Quibbles about measurement aside, I agree with Steven Pinker that humans have become astoundingly more peaceful over the past several centuries. What I found unsatisfactory about Better Angels of Our Nature is precisely the rationalist element of Pinker's explanation for the decline of violence which Peter Singer singles out for praise, and nicely summarizes, in his laudatory NYT review:

To readers familiar with the literature in evolutionary psychology and its tendency to denigrate the role reason plays in human behavior, the most striking aspect of Pinker’s account is that the last of his “better angels” is reason. ...

Pinker’s claim that reason is an important factor in the trends he has described relies in part on the “Flynn effect” — the remarkable finding by the philosopher James Flynn that ever since I.Q. tests were first administered, the scores achieved by those taking the test have been rising. ... One theory is that we have gotten better at I.Q. tests because we live in a more symbol-rich environment. Flynn himself thinks that the spread of the scientific mode of reasoning has played a role.

Pinker argues that enhanced powers of reasoning give us the ability to detach ourselves from our immediate experience and from our personal or parochial perspective, and frame our ideas in more abstract, universal terms. This in turn leads to better moral commitments, including avoiding violence. It is just this kind of reasoning ability that has improved during the 20th century. He therefore suggests that the 20th century has seen a “moral Flynn effect, in which an accelerating escalator of reason carried us away from impulses that lead to violence” and that this lies behind the long peace, the new peace, and the rights revolution.

Singer is right that "the most striking aspect of Pinker's account" is the heavy explanatory weight he loads onto reason. That evolutionary psychology denigrates the role of reason in human behavior is not incidental. According to the sort of evolutionary psychology to which I take Pinker to subscribe, there is no such thing as capital 'R' Enlightenment reason. This is the objection I pressed in my ridiculously short (and I'm sorry to say oversimplified) review of Better Angels for The Daily:

Pinker is among the world’s foremost evolutionary psychologists. The prevailing Darwinian view, to which Pinker subscribes, is that the mind is a cluster of interrelated “modules” selected by evolutionary pressures to perform specialized tasks. But there is no general-purpose reasoning module. Rationality, in its abstract, impartial Enlightenment glory, does not come naturally to us. Pinker understands this. “Humans were not, of course, created in a state of Original Reason,” he says.

But then how does reason develop? Pinker expects that “as collective rationality is honed over the ages, it will progressively whittle away at the short-sighted and hot-blooded impulses toward violence, and force us to treat a greater number of rational agents as we would have them treat us.”But how did rationality get this far? How did it get off the ground in the first place? How does it spread? How does it stick? If the progress of morality depends finally upon the progress of reason, Pinker owes us a fuller account of how mental modules meant for other purposes were made to work in concert to produce something like Enlightenment reason. He owes us an account of how this discipline of mind is cultivated, “honed over the ages,” and passed along culturally. If the key to the astounding decline in human cruelty and death is the cultural evolution of reason, I’d like to hear rather more about it.

Pinker's account of the development of reason seems to me quite thin. Here's much of what he has to say about it: