articulett said:
Oh. Well then, that explains it. And for now, that's the explanation that fits the data best.
but wait...does that mean...that...ewww...I feel so dirty.
Exactly. Kleinman denies that the gratification is sexual, so perhaps I'm a bit off. But in any case, isn't his behavior eerily like those old experiments where the rat would continue to press the lever, over and over and over again, long after the reward was discontinued?
Speaking of data, Kleinman has posted some results of varying the site width, but only in ranges where Rcapacity remains greater than Rfrequency, where varying the site width has little effect. This is yet more deliberate deception.
Here are two series varying only the binding site width.
genome length 1024
number of binding sites 8
Rfrequency = 7
mutation rate 1/512 bases
weight width 5
population 64
site width / Rcapacity / generations to perfect creature
10 / 20 / 5,746
9 / 18 / 7,186
8 / 16 / 5,073
7 / 14 / 2,778
6 / 12 / 5,180
5 / 10 / 6,848
4 / 8 / 12,298
3 / 6 / no correct binding sites in >7,000,000 generations
genome length 4096
number of binding sites 8
Rfrequency = 9
mutation rate 1/512 bases
weight width 5
population 64
site width / Rcapacity / generations to perfect creature
10 / 20 / 14,459
9 / 18 / 19,907
8 / 16 / 18,228
7 / 14 / 18,390
6 / 12 / 16,396
5 / 10 / 75,384
4 / 8 / no correct binding sites in >550,000 generations
Notice that when Rcapacity is less than Rfrequency no binding sites evolve, and when it's close to (but greater than) Rfrequency the number of generations to evolve a perfect creature increases sharply.
Now imagine what would happen if you ran a series of trials with the site width fixed at 5 (and hence Rcapacity fixed at 10), the number of binding sites fixed at 8, with a series of increasing genome lengths. At length 1024 you'd see (from the data above, first series) 6,848 generations, and at length 4096 you'd see (from the data above, second series) 75,384 generations. Quadrupling the genome length appears to increase the generations to convergence elevenfold. You might conclude that the convergence rate slows down with genome length by a power of 2 or so, or perhaps even exponentially, especially once you see that the next fourfold increase in genome length (making Rfrequency exceed Rcapacity) fails to converge at all. You might conclude that, that is, until you realized your error of scaling up other parameters without suitably scaling up the binding site width. (Analogy: if you scale up a scale model train, but forget to scale up the wheels in proportion, it won't run very well.)
This is exactly what Kleinman did, though at different numbers (except for the part about realizing his error, though it was pointed out to him months ago). It's what his entire argument is based on. It was an understandable error at first, and I can see why he would be disappointed when he thought he'd found an interesting and important result and turned out to be wrong, but now he's just pushing the lever over and over and over.
I talk about his motives for making false claims because all his claims have been so thoroughly refuted that there's nothing else left to talk about. It is, however, an interesting topic. If there's no gratification involved, then perhaps it's because he believes in a religion, such as Discordianism, SubGenius, or certain branches of Satanism, in which lying is approved of.
Respectfully,
Myriad