There's one big omission in Tyler Cowen's argument that peace is bad for growth - opportunity cost.
The conventional neoclassical view is that £1 spent on the military is £1 not spent on something else, and that something else might well involve more productive activity than having soldiers sitting in barracks. This was the point made by Frederic Bastiat in his 1850 essay, That Which is Seen, and That Which is Not Seen:
When a tax-payer gives his money either to a soldier in exchange for nothing, or to a worker in exchange for something, all the ultimate consequences of the circulation of this money are the same in the two cases; only, in the second case, the tax-payer receives something, in the former he receives nothing. The result is - a dead loss to the nation.
Against this, Tyler's arguments seem questionable. His claim that "the very possibility of war focuses the attention of governments on getting some basic decisions right" might surprise those of us who suspect government is dysfunctional even in matters of life and death; the phrase "snafu", remember, is military slang.
And whilst he is right to point to some innovations that arose from militarism, doing so runs into Bastiat's injunction to consider the unseen. From a neoclassical viewpoint, military spending diverts resources from cilivian spending, some of which would have produced other innovations. It's possible that a bigger civilian economy would generate faster growth by "learning by doing effects", or simply because the private sector is on average better at innovating than the government.
All of these objections, however, rest upon a questionable assumption - that resources are fully employed so that higher military spending means lower civilian spending. If resources are unemployed, the opportunity cost argument fails so Tyler might be right: for this reason, he is too hasty to dismiss Keynesian considerations.
In this context, Tyler has some interesting predecessors - Marxian underconsumptionists such as Paul Baran and Paul Sweezy. Back in 1966 they argued that capitalist economies tend to stagnate because capitalism's massive potential to produce was not matched by growth in potential demand. Military spending, they argued, helped fill this gap and thus boost growth:
If one assumes the permanence of monopoly capitalism, with its proved incapacity to make rational use for peaceful and humane ends of its enormous productive potential, one must decide whether one prefers the mass unemployment and hopelessness characteristic of the Great Depression or the relative job security and material well-being provided by the huge military budgets of the 1940s and 1950s. (Monopoly Capital, p210)
Keynesians might object that civilian public spending would also boost aggregate demand. But, said Baran and Sweezy, from a capitalist point of view military spending was superior. It gave fat, reliable profits to government contractors in a way that, say, paying teachers doesn't. And a militaristic society fosters a culture amenable to capitalists, of conformity to authority.
So, who's right - Tyler and the Marxian underconsumptionists or the neoclassicals? Personally, I think this is another area where there are no universal truths in economics. Militarism might well have been bad for economies in the full-employment late 60s and inflationary 70s. But it's plausible that it helped in the deflationary 30s. And it could be that today has more in common with the 30s than the late 60s.