From: Denis Hennessy <denis@...>
Date: 2008-01-18T21:14:13+09:00
Subject: Re: [SUMMARY] Counting Cards (#152)

On 18 Jan 2008, at 09:19, Gareth Adams wrote:

> Ruby Quiz <james <at> grayproductions.net> writes:
>
>> 	    size.times do |i|
>> 	      j = rand(size)
>> 	      @shoe[i],@shoe[j] = @shoe[j],@shoe[i]
>> 	    end
>> (the code is just a longhand form of @shoe.sort_by { rand }).
>
> Easy to think that, but this is the exact same definition of a  
> "naive algorithm"
> demonstrated recently and very well summed up by Jeff Atwood:
>
> http://www.codinghorror.com/blog/archives/001015.html
>
> Gareth

That's very interesting. The core of the bug is the the original  
algorithm generates 52*52 possible outcomes, instead of 52! outcomes.  
The non-randomness occurs because 52*52 is not evenly dividable by 52!.

Also interesting that you can learn new things by reading C# code...

/dh
>