When I say it is more annoying I mean, it is not only a prime number which makes it much harder to rescale it's exponents which are also annoying numbers such as 121,1331,14641. Plus you would have to manually offset it rather than using a simple reduction of the exponent to achieve similar effects.
... I suppose using 11 as the exponent could work however it is more of an annoying number than anything which is why I preferred 10, plus logic says that it should be a factor of 10, ...
Your mathematical skills are in evidence now.
Also Xolo why not try a simple 10% ie 0.09*10^(L-M) instead of 0.1*10^(L-M), that should be enough leeway to not constantly be negative.
Maybe I shouldn't write this in a hurry, but doesn't produce 0.1^(M+2-L)*10 the same results as 0.1*10(L-M)? So the formula could easily be adjusted to whatever factor that seems appropriate for a chain length step-up. Say it were a probability of 1/11 to find a chain L+1 in comarison to find a chain L, you could calculate that part of the weight formula as (1/11)^(M+2-L)*10.
That would replace a linear behaviour...
I don't see why we need to rescale the exponents and if ever necesary this could be done with logarithms, or not?