There's an external cost (or benefit) to everything. Fully calculating the impact of a given action on everyone else would be absurdly computationally intractab

There's an external cost (or benefit) to everything. Fully calculating the impact of a given action on everyone else would be absurdly computationally intractable, particularly for extremely diffuse or probabilistic effects. So, the question to ask must be not "what are the externalities of this action", but "what externalities are significant enough to take into account"? The author of the article is concentrating on obvious costs felt independently by individuals, and ignoring diffuse non-economic benefits (such as... not accumulating huge piles of trash filled with difficult-to-recover resources). He's quite possibly correct anyway that it's a poor trade-off, or that it's at least poorly implemented, but he really isn't making a persuasive case, especially given the lack of any actual non-anecdotal evidence of what the costs and benefits are.