Amid the various recent
whacks at considerations of Gladwell lately, I find this one, by Razib Khan, particularly helpful in defining what sometimes goes amiss with Gladwell — and the danger that waits every science writer:
[Gladwell's problem is that] out of the possible set of ideas and models, only a subset can be turned into an interesting piece of prose, and only a subset are actually non-trivially true (that is, they stand the test of the time, not just falling below the p-value for the purposes of getting published once, and, add something which isn’t a mathematically fluffing up of something we already knew verbally or intuitively). The intersection between the two subsets is rather small proportion of the peer-reviewed literature at any given time.
There’s a parallel in science, of course. It’s the tension between explaining a phenomenon (say, coral reef formation) in the most parsimonious, conservative way possible and explaining it in the most imaginative way that can be squared with the facts. That tension drives science. Science writers best heed it as well, and watch their steps lest they tread too far afield.