The Fool wrote: 
>  
> Here's how you figure out how random something is: 
> convert the data to binary numbers in  a file and 
> then compress that file with some compression 
> program.  If the compressed file cannot compress 
> the file (0% compression) (note compressed file 
> will be bigger than uncompressed file), then the 
> data is completely random / equivalent the output 
> of a good encryption routine. 
>  
I think this is not what random means. This 
might be a good test of randomness, except that 
only _deterministic_ processes could generate 
output that would _always_ match this criterium. 
 
Let me explain: a random 6-sized die is one that 
_will_ generate a sequence of 1 million 6s, 
about once every 6^1,000,000 times. 
 
A process that would never fail in that compression 
test would never generate such sequence. 
 
Alberto Monteiro 
 

Reply via email to