;D Lovely graphs!
I knew there would be a few kinks in the line but it didn't really bother me. There is so much inelegance in this mission that one more example didn't seem to matter. Also, like Watson-Watt, I'm a great believer in "second best today". Actually those lines are pretty good: the only problem is the kink in the low 20s. Even that is unavoidable with a crude system like this: the problem them was to do with the rounding of small random numbers to whole numbers, which is how loons come. There isn't much point in generating a command "random 1" if you have to round it up to 1 afterwards. ...
These imperfections are [deliberately] overwhelmed by the double randomness: not only is it random (within limits) but its different on every iteration. Consequently, even if you know the algorithm - if you can apply such a grand word to this - you still have no idea when it will happen. The best part is that the more iterations there are the more likely it is: in other words, if you spend a long time looking and can't find them, the more likely it is that the random _base will come good for you.
This was a classic example of necessity being the mother of invention. The original version had a sort of _base=random 22 thing, but after Sneaker played (and had such a different _startCount from Planck) I realised that something slightly more sophisticated was required. This bag of knitting just grew out of hair-tearing: ah that's it ... no it isn't, what if its X, damn, well ok, I'll bolt on this ... no bugger that doens't work either, right, THAT should do it. Again, if I'd known what was going to happen I'd have drawn the graphs first and then figured out the function. It's only y=mx+c after all.
And yes, _base will always be between 0 and 22. I worried more about the upper limit than the lower: the lower fell out of the game but the upper was just a finger in the air job.
And the sig line comment was a joke of course, put it back.

Not least because it works: 100 people have d/l Lookout! in the last few weeks.