So for me, the motivation for categorizing altruistic projects into buckets (e.g., classifications of philanthropy) is to notice the opportunities, the gaps, the conceptual holes, the missing buckets. Some examples:
If you divide undertakings according to their beneficiaries and you have a good enough list of beneficiaries, you can notice which beneficiaries nobody is trying to help. For example, you might study invertebrate welfare, wild animal welfare, or something more exotic, such as suffering in fundamental physics.
If you have a list of tools, you can notice which tools aren’t being applied to which problems, or you can explicitly consider which tool-problem pairings are most promising. For example, ruthlessness isn’t often combined with altruism.
If you have a list of geographic locations, you can notice which ones seem more or less promising.
If you classify projects according to their level of specificity, you can notice that there aren’t many people doing high level strategic work, or, conversely, that there are too many strategists and that there aren’t many people making progress on the specifics.
More generally, if you have an organizing principle, you can optimize across that organizing principle. So here in order to be useful, a division of cause areas by some principle doesn’t have to be exhaustive, or even good in absolute terms, it just has to allow you to notice an axis of optimization. In practice, I’d also tend to think that having several incomplete categorization schemes among many axis is more useful than having one very complete categorization scheme among one axis.
So for me, the motivation for categorizing altruistic projects into buckets (e.g., classifications of philanthropy) is to notice the opportunities, the gaps, the conceptual holes, the missing buckets. Some examples:
If you divide undertakings according to their beneficiaries and you have a good enough list of beneficiaries, you can notice which beneficiaries nobody is trying to help. For example, you might study invertebrate welfare, wild animal welfare, or something more exotic, such as suffering in fundamental physics.
If you have a list of tools, you can notice which tools aren’t being applied to which problems, or you can explicitly consider which tool-problem pairings are most promising. For example, ruthlessness isn’t often combined with altruism.
If you have a list of geographic locations, you can notice which ones seem more or less promising.
If you classify projects according to their level of specificity, you can notice that there aren’t many people doing high level strategic work, or, conversely, that there are too many strategists and that there aren’t many people making progress on the specifics.
More generally, if you have an organizing principle, you can optimize across that organizing principle. So here in order to be useful, a division of cause areas by some principle doesn’t have to be exhaustive, or even good in absolute terms, it just has to allow you to notice an axis of optimization. In practice, I’d also tend to think that having several incomplete categorization schemes among many axis is more useful than having one very complete categorization scheme among one axis.