 'good' streams are more likely to occur because they are not random (ie. determined by calculation) and have to balance each other. So pseudorandomness is not so fine from the probability of your next player action success point of view - micro view.

I think that Rynkky's idea of separate RNG for both teams is quite fine - everyone has its own bad/good streams.
As well as "one common" RNG shared for all FUMBBL games could be - everyone competes for others bad/good streams.

My point of view is that I can blame RNG for failed GFI because I'm helpless and I know that my fate is predetermined.

PS
If you don't understand anything, please do not say that it's partialy true. Ask for clarification first.

_________________
Hold him, thrill him, kick him, kill him!!!
Display posts from previous:     
 Jump to:   
All times are GMT + 1 Hour
Post new topic   Reply to topic
View previous topic Log in to check your private messages View next topic
Goto page Previous  1, 2, 3, 4, 5, 6  Next
 FUMBBL Forum Index » FUMBBL » General Chat