A proper law

[Ian G Batten]

On Thu, 06 Mar 2003, Peter Tomlinson wrote:

> In my experience, the processing uses a fixed algorithm, and is not an
> attempt to spot non-randomness. One practitioner in this field told me to
> print out the bit stream as a 2D plot (e.g. just fill a page with the bits)
> and check it by eye for patterns.

That's bad advice (in those bald terms, at least).  There are repeatable
experiments which take 1000 random points on a page, and get people to
compare them with 1000 points which are generated by plotting a regular
grid and then purturbing each point by a random distances, whose mean
value is less than the initial grid spacing.  People tend to see the
latter as ``more random'', as it evinces fewer clumps and empty
regions.

ian