Ticket to Ride France Ticket to Ride France

Forums

Search
Forums » Memoir '44 Online - English » Dice Rolling Algorithms Cannot Be Right
Show: Today's Posts 
  
AuthorTopic
Dasher
Senior Member
Captain

User Pages
Posts: 130
Registered:
March 2006
Dice Rolling Algorithms Cannot Be Right Wed, 16 March 2011 21:16
Over the course of some 16 games of "Hellfire pass", the statistics for the computer show an average superiority in dice rolling percentages of 16.5% above the averages for the human player.

That is, if Human Player achieved 24% Infantry, 10% Armor and 16% Grenades on his die rolls, the computer averaged 40% Infantry, 26% Armor and a staggering 32%(!) Grenades on its rolls.

Often, the computer had twice as many Grenade symbols rolled as it did Infantry!

I don't know much about programming, but I do know dice (Vegas, baby), and I can't help but think those kinds of results have got to be the result of some form of manipulation in the code.

First off, "Dice are blind".

When the computer is rolling dice for itself, is each die result generated randomly without regard to the other? That is, are the dice programmed to avoid duplication of certain results?

Second, "Dice live and die alone".

Is each die facing predicted independently? i.e., are the two faces showing Infantry simulated as two faces with a 16.666% probability each, or are they generated as a single probability of 33.333%? ?

Finally (and most important), "Dice have no memory".

Is there some mechanic for how well the computer or the human player has rolled in the past, manipulating probability to "even out" the odds with each result generation ?

I'm just curious, because, I give you my word, I don't (think I) suck at this game, but i have yet to beat the computer, and while I am fine with that if the game is truly balanced, I'm not so keen on the notion if there are "Sid Meier Mechanics" running hidden in the background.
      
Nygaard
Senior Member
Général de brigade

User Pages
Posts: 1010
Registered:
May 2006
Re:Dice Rolling Algorithms Cannot Be Right Wed, 16 March 2011 21:24
... So how do you explain I keep rolling better than the computer?
      
rasmussen81
DoW Content Provider
Railway Tycoon

User Pages
Posts: 8148
Registered:
July 2007
Re:Dice Rolling Algorithms Cannot Be Right Wed, 16 March 2011 22:02
Darion Leonidas wrote on Thu, 17 March 2011 00:16


I'm just curious, because, I give you my word, I don't (think I) suck at this game, but i have yet to beat the computer, and while I am fine with that if the game is truly balanced, I'm not so keen on the notion if there are "Sid Meier Mechanics" running hidden in the background.


You're asking some good questions and I have no answers about the way the dice are programmed...one of the crew will have to speak on that if they have time.

However, I do know that the AI usually has a style of play that can be murder if you don't take it into account (believe me, I've lost my share of battles against ever version of Johnny).

Stick with it!! Cool
      
john
-= Crew =-
Cadet

User Pages
Posts: 61
Registered:
October 2002
Re:Dice Rolling Algorithms Cannot Be Right Wed, 16 March 2011 23:21
Darion Leonidas wrote on Wed, 16 March 2011 13:16

When the computer is rolling dice for itself, is each die result generated randomly without regard to the other?


YES (also true when rolling dice for human players)

Darion Leonidas wrote on Wed, 16 March 2011 13:16

That is, are the dice programmed to avoid duplication of certain results?


NO

Darion Leonidas wrote on Wed, 16 March 2011 13:16


Is each die facing predicted independently? i.e., are the two faces showing Infantry simulated as two faces with a 16.666% probability each, or are they generated as a single probability of 33.333%? ?


It is not possible to perform an experiment that would distinguish one approach from the other.
The code generates a RANDOM integer between 1 and 6 and assigns both 1 AND 2 to an Infantry. The information about weather the infantry came from a 1 or a 2 is LOST. (like with REAL dice)

Darion Leonidas wrote on Wed, 16 March 2011 13:16


Is there some mechanic for how well the computer or the human player has rolled in the past, manipulating probability to "even out" the odds with each result generation ?


