The problem is that individual rational optimization can and does lead to collective deterioration. In the abstract: If a given adoption has an individual utility of +1, and a collective detriment of -0.01, then for each rational individual actor in non small populations the decision to adopt is a net improvement even though the resulting utility for all is far below 0. This is why the old excuse of 'vote with your wallet' is nonsense.
You indicate rightly that this is a 'hard problem'. Especially because it goes against the new 'geocentrism' of 'market' dogma. But it being 'hard' should not be a ticket to just giving up.
No, I to think he's saying that -0.01 applies to everyone. Each individual choice is actually causing a loss of 80 million in utility (vs a gain of 1), but that loss is mostly borne by other people, so they are individually better off making that choice.