An enormous amount of credit goes to johnswentworth who made this new post possible.
This is a framing practicum post. We’ll talk about what selection incentives are, how to recognize selection incentives in the wild, and what questions to ask when you find them. Then, we’ll have a challenge to apply the idea.
Today’s challenge: come up with 3 examples of selection incentives which do not resemble any you’ve seen before. They don’t need to be good, they don’t need to be useful, they just need to be novel (to you).
Expected time: ~15-30 minutes at most, including the Bonus Exercise.
What Are Selection Incentives?
Imagine trying to find great/popular posts on LessWrong. We look for things like high karma values, high number of comments, or a well-known writer. We don’t really look at the contents of the individual posts (yet), we just look at an overall “score” that can help us to choose posts. This overall “score” mechanism encourages writers to write posts that could potentially achieve high “scores”, for instance broad-interest posts, thought-provoking posts, controversial posts, etc, regardless of what the actual purpose of the writer is.
This is a selection incentive: Something is chosen based on some criteria or a known process. For instance, posts are chosen based on an overall “score”: high karma values, high number of comments, etc. On the other hand, presenting ideas or transferring knowledge (not pursuing high karma values) might be what the writers actually want. But, the readers’ selection criteria are there regardless of what writers actually want in the first place.
Another example is corporations maximizing profits. The founder of the corporation has something in mind, for instance sending humans to space, producing the most affordable cars to the mass population, etc., and they may or may not be trying to maximize profit. What happens in the real business world, however, is that businesses live or die based on how well they maximize profits. Businesses are selected on the basis of how well they maximize profits, regardless of what the founders actually want.
What To Look For
In general, selection incentives should spring to mind whenever something is chosen based on some criteria or a known process. We want to know what factors cause something to be more or less likely chosen. A few ways this can apply:
We are selecting/choosing something based on some criteria.
We see systems which grow or die, and we want to know what causes the system to grow/die faster or slower.
Useful Questions To Ask
In the post selection example, posts with high karma values or high number of comments are more likely to be chosen than the ones with low karma values or low number of comments. But the post writers imagine a post that can present ideas/thoughts, transfer knowledge, or initiate a communication. High karma values may correlate with great posts, but may not align with what the writer actually wants, i.e., transfer knowledge and/or initiate conversation between the writer and the reader. What the writer actually wants diverges from what the selection criterion selects for/incentivizes.
In general, if an agent is involved, we want to know how the things the agent wants diverge from what the selection criteria “want”.
Come up with 3 examples of selection incentives which do not resemble any you’ve seen before. They don’t need to be good, they don’t need to be useful, they just need to be novel (to you).
Any answer must include at least 3 to count, and they must be novel to you. That’s the challenge. We’re here to challenge ourselves, not just review examples we already know.
However, they don’t have to be very good answers or even correct answers. Posting wrong things on the internet is scary, but a very fast way to learn, and I will enforce a high bar for kindness in response-comments. I will personally default to upvoting every complete answer, even if parts of it are wrong, and I encourage others to do the same.
Post your answers inside of spoiler tags. (How do I do that?)
Celebrate others’ answers. This is really important, especially for tougher questions. Sharing exercises in public is a scary experience. I don’t want people to leave this having back-chained the experience “If I go outside my comfort zone, people will look down on me”. So be generous with those upvotes. I certainly will be.
If you comment on someone else’s answers, focus on making exciting, novel ideas work — instead of tearing apart worse ideas. Yes, And is encouraged.
I will remove comments which I deem insufficiently kind, even if I believe they are valuable comments. I want people to feel encouraged to try and fail here, and that means enforcing nicer norms than usual.
If you get stuck, look for:
Systems in which something is chosen based on some criteria or a known process.
Systems which grow/die and what makes them grow/die faster or slower.
Bonus Exercise: for each of your three examples from the challenge, explain:
What strategies do the selection criteria incentivize for the things being selected?
If an agent is being selected, how do the things the agent wants diverge from what the selection criteria are?
If the agent is making the selection, how is it different from the selection criteria?
This bonus exercise is great blog-post fodder!
Using a framing tool is sort of like using a trigger-action pattern: the hard part is to notice a pattern, a place where a particular tool can apply (the “trigger”). Once we notice the pattern, it suggests certain questions or approximations (the “action”). This challenge is meant to train the trigger-step: we look for novel examples to ingrain the abstract trigger pattern (separate from examples/contexts we already know).
The Bonus Exercise is meant to train the action-step: apply whatever questions/approximations the frame suggests, in order to build the reflex of applying them when we notice selection incentives.
Hopefully, this will make it easier to notice when a selection incentive frame can be applied to a new problem you don’t understand in the wild, and to actually use it.