NO
      
sam1812
Senior Member
Brigadier General

User Pages
Posts: 2649
Registered:
August 2006
Re:Dice Rolling Algorithms Cannot Be Right Wed, 16 March 2011 23:39
A few weeks ago, on another thread, Yann posted the total number of occurrences of each of the five shapes, from the huge number of games that had been played up until then, and the totals were right on target. So the aggregate distribution of rolls is fine.

A number of players have wondered if the dice may be unusually streaky. To the best of my knowledge, no statistical randomness test has been performed.
      
Phread
Senior Member
Railway Tycoon

User Pages
Posts: 1777
Registered:
December 2008
Re:Dice Rolling Algorithms Cannot Be Right Thu, 17 March 2011 00:05
Nygaard wrote on Thu, 17 March 2011 10:09

... So how do you explain I keep rolling better than the computer?



Because you paid the premium for the "good dice" Shocked Laughing Shocked Laughing
      
eldonion
Senior Member
Minesweeper

User Pages
Posts: 159
Registered:
December 2007
Re:Dice Rolling Algorithms Cannot Be Right Thu, 17 March 2011 01:19
Phread wrote on Wed, 16 March 2011 23:05

Nygaard wrote on Thu, 17 March 2011 10:09

... So how do you explain I keep rolling better than the computer?



Because you paid the premium for the "good dice" Shocked Laughing Shocked Laughing



Where do i buy these "good dice"!!!
      
Phread
Senior Member
Railway Tycoon

User Pages
Posts: 1777
Registered:
December 2008
Re:Dice Rolling Algorithms Cannot Be Right Thu, 17 March 2011 01:25
If you can't see the checkbox for "good dice" then you aren't eligible for "good dice."
And if you have to ask where the checkbox is you aren't eligible either.

It's not what you know but who you know that counts.
      
Clexton27
Senior Member
Arnhem Tree

User Pages
Posts: 3389
Registered:
February 2007
Re:Dice Rolling Algorithms Cannot Be Right Thu, 17 March 2011 17:13
Quote:

It's not what you know but who you know that counts.


What Phread means is that he does a lot of praying before he tosses those babies...
Cool

[Updated on: Thu, 17 March 2011 17:15]

      
Phread
Senior Member
Railway Tycoon

User Pages
Posts: 1777
Registered:
December 2008
Re:Dice Rolling Algorithms Cannot Be Right Thu, 17 March 2011 23:22
stevens wrote on Fri, 18 March 2011 05:58

Quote:

It's not what you know but who you know that counts.


What Phread means is that he does a lot of praying before he tosses those babies...
Cool


Laughing Laughing Laughing ROTFLOL Laughing Laughing Laughing
      
Nygaard
Senior Member
Général de brigade

User Pages
Posts: 1010
Registered:
May 2006
Re:Dice Rolling Algorithms Cannot Be Right Thu, 17 March 2011 23:39
Phread wrote on Thu, 17 March 2011 23:22

stevens wrote on Fri, 18 March 2011 05:58

Quote:

It's not what you know but who you know that counts.


What Phread means is that he does a lot of praying before he tosses those babies...
Cool


Laughing Laughing Laughing ROTFLOL Laughing Laughing Laughing


Which of course make the important question: What divine being does Phread worship? (I'm betting it's Joe Pesci)
      
Phread
Senior Member
Railway Tycoon

User Pages
Posts: 1777
Registered:
December 2008
Re:Dice Rolling Algorithms Cannot Be Right Fri, 18 March 2011 00:59
Nygaard wrote on Fri, 18 March 2011 12:24


Which of course make the important question: What divine being does Phread worship? (I'm betting it's Joe Pesci)


Nope. The Risen Jesus Christ.

But God, the Father, Son, and Holy Spirit doesn't help with M44 die rolling.
God is independent and supports us while we make our own luck.

[Updated on: Fri, 18 March 2011 01:00]

      
    
Previous Topic:Closest Enemy Unit ONLY? Pleas say it ain't so...
Next Topic:That's a first
Goto Forum: