|
RE: Will I only get males?
Ah. But, remember, we're talking about computer programs. The intention of random number generators is to produce a 'fair coin'. A lot of very good mathematicians have put a lot of work into designing those random number generators. Sure, technically, since they are generators, and so, by definition, cannot be 'truly' random. And, yes, some functions designed in the '80s were shown in the 90s to be weak on producing unpredictable results. But, neither you nor I are one of those scary good mathematicians qualified to either design OR test those functions.
What I have done, and what KittyCatS has done (yes, we talked at length about it, years ago), is perform the basic tests to ensure the results appear fair.
Yes, you can hypothesize all sorts of alternate implementations. And you're free to waste your time testing to see if any of them were used. But the code is simple. And, even using the weakest random number generator available, rand(), will produce results which will require millenia for you to 'crack' ..and that's assuming you have results for EVERY cat's gender. The code is simply "if rand(0,1) == 0 then gender = female else gender = male" or something very much like that. And all tests performed by KittyCats (who DOES have access to 100% of the results) and me (using randomly selected samples) indicate the results produce 50/50 distribution. within the statistical limits imposed by the sizes of the datasets.
If you have results indicating otherwise, instead of hypothesizing complex systems which might have been used, and you feel your results indicate the distribution is not 50/50, then you'd better spend your time better examining the sources of bias in your dataset. For example, you're selecting only 'pure' from 'pure' parents, which means you're, by your admission, discarding a number of offspring. And I doubt you're using randomly selected parents, either, since you're attempting to breed 'pure' lines. So, it seems to me you're likely more often selecting, even if subconsciously, males-from-this-line to breed with females-from-other-lines and ignoring that YOUR selection bias will, naturally, produce lines biased to one gender or the other.
As for weighting systems and multi-factor systems, there are simple tests which can be (and have been, repeatedly) performed which can show whether they exist. And, as I said, KittyCatS and I (along with a few others) have performed these tests. The results have consistently been presumptive-negative for information content in gender results .. in other words, within the limits of sample (which, remember, is 100% coverage for KittyCatS) there is no information content in gender .. which means, to a very high degree of confidence, there is no weighting and no undiscovered, obscure, multi-factor systems.
|
