I agree that private innovation almost always outpaces public innovation. But it took massive government to get the ball rolling in the first place. You need to keep this in mind next time you make silly overgeneralizations like "all subsidies are bad."
Why didn't we see ARPAnet get huge in the early 1980s? The real bottleneck to internet adoption was the lack of computing power (memory and CPU), which were necessary for the rise of graphical interfaces, which were necessary for lots of people to be able to use computers. These things were developed rapidly, and privately. The limits that were pushed to do this were not government edicts, but physical constraints (heat dissipation, die shrinks.) Remember that Xerox PARC was a private research lab!
The "lead" you're talking about sat idle for twenty years until cheap computers got good enough to help normal people. Therefore that "lead" was essentially wasted according to any sensible economic theory.
(I think that ARPAnet was technically great, and TCP/IP, sockets, Unix, etc., are all fine technologies. All I'm saying is that we're at no loss economically if technological development is paid for in a moral way, rather than the way we do it now.)
There are lots of ways to do networking, and government subsidy didn't pay for all of them. (Even if HTTP didn't exist, gopher would have filled in just fine!) Nor did it pay for the commercialization of the web browser, which is what made it truly practical for people to use the web. (Remember that before Netscape, browsers from academia didn't support JPEG, and javascript hadn't been invented yet.)