Phrased this way, it starts to sound like an ML problem. Optimize based on a cost function, now what cost function do we use? Having people choose to maximize

Phrased this way, it starts to sound like an ML problem. Optimize based on a cost function, now what cost function do we use? Having people choose to maximize happiness seems to be leading to worse outcomes by many objective standards. The monetary costs are high for the society in aggregate (and the externalities are not distributed according to the costs), health outcomes seem to be getting worse even to the extreme of lowering life expectancy, and so on. Purely seeking individual happiness seems like the wrong cost function based on that. Of course, focusing only on economic cost to society would be even worse, and health outcomes tend to be based in large part on individual happiness. I guess I'm saying that as you've characterized things, you make it sound like the "right" position is the correct one, but there seem to be a lot of things about that position that aren't so great. To be fair, you said "less total utilitarian happiness," but I thought it might help to give examples.