I worry that some EAs consider certain interventions to be interventions with high expected value (high EV) because the intervention tackles a major EA cause area, rather than using major EA cause areas as a tool to identify high EV interventions. This strikes me as “thinking in the wrong direction”, and seems wrong because I think we should expect there to be many, many potential interventions in global health and development, reducing existential risk and improving animal welfare that have low expected value (low EV).
As a result of this error, I think some EAs overvalue some interventions in major EA cause areas, and undervalue some interventions that are not in major EA cause areas.
Because of one of the problems with the ITN framework (that we seem to switch between comparing problems and interventions when we move between importance, tractablity and neglectedness), I think it may help and be more accurate to view the major EA cause areas as areas where high EV interventions should be easier to find, and to view other cause areas as areas where high EV interventions should be harder to find.
Thinking in these terms would mean being more open to interventions that aren't in major EA cause areas.
The main examples where I think EAs may underestimate the EV of an intervention because it doesn't involve a major EA cause area are those where a particular form of activism / social movement / organisation could potentially be made more efficient / effective at a low cost. There are probably many such examples, with some having much greater EV and some having much smaller EV, but 2 examples I'd provide are: a) starting a campaign for the USA to recognise Palestine (https://forum.effectivealtruism.org/posts/qHhLrcDyhGQoPgsDg/should-someone-start-a-grassroots-campaign-for-usa-to) b) identifying areas and ethnic groups internationally at greatest risk of genocide / ethnic violence and trying to redirect funding for western anti-racism movements towards these areas
From discussion in comments: One general point I'd like to make is if a proposed intervention is "improving the efficiency of work on cause X", a large amount of resources already being poored into cause X should actually increase the EV of the proposed intervention (but obviously, this is assuming that the work on cause X is positive in expectation).