Why we look at the limiting factor instead of the problem scale

Scale, or im­por­tance, is held as one of the 3 crite­ria to con­sider when eval­u­at­ing an in­ter­ven­tion for promis­ing­ness. With the idea be­ing that large scale prob­lems might sug­gest area area will be more effec­tive to work, on as­sum­ing it also scores well on the other crite­ria. Some in­ter­ven­tions are pred­i­cated on very strong scale ar­gu­ments, such as far fu­ture or wild an­i­mal suffer­ing. How­ever, we (Char­ity En­trepreneur­ship) have found that scale speci­fi­cally is quite a poor in­di­ca­tor of the promis­ing­ness of an area.

Or­ga­ni­za­tions tend to be in­cen­tivised to scale both from an im­pact per­spec­tive and from a per­sonal per­spec­tive. How­ever, all or­ga­ni­za­tions, even­tu­ally, hit a limit­ing fac­tor that makes it hard to scale faster. For some or­ga­ni­za­tions, it might be the to­tal scale of the is­sue. For ex­am­ple, per­haps when work­ing on a nearly erad­i­cated dis­ease, the limit­ing fac­tor might just be how much more of the prob­lem there is left to deal with. This would be a case when the tra­di­tional use of scale comes to al­most the same re­sult as the limit­ing fac­tor model. Both mod­els sug­gest this would not be a great area to work on due to the prob­lem scale be­ing quite small (cur­rently).

How­ever, there are many times when it di­verges. For ex­am­ple, cer­tain is­sues might have a mas­sive prob­lem scale, but be quickly limited by some other fac­tor. Givewell has talked about surg­eries be­ing limited by the sup­ply of sur­geons—this is not a scale of prob­lem is­sue speci­fi­cally, but it is a limit­ing fac­tor. A scale model might sug­gest that if there is a ton of surg­eries still to be done then this is a worth­while is­sue to fo­cus on. But a limit­ing fac­tor model would sug­gest that it would be quickly capped by the num­ber of sur­geons. Below is a very sim­plified com­par­i­son.

Scale model

Limit­ing fac­tor model

As you can see, the re­sults end up be­ing quite differ­ent: surg­eries are limited by the tal­ent pool far ear­lier than vac­ci­na­tions be­come limited by any­thing else. The mag­ni­tudes of the cat­e­gories were set to be more cross-com­pa­rable (e.g. 1 mil­lion com­pares to 10 full-time staff). And both get limited sev­eral times be­fore their prob­lem size limit. In this case, it would not even mat­ter if surg­eries had 10 times the scale (say, 200 mil­lion peo­ple af­fected) if they will be stopped by tal­ent, lo­gis­tics, and fund­ing be­fore they can even help 20 mil­lion peo­ple. Th­ese num­bers are es­ti­mates, but given our work and re­search in these ar­eas we are con­fi­dent both of these will hit a limit­ing fac­tor far be­low their prob­lem size (or what is of­ten referred to as ‘Scale or im­por­tance’).

A com­mon re­sponse is that scale is only one of the fac­tors con­sid­ered when eval­u­at­ing an in­ter­ven­tion. How­ever, things like ‘lo­gis­ti­cal limit’ fall un­der the radar of tractabil­ity/​solv­abil­ity, and there are two prob­lems with this. First, tractabil­ity is not re­ally cur­rently used in this way. Right now, lots of claims are be­ing made along the lines of “cause X should be fo­cused on more due to it hav­ing a huge prob­lem size” with no fur­ther refer­ence to tractabil­ity. Se­condly, tractabil­ity is of­ten seen as the speed at which you will make progress as op­posed to a spe­cific fac­tor that will stop growth from oc­cur­ring. For ex­am­ple, an in­ter­ven­tion could be very shovel ready, but only at a small scale be­fore its limit­ing fac­tor comes into play. If we go back the surgery char­ity ex­am­ple, it could be that 3 sur­geons want to start a char­ity, and for them it is very tractable and shovel ready, since their marginal effort is great and, at a very limited scale, their solv­abil­ity rate is very high. The en­deav­our will run into spe­cific “scale” is­sues in­volv­ing a limit­ing fac­tor of ‘hiring other sur­geons’, And this ends up feel­ing more like a scale is­sue rather than a tractabil­ity one.

Another com­mon con­cern might be fund­ing limits, which on their own is are not hard limits like prob­lem size. How­ever, while some­thing like a fund­ing limit can be im­proved upon with more fundrais­ing and field build­ing, this is not an easy task. For the money rank­ing, I do not think you would just put down the amount you have fundraised for the area, but a rea­son­able bound for how much could be fundraised, while tak­ing into ac­count the cur­rent donor space and a rea­son­able amount of time (e.g. 2-5 years). This can change over time, but so can the size of the prob­lem: fac­tory farm­ing is a grow­ing prob­lem and global poverty is a shrink­ing one, but that does not change the im­por­tance of hav­ing sense of their scale.

A claim that I hear a lot in the an­i­mal space is that Wild an­i­mal suffer­ing is such a huge scale prob­lem that we should se­ri­ously con­sider work­ing on it. Many peo­ple would sug­gest that Wild an­i­mal suffer­ing is an is­sue with a much larger scale than some­thing like vac­ci­na­tions based on the pure num­ber of be­ings af­fected. This claim is definitely true, since there are trillions of wild an­i­mals and only ~20 mil­lion peo­ple in need of any sin­gle vac­ci­na­tion. But if we look more closely at its limit­ing fac­tors, I think this claim is pretty mis­lead­ing.

The prob­lem size limit is in­deed huge. In fact, it has to be cropped, or oth­er­wise it would go all the way to the top of this post and the other sec­tions would be­come im­pos­si­ble to see. How­ever, that is not re­ally what mat­ters. Even if wild an­i­mal suffer­ing is a huge prob­lem and even if there is only a very limited amount of fund­ing and tal­ent that wants to work in the area, you will bump into prob­lems with scal­ing far be­fore it starts to mat­ter if there are a billion wild an­i­mals or a billion billion. In the prac­ti­cal cases, when com­par­ing these in­ter­ven­tions, if a char­ity were founded in both of these ar­eas, the vac­ci­na­tion char­ity would be able to get to a much larger scale than the wild an­i­mal suffer­ing-fo­cused char­ity could. A claim that “we should work on in­ter­ven­tion X due to its mas­sive prob­lem scale” seems quite in­ac­cu­rate. Th­ese sorts of ar­gu­ments are ex­tremely com­mon for wild an­i­mals speci­fi­cally and more broadly in EA.

Re­search­ing to de­ter­mine the causes’ limit­ing fac­tors gen­er­ally ends up with more time be­ing spent on con­sid­er­ing the fun­der and tal­ent space com­pared to, for ex­am­ple, map­ping out the spe­cific num­ber of an­i­mals af­fected by any given in­ter­ven­tion. And it usu­ally ends up with a set of fairly differ­ent in­ter­ven­tions look­ing promis­ing.

Of course, the per­spec­tive changes de­pend­ing on what you are look­ing to do in a given area. For ex­am­ple, when con­sid­er­ing donat­ing to a char­ity, the main thing ex­am­ined by Givewell is room for fund­ing. They are look­ing for an in­ter­ven­tion for which its limit­ing fac­tor to do­ing more good is fund­ing. Some in­ter­ven­tions might be very promis­ing, but due to tal­ent, lo­gis­ti­cal prob­lems, or the size of the prob­lem limi­ta­tions, even if Givewell gave them more fund­ing, they would not nec­es­sar­ily be able to cre­ate more im­pact. Or­ga­ni­za­tions with room for more fund­ing are gen­er­ally stopped not by other fac­tors but by ‘room for fund­ing’ it­self. Due to Givewell be­ing a fun­der, it makes sense for them to take a care­ful look at room for fund­ing as that is the limit­ing fac­tor they can im­prove. As a char­ity en­trepreneur, what the fund­ing space looks like for a given or­ga­ni­za­tion should play a big role in what char­ity to found. As an em­ployee at an NGO, you mostly have to con­sider how much of a limit­ing fac­tor tal­ent is for the or­ga­ni­za­tion.

Th­ese mod­els are re­ally sim­plified on both the scale and the limit­ing fac­tor side, and I think there are other pos­si­ble ways to use scale differ­ently (or use tractabil­ity in a differ­ent way to cover some of the same con­cerns). I do not think 100% of EAs use a sim­ple scale-based way of look­ing at prob­lems, but I do think a large % of EAs use a fairly sim­ple “size of the prob­lem” based way of con­sid­er­ing scale with­out thought to the % of the prob­lem they can solve, given the first limit­ing fac­tor that will stop growth/​progress.