You wish to create one monolithic gene pool by interbreeding as much as possible, whereas I wish to create a homogeneous population by isolation.

You misunderstand completely if you think that I have any wishes at all. I'm describing how I believe human genetics has actually worked, and my approving comments have to do with the efficacy of the "genetic engine" compared to passive radiation and differentiation. That sets the context for my view of your wish to create a homogenous population: Diversity and hybridization are simply a better way to produce a highly-evolved species in the shortest period of time.

One privilege of being a dilettante is that I can cheerfully drag quantum mechanics into a discussion of evolutionary biology. You speak disparagingly of a "scatter gun effect" and an "alchemist madly mixing and matching compounds", but yet it's emerging that universe itself evolved in exactly that way; which is to say, according to principles conceptually the same as biological evolution. Random quantum fluctuations provide an input analogous to mutation, and the laws of physics act upon those fluctuations to select those that "survive". The simplest example is the way that random quantum mass fluctuations were acted upon by the law of gravity to select for clumping, which, continued over a long enough period of time, produced galaxies and stars and planets.

This adds a lot of power to my old aphorism that "physics begets chemistry begets biology". It appears that the principle of selection from random input is universal and in fact is the source of the universe's power to produce, indeed its apparent "affinity" for, complexity.

So in fact, Lee, the "scatter gun effect" and "mixing and matching compounds" is a fundamental and powerful force of nature. It's your loss to disparage it.

To step back and put it in perspective, biology is only now emerging from a distinctly Newtonian and mechanistic world view into the quantum and relativistic understanding of the universe. Those hierarchical taxonomies and the idea of distinct species and bloodlines and the notion of engineering genes, all those things are fading away. The view I'm giving of the human genome as a "cloud" or a "flow" is the more accurate one. "Family trees" based on a few hundred isolated fossils are, we now understand, absurd fantasies. Homo erectus is a hallucination, just a name given to a few fossils which represent no more than a snapshot in time of one local human population. Across the whole continent of Africa, hominid genes would have been constantly diverging and recombining in real time, every generation. Hominids didn't decide every few thousand years to hold a speciation event; there is no such thing as a "speciation event", only the constant accumulation of tiny changes that you recognize as speciation only with 20/20 hindsight.

The gene flow chart we've been discussing shows only the recent movements of homo sapiens sapiens. But what led to the dominance of h.sap.sap by ~70,000 years ago? Why is there only one species of hominid? Why aren't we sharing the world with multiple species in the genus Homo, or at least with multiple subspecies of homo sapiens?

I know the simplistic explanation, that we're just the most violent and bloodthirsty of the lot, but that explanation doesn't hold water. If there was a replacement of other hominids by humans, it had to have happened essentially simultaneously all over the world; as if it were a preplanned campaign of genocide. Otherwise you'd see genetic artifacts of surviving populations, and the conquerers themselves would have had time to drift and differentiate. A replacement event might have reduced the number of hominid species and subspecies, but it couldn't, by itself, have resulted in only one survivor.

As noted above, we're not on solid ground anyway talking about hominid species, and I think that a better explanation of the outcome is to discard replacement entirely and look to the disciplining effect of a central genetic engine to maintain a degree of standardization of the whole genome, keeping it centered about a common set of characteristics. Instead of fragmentation into many isolated gene pools, the engine maintained a common pool which evolved on average over a long time. The hallucination of hominid species, as stated, is just an artifact of taking a snapshot of the gene pool in one location at one time.

Or, consider it this way: As long as migration time was determined by hominid walking speed, the engine was "clocked" at a relatively long cycle time, and that permitted wide variation within the genome. The faster we learned to travel, the shorter became the cycle time, and because variations were checked in to the central library more often, genetic diversity was diminished, and the genome started to become uniform enough to be called a species. In modern times I think that one reason we're unlikely to see further physical evolution is because the engine's cycle time is very short nowadays and very little average change is possible.

That's it in a nutshell: There are no precursor species to humanity, there's just one big cloud of genes that's evolved over time and gradually converged on us. No species differentiation, no replacement, no radiation and divergence. Just the cloud of hominid genes disciplined and shaped by the genetic engine of Africa.