How to give good advice

Hi everyone,

This is a free issue of Overthinking Everything. There are also paid issues, with the two going out on roughly alternating schedules. Paid issues tend to be a bit more speculative and at the cutting edge of things I’m still trying to figure out for myself, while free issues tend to be about topics that I feel I’ve already figured out reasonably well. The last paid issue was How does good taste work?, in which I reflect on the nature of taste and talked about my desire to improve my taste in fiction, and prior to that I wrote Some thoughts on ethics and narrative, which was about how much of our understanding of the world we learn from fiction.

It may seem that my plan to write this month without a schedule is going badly, but actually it’s going great, it’s just that most of that writing has been on the notebook blog, where I tend to write things that are more fragmentary. I’ve been writing something almost every day (and on days I haven’t, I’ve written something elsewhere). I should probably have mentioned that I was doing this in an earlier newsletter issue, but didn’t. Sorry.

One thing I wrote recently was “Would you like some advice?”, in which I implore people to ask if advice is welcome rather than just launching into it in response to someone telling them about a problem.

Today I’d like to talk about what to do if they say “Yes”, because I think most people are quite bad at giving advice, and it’s mostly the result of not knowing how to do it well.

Specifically I’m going to be talking about the sort of advice where someone has a reasonably well defined problem that you don’t have much context for, and you want ot suggest things that will help them solve it.

I think most people should be able to do this well, and also that most people don’t, so hopefully I can give you some useful pointers.

You need to be more useful than a duck

Yesterday I wrote First decide what’s good enough, in which I suggested separating your plans on how to do something into two parts by asking the following questions:

  1. How good do you need the thing to be?

  2. How good would you like the thing to be?

You should only worry about the second once you’ve ensured the first.

When giving out advice, your benchmark for how good you need to be is that you should be more useful than a rubber duck. Most people fail to clear this bar.

If you have no clue what I’m talking about, I can recommend this great piece by Siderea explaining rubber duck debugging as a general practice, but one of the most salient parts is referencing Ask the duck by hwrnmnbsol1:

Bob pointed into a corner of the office. "Over there," he said, "is a stuffed duck. I want you to ask that duck your question."
I looked at the duck. It was, in fact, stuffed, and very dead. Even if it had not been dead, it probably would not have been a good source of design information. I looked at Bob. Bob was dead serious. He was also my superior, and I wanted to keep my job.
I awkwardly went to stand next to the duck and bent my head, as if in prayer, to commune with this duck. "What," Bob demanded, "are you doing?"
"I'm asking my question of the duck," I said.
One of Bob's superintendants was in his office. He was grinning like a bastard around his toothpick. "Andy," Bob said, "I don't want you to pray to the duck. I want you to ASK THE DUCK YOUR QUESTION."
I licked my lips. "Out loud?" I said.
"Out loud," Bob said firmly.
I cleared my throat. "Duck," I began.
"Its name is Bob Junior," Bob's superintendant supplied. I shot him a dirty look.
"Duck," I continued, "I want to know, when you use a strap hanger, what keeps the sprinkler pipe from jumping out of the strap when the head discharges, causing the pipe to..."
In the middle of asking the duck my question, the answer hit me. [...]
I turned to look at Bob. Bob was nodding. "You know, don't you," he said.

When someone explains a problem clearly enough to understand, they often explain it clearly enough that they can see the solution, because by spelling out all the details you draw your attention to the important parts of the problem, and build a coherent picture that makes it obvious how to solve it.

This doesn’t always work, but it works a lot. In contrast, many other commonly employed advice giving strategies fail far more often than they succeed.

An interesting example of this crossed my path recently, in which professors Agnes Callard and Ayelet Fishbach discuss advice giving:

One thing that comes up in this is that Fischbach cites some research showing that giving advice is more useful than receiving advice. They got together a bunch of unemployed people and did one of two things:

  1. They gave them advice on job seeking.

  2. They asked them to give advice on job seeking.

The second proved significantly more useful than the first. Their hypothesis for why this was the case was that the job seekers already knew what to do but needed motivation to do it.

My hypothesis for why this was the case is that most people’s job seeking advice is shit, and I’d expect academics to be particularly bad at advising unemployed people on how to find a job, because they have very different backgrounds and the academic job market is very weird.

In contrast, in order to give advice, the unemployed people had to first articulate the problems they were trying to solve, basically taking them through an episode of rubber duck debugging. This was more useful than just being told what to do, because it could draw on an actual understanding of the problem.

This is what I mean by being less useful than a rubber duck: Unless you give people the space they need to properly articulate the problem, you’ve taken away a valuable opportunity for them to work through the details themselves, and also you probably don’t understand the problem well enough to help.

How to take the rubber duck’s job

You can be more useful than a rubber duck, I have faith in you. This starts by doing what the rubber duck does, only better.

Rubber duