@natematias @dansinker @danhon @jessamyn @darius Cost-benefit is where my mind goes first, too… but at the risk of "responsibility-splaining" to folks with more experience, if the analysis is done uncritically, it's easy to mistake "we" for a constant, when "we" live within a varied matrix of interlocking oppressions.
"Our" harms will be unevenly distributed.
Analyzed uncritically, cost-benefit data'll show you that being Black & gay & an immigrant is inherently high-risk, when our problem is having far too many systems which each quite efficiently route harm toward such a person.
Right now, the lack of QTs seems to act as a Jersey barrier, with just enough friction to safely redirect low-effort, but occasionally intentional harms.
Taking the analogy further, hopefully there's an elegant design to let traffic cross lanes and change directions more safely… but any data analysis needs to be careful about what it mistakes for a constant, whose benefits / harms "count," & how they do.