Adverse advice selection
January 25, 2022 — November 15, 2022
C&C the Reverse advice you hear idea, although I think the framing in my head is broader and better.
The concept I want to get at is something like the one in Bravery Debates: “I feel pretty okay about both being sort of a libertarian and writing an essay arguing against libertarianism, because the world generally isn’t libertarian enough but the sorts of people who read long online political essays generally are way more libertarian than can possibly be healthy.”
We can imagine advice as a statistical problem, where we need to get external validity in our advice. If we are also learners, should we not be concerned about how our own hypotheses generalize to the wider world outside our experience?
Every pundit has a model for what the typical member of the public thinks and directs their advice accordingly. For many reasons, the pundit’s model is likely to be wrong. The readers of various pundits are a self-selecting sample, and the pundit’s intuitive model of society is distorted. Even if they surveyed their readership, which is difficult to do right anyway.
So all advice like “People should do more X” is suspect because the advice is based on the author’s assumption that the readers are in class A but they could easily be in class B, who maybe should do less X, possibly because X does not work for class B people in general, or because class B people are generally likely to have done too much X and maybe need to lay off the X for a while.
This is a problem for general advice. The living world, and especially the social world, is made of adaptive systems, which are characterised by complicated control problems where the answer is often not “always more X” but rather “Hit the sweet spot of the perfect amount of X, not too little, not too much” where the “perfect amount” is a function of the context.
1 Alpha
Rohit draws a distinction of use in Beware The Idle Kantian where he describes the Idle Kantian problem:
Categorical imperative misapplied from thinking the question “if everyone did as I, would it be moral”, into the statement “if everyone did as I think it would be moral”.
To be an Idle Kantian is to believe that any proof that an action isn’t a universal law means that it’s not lawful and should be toppled forthwith.
The insight that we have trouble working out what the universal aspects of our platform are is nice. Indeed, sometimes we are OK at this:
… There are entire professions dedicated to not being Idle Kantians. Finance is one, where your entire edge often revolves around being able to do an action such that, if everyone else did it too, it will be made irrelevant. Writing is another, where the novel you write or the essay you write has to be one that you wanted to write, not the one that would be best if everyone else wrote too.
I believe the verbal shorthand for this is seeking alpha.
Sam Atis, in Most advice is pretty bad, shows the needle we might try to thread here:
I think good advice has three main components:
- It is not obvious
- It is actionable
- It is based on some true insight
2 Incoming
Almost all advice is contextual, yet it is rarely delivered with any context.… Without understanding the context, the advice is meaningless, or even worse, harmful. If those folks had followed their own advice early on, they themselves would likely have suffered from it. It is hard to escape this trap. We may be the culmination of our experiences, but we view them through the lens of the present.
Spencer Greenberg, in a newsletter, hypothesizes:
There’s a hierarchy of skill levels regarding what people do when someone comes to them for help solving a problem:
- Reaction: suggest what first comes to your mind after the person tells you about their challenge.
- Personal anecdote: say what worked for you in a similar-sounding situation.
- Patterns: give advice about examples of what tends to work for others in similar-sounding situations.
- Empathy and Patterns: listen closely and attentively, empathetically ask questions, and explain what you’ve seen work for others in a similar situation.
- Empathetic Analysis: empathetically ask questions, listen to their proposed options for what they might do, propose one to two other potential options, then discuss the most important advantages and disadvantages of all options.
- Model: empathetically ask questions until you have a causal model. Explain it to the person to see its weaknesses. Get their help improving it. Use it to predict a few things that may work, and discuss the pros and cons of each of them. Encourage the person to pick what to try first/second/etc.
DRMacIver’s Notebook: Disingenuous Advice
The characteristic features of this sort of disingenuous advice are:
- It’s not exactly bad advice. It’s important that it could, in theory, solve the problem.
- It tends to be quite one-size-fits-all. It shows very little engagement with the actual problem—if you were going to put in enough effort to tailor your advice, you’d probably make at least some effort to give reasonable advice.
- It tends to be quite simple, so as to make it seem easy. It’s the sort of thing that you could very naturally use the word “just” in.
- It seems to encourage closing off further conversation until you’ve tried to follow the advice. Reasons why the advice won’t work sound like excuses.
- It is, in fact, very hard to follow this advice and even following the advice isn’t guaranteed to work.