phresnel: It was actually a vague reference to the article posted by marcjulian:

Quote:Original post by marcjulian
Hi,

you should read the article "The psychology of randomness" of Shawn Hargreave.

http://blogs.msdn.com/shawnhar/archive/2009/12/17/the-psychology-of-randomness.aspx

:)

Thanks for all the additional information guys, and the article that was linked to, it was an ejoyable read.

Kind Regards
David

I should have posted this earlier, but I remembered about it now: A long time ago I saw an easy method to determine if clustering data is random, more uniform than random or more clustered than random. The number of times that each value appears in you sequence should follow something very close to a Poisson distribution. A Poisson distribution has the peculiarity that the variance and the average are the same value. One can use the ratio of the two to detect departures from randomness (i.e., if the ratio is very far from 1, the numbers are not normal). I believe I read about this in the context of archeology, but it should apply just fine to your examples.

I made a quick Perl script to compute this ratio for each of your examples, and I got the following ratios:

1.03448275862069
1.10344827586207
0.96551724137931


I haven't thought about what distribution this ratio should have for random input, but these numbers don't look very far from 1 to me...