One of the most important aspects of EA is our collective effort to understand which paths or actions, in given contexts, seem to do the most good. This I'll call prioritization research for this post. Below I show some cases of common goods (anything that potentially helps everyone doing prioritization research) and show a potentially important division of them to rival and non-rival goods.
Be warned! I don't have any direct conclusions and this might not be important at all. I have allocated 2 hours to think through this, mostly as freewriting, and decided to publish to see if this might be interesting to pursue further.
Prioritization research includes, among others: personal career choice, developing individual donation strategy, coming forward with the best policies as a part of an advisory committee, or figuring out what are the most pressing global challenges. These are all similar in their underlying goal - to figure out which choice does the most good - but they differ tremendously in the reasoning processes involved.
I'd like to understand better what are the commonalities. I'd like to construct a better mental model of what is needed to do good prioritization in any significant case and to understand what are the common knowledge and methodologies that we can improve on as a community, or any other lesson we can take. Let's look at some examples.
Let's say that we want to make a decision among several known choices.
- One common heuristic is to consider each possible choice as a part of a solution to some global problems (or causes) and filter by the importance of these global problems. This is a very quick and rough way to identify less promising choices and then to eliminate them or require more positive evidence to investigate them further.
- First common good: understanding of cause areas.
- Another is to make a rough cost-effectiveness analysis for each choice and compare the results. This strategy is more advanced and requires many things in order to work well in practice.
- Common goods:
- CEA frameworks and understanding of methodological problems.
- metrics for good outcomes (QALYs, lives saved, ..) and comparisons among them.
- (self-)training resources.
- tools like Guesstimate and existing templates.
- subject-matter data, estimations, and knowledge
- Common goods:
There are other methodologies for comparing choices, with broadly similar common goods. Of these, I'd like to mention Expert Interviews and direct reasoning (Informed Considerations) that raise two more common goods:
- an approachable network of experts.
- engagement on a platform for discussing and delegating questions (e.g. a peer network or the forum).
These two common goods differ in an important quality than those previously mentioned. In both of the last two goods, the user is costing others by taking valuable time and attention from experts or people on the platform (or even causing them to disengage with EA) - they can be thought of as rival goods. In the earlier cases of common goods though, anyone can use them (mostly) without problems - they are non-rival goods.
Rival goods can be allocated better through some form of coordination or designing constraints (social or technical) into the allocation mechanism. Also, note that other important rival goods are time and skills dedicated to increasing any of the common goods above. I am unclear about the value of this distinction, but I decided to publish here instead of thinking more about useful implications- if there are any. I am pretty sure that cataloging all such common goods would be useful, and that prioritizing between them according to the research community needs would be very important.