I haven't looked at the original criticism, but the "basic idea" as you describe it seems to introduce a source of bias: we have more visibility of luckily avoided ways in which things could have gone badly for recent events than for older ones, so if you try to take those into account then you will skew the change over time in the direction opposite to the one Pinker claims.

(If you also look for unluckily avoided ways in which things could have gone well then maybe the bias goes away.)