What Works and for Whom

I’ve been reading a bunch of Slate Star Codex and thinking a lot about what makes something “instrumentally useful” i.e. it helps you survive or complete some goal that will reinforce its use; both of these have a lot to do with natural selection, probably one of my top 5 favorite concepts.

One of the classic mistakes, which I find myself making again and again, is to consider an organism or object to be the direct product of the natural selection. The Giraffe has a long neck because long necks were selected for. Tupperware is light and flexible because heavy and brittle containers broke a lot and no one bought more once there were other options. Weed keeps getting stronger because people keep paying more for stronger weed.

Except, Giraffe’s genes really keep winning this just as much as the Giraffe’s do. The companies that only sell breakable containers switched over or went out of business. The stores that have access to stronger weed are being naturally selected for.

A different but related concept is the idea of being “instrumentally useful” i.e. when an attribute actually gets things done. It might sound good to have a very nice set of silverware that your in-laws will compliment when you have them over for dinner, but if you dread polishing the set so much you always avoid bringing them out they might not be instrumentally useful.

I like this way of thinking about whether abstract technologies, like words or concepts or social protocols, are worth inventing or worth destroying. But the above observation about the mistakes I make regarding evolution makes me think: instrumentally useful to whom? If we invent a technique where people can change the way they speak to make it easier for people who are face blind to recognize social subtext, most people won’t use it because it requires extra energy to learn and use and forces them to give up on certain levels of nuance.

If the government tried to incentivize this new way of verbalizing social subtext by giving money to companies that enforced the use of this new technique, one highly likely outcome is that companies would do a terrible job unless the government had an effective audit system. The company wants the money, but it doesn’t want to force employees to do more things and make themselves a less attractive company to be in. On the other hand, sometimes companies make a point out of engaging with accessibility concerns and specifically tries to attract employees who see this kind of accessibility concern as pro, though often enforcement is actually still low in these companies because they care more about signaling than actual enforcement. If the government does audit companies it plans to give such awards to and the company does enforce the use of this verbalizing technique and fires people who don’t use it, then it becomes instrumentally useful for all, and becomes common company-wide.

This kind of hierarchical process makes me think that in order to make anything interesting happen, you have to incentivize it to be instrumentally useful on all the levels that have agency in this situation.