On Giving Advice

I’ve been thinking about what it means to “give advice” to someone. The context I’m thinking of is, say a person looking back at their life and thinking about what they would have liked to tell their past self. Or people who have been successful trying to distill their crystallized nuggets of wisdom to other people. The point is, you’re trying to give information to someone about how they should be looking at the world, probably with respect to the decisions they’re making.

In my mind, the typical “giving advice” approach looks like this: You go through many experiences in your mind and you try to isolate common threads between things. You’re trying to look for a general conceptual framework that runs through many of the good decisions you make, and the resulting generalization is your advice.

I think that the typical approach towards giving advice really doesn’t work.

For one, when you do the thing I outlined above, what ends up happening is you often come up with a generalization like “starting tasks is half the battle”, which sounds quite trite. So people often file it away under their “boring advice” mental drawer. You could obviously spend time trying to explain the idea of In Defense of the Obvious, but I think that if the person can follow the type of argumentation you’re using to justify the meta-contrarian stance of still following Obvious Advice despite it being Obvious, then they’re likely already in a good position.

I think a slightly better reframe is to ask yourself “What sort of general decision-making procedure would I have needed in order to make more good decisions of the kind that I am thinking of?” rather than “What is the common thread underlying all these good decisions?”

Still, this doesn’t really change the fact that giving people this type of advice seems suboptimal, even if the receiver is in the right frame of mind to actually update, because generalized principles will likely miss out on the nuances that were afforded to you by the very experiences that you used to generate the advice in the first place.

In addition, both the “‘I already get it’ slide” and inferential distances will be working against you. So you really can’t even just give your conclusions, i.e. your generalized principles or decision procedure outright, because they’ll likely not value it as strongly.

What’s left then? I think this leaves us back to the general domain of pedagogy. In general, I’m fairly pessimistic about radically changing people’s worldviews in short amounts of time, and it feels like that’s what advice is generally trying to do:

You see a person who’s obviously got an incorrect way of looking at the world, and your goal is to rectify that, by changing their worldview, so they can make better decisions.

Giving advice, then, is less about taking people straight to your end point, but more about giving them a whole bunch of intuitions about every step you took. If you have enough time, you can probably speed them through the inferential steps / motivations that led you to derive whatever advice you gave them, so you can boost the receivers intuitive feelings towards the advice.

Otherwise, if you don’t have enough time to do an inferential distance speed-run, I think the best thing you can do is provide them with either an object-level skill or an intuition pump that allows them to iterate and self-improve.

Your goal then, with advice, becomes less about telling them where to not screw up (because that’s pretty hopeless), but instead to give them the tools such that they become more able to go out into the world and improve.

You want to be a catalyst for their further growth, rather than a dictator.



  1. Compare with my model:

    * we are optimizing value of a multi-dimensional, non-linear

    * we seem to have 2 main approaches (shift a little, improve local
    optimum; or big jump, then look for another local optimum).

    From this emerge roughly 3 levels of giving advice:

    1. **Local advice.**
    Info on which parameters to experiment with to move
    closer to a local optimum.

    2. **Jump advice.**
    Also examples of local optimums found by different people,
    and what they changed to make a big jump.

    3. **Total advice.**
    A model of how everything works,
    which predicts what happens when you change parameters,
    and makes it possible to find a global optimum.

    Always needed:

    – acknowledgement that the search depends on individual stuff
    (like the starting point)

    – remembering that when you get a good result,
    it might be harmful for others to repeat what you did
    (depending on their position)

    Liked by 1 person

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s