BitKeeper, Linux, and licensing disputes: How Linus wrote Git in 14 days (Graphite blog)

Eh... I think this is a better illustration of the benefits of pragmatism. Linus happily used closed source software right up until it became a problem, and then replaced it with FOSS when it proved to be untenable. The kernel got several years of Linus's undivided attention out of that arrangement, and paid basically nothing in return (some shouting matches on LKML notwithstanding). Perhaps there is an argument that the world as a whole would have benefited from Git coming into existence earlier... but:

1. It is not our place to tell Linus what he should or should not work on. If you wanted Git to exist earlier, you should've done it yourself (or paid someone else to do it, which is exactly what some people did, hence BitMover's revenue stream for many years).
2. Linus has a finite amount of time in a day that he can spend on programming, project management, and related work. He spends a non-zero amount of that time on Git, which means he can't spend it on the kernel or anything else. So determining whether the early arrival of Git would have been better would require us to figure how the kernel's development path would have changed, assuming Linus had less time for it, and comparing that against the benefit of an early Git. This is not straightforward to calculate.
3. Would Linus have developed Git, if he had never seen BitKeeper? I mean, probably eventually *somebody* would have (Merkle trees were patented in 1979, so the necessary tech mostly already existed), and I think it is even reasonable to grant that that person could plausibly still have been Linus, but would that person have been Linus in 2002? If we suppose that BK does not get adopted, and Git does not get developed (by Linus or whoever else) until 2005 or later, then what happens to the kernel from 2002 to 2005?

TL;DR: Alt-history is a complicated exercise which usually lacks clear winners and losers, if you're doing it right.