This is ancient and i don't know much about many of the topics touched. But it made me wonder about four questions, which seen crucial to the argument:
To unpack: One drawback of self reported life satisfaction and affect using Likert scales is the implicit request for a respondent to compare their current state against some standard. That standard may change with major life events, rendering comparisons before and after invalid. An alternative, direct way of estimating utility would be to ask bereaved parents: how much would you be willing to give to get your child back?
What's the utility cost of disability (rather than death) caused by malaria?
What's the evidence regarding the long term cost effectiveness of established mental health interventions such as CBT?
How well do we understand the concept of 'happiness' or what makes a 'good life' and how well do we need to understand these things to do different forms of good?
To identify the most efficient way of increasing net happiness for humankind overall, a lot of things need to be known, first of all what happiness is and how it can be quantified appropriately. For the psychometric concerns outlined above (and others), I think our understanding is a lot more rudimentary than that. If one accepts a 'little-gnostic' view regarding human happiness, it seems sensible to prioritise cases which are universally accepted as extreme forms of suffering - such as the loss of a child.