Lindy is right about some things, but ridiculously wrong about others. In particular about helmets he does the usual nerd idiocy of thinking that if he just heard about survivor bias, certainly nobody else did. I mean, the allies may have been good enough at statistics to determine the exact armament production of Germany, or break several nations' encryption, but surely they were stupid enough to never have heard of survivor bias, if Lindy wasn't there to tell them about it.
Derp.
In reality, there is another well documented phenomenon which Lindy seems to have never heard about, namely that people adjust their behaviour according to the perceived level of risk. If they think that X makes them safer (whether justified or not,) they'll take higher risks than without X, pretty much placing themselves at the same level of risk they find acceptable.
Basically if you put a railing next to a cliff edge, more people go up to the cliff edge to look down.
This can get as ridiculous as that people who were told they were given a lucky ball played golf more aggressively. Which can be unimportant or even good if we're talking about just a golf game with no extra stakes, but bad if you poke your head out of the trench when there's a sniper around. And the use of snipers rose sharply through WW1.
And in WW1 they discovered just that: soldiers who thought that the helmet can stop bullets were actually more inclined to poke their head out and get shot in the head. THAT is why they briefly considered whether or not it's a better idea to recall the helmets.
Of course, it IS possible for the effect to be smaller than the actual benefits. E.g., a railing may still stop more lives even if it makes more people come to the edge and lean over it. E.g., in the case of helmets, it still prevented more deaths from getting hit in the head by a big rock thrown around by a heavy artillery shell, even if, yes, there was an actual increase in soldiers shot in the head.