My reading of the article’s core argument is the following: longtermism assumes that it is possible to perform an accounting of future outcomes, and from this expected value calculations, even if just in sketch, that can be used to allocate resources in the present, so as to maximise total “good”. This process is incredibly prone to bias, and in reality appears to amount to little more than selective speculation about its proponents’ desired futures, and a corresponding proposal for the targeting of present resources.

I find this argument compelling - most methodologies for inter-temporal decision making acknowledge the large uncertainties associated with the future, and, either explicitly or implicitly, avoid making claims about the distant future, either in terms of what plans we should make for the future, or in how future states should affect our current plans.

There is no reason to believe that the proponents of longtermism have addressed this issue - indeed if they had it would be a very significant breakthrough, and we could apply their method to more immediate problems as well. What seems much more likely is that they have not addressed the issue at all, and that their purported accounting is really just a regurgitation of their prior beliefs, and thus, as the article suggests, consequently serves only to justify the existing behaviours of those who subscribe to it.