I once had a lecture on the relevance of freshwater ecology for human water supply. The lecturer brought up chlorination byproducts. Basically, freshwater is treated using chlorine (and a bunch of other processes) to disinfect the water prior to human consumption. This is a big reason why I (and probably you) have not contracted typhoid or cholera from drinking water. While making the water undeniably safer, chlorination produces small quantities of unintended chemicals, called chlorination byproducts.

Now, this lecturer said, chlorination byproducts are a little bit dangerous. How dangerous? Well, the percentage of people who would die from consuming chlorination byproducts every year is a tiny fraction of one percent.

But, the argument went, if you consider the case of China, then you multiply this tiny fraction of one percent by over a billion people, and you end up with ~800 people per year.

I think this is a fallacy. Multiplying the relative scale (a tiny fraction of one percent) by the intervention population (over a billion people) gives you an absolute number (~800) that seems large when you have no context. But if you also multiply other interventions, with much larger relative scales, by that same large intervention population, you end up with larger absolute numbers still.

The upshot: when comparing interventions within a single context, it is really important to a) consider many interventions, and b) compare the relative effectiveness of each.

The problem is considering interventions in isolation and then seeing a big number as impressive. A better approach is to compare as many different opportunities as you can think of, and pick the ones that top the list - these will be big in Luxembourg and big in China.

It’s still important to consider the absolute numbers - you should consider both. The absolute number will enable you to calculate that, say, fish welfare is higher-scale than camel welfare. Or even that China is higher-scale than Luxembourg - which is certainly correct. But the relative number will let you pick the best intervention within that large target population.

It’s also still important to consider the context. Maybe you have really good political connections in Luxembourg and you don’t speak a word of Chinese. This could mean that an ask is still bigger in Luxembourg. But no matter where you go, working on water chlorination byproducts will still probably be near the bottom of your list.