Theoretical models are nice (I did some for my dissertation), but real-world is the test--does it match the model or not?
Here are a bunch of species (below); these are data points which can't be ignored. Your model must take these into account or it is useless.
You ask how many mutations it takes? As many as it takes. Make your model fit the real world and it has a better chance of being taken seriously.
A model which says "it can't happen" is useless in the face of empirical data that shows that "it did happen."
Figure 1.4.4. Fossil hominid skulls. Some of the figures have been modified for ease of comparison (only left-right mirroring or removal of a jawbone). (Images © 2000 Smithsonian Institution.)
- (A) Pan troglodytes, chimpanzee, modern
(B) Australopithecus africanus, STS 5, 2.6 My
(C) Australopithecus africanus, STS 71, 2.5 My
(D) Homo habilis, KNM-ER 1813, 1.9 My
(E) Homo habilis, OH24, 1.8 My
(F) Homo rudolfensis, KNM-ER 1470, 1.8 My
(G) Homo erectus, Dmanisi cranium D2700, 1.75 My
(H) Homo ergaster (early H. erectus), KNM-ER 3733, 1.75 My
(I) Homo heidelbergensis, "Rhodesia man," 300,000 - 125,000 y
(J) Homo sapiens neanderthalensis, La Ferrassie 1, 70,000 y
(K) Homo sapiens neanderthalensis, La Chappelle-aux-Saints, 60,000 y
(L) Homo sapiens neanderthalensis, Le Moustier, 45,000 y
(M) Homo sapiens sapiens, Cro-Magnon I, 30,000 y
(N) Homo sapiens sapiens, modern