coloin wrote:Well.....great work by Red Ed and Ocean - I can just about see how they are getting their results - very impressive !

from Ocean's data - it would appear that the average size minmal puzzle would appear to have 27 clues !

Hopefully we shall soon have confirmation regarding the number of puzzles per individual grid.

It seems that the method used will bias the distribution towards high number of clues. How much this bias is, depends on some currently unknown properties.

Meanwhile the sampling continues, and estimates for 30s (upper end) and 20s (lower end) now emerge. But will not report new numbers yet, because of the aforementioned uncertainties.

In addition to slow convergence, the numbers r and m converge to some fixed values different from the 'true mean', because of 'biased sampling'. If we could reliably estimate the error bounds, we could also calculate upper/lower bounds for M(C), the number of minimals with C clues.