The best part about Pinker's book is he actually uses data to backup his thesis. Gray is just rambling and appealing to emotion.
If you look at it in terms of what percentage of people are acting non-violently towards each other, 90% of the population being peaceful versus 99.98% seems to show a clear positive movement.
It not only shows that, as a whole, we are more capable of being non-violent, it also provides more opportunities for people to move away from violent situations into peaceful ones.
I'm open to criticism of Pinker, but I don't think this article proves that Pinker's position is totally flawed.
1000 pages is a lot of cherry picking. And let's face it, at least his thesis is declared and we can interpret his data selection in that light. Gray offers no quantitative evidence. As far as I can tell, he's trying to cow the reader with a lot of "I read way more political philosophy than any a' y'all." Taleb offers a statistically oriented critique, but again offers no data, and apparently no proof-reading.
If anything, the evidence we have from Gray and Taleb is N=2: 2 ostensibly smart fellows found enough grist in Pinker's work to write a publishable response without gathering any data themselves. That alone suggests Pinker's thesis has merit.
My own view is that the thesis is not important. What is really at issue is the subtext that we should not be too worried about western violence because we are doing better than ever before and that violence is decreasing "naturally", neither of which is true or follows from the claim that violence has been historically in decline.
That is why double-blind studies exist, and one of the key parts in data-based social studies programs contains whole classes on building experiments: data is easily abusable. While Pinker might have charts, without understanding the causality of those charts - in depth! - all it becomes is propaganda for Pinker's ideas and a quick way to sway nerds.
(not to say that Pinker is right or wrong, but having data doesn't add credibility ipso facto).
From conversations with social scientists I came to the conclusion, that their means of analysis are rather primitive. In most cases they also have no quantitative models (like in physics) to make predictions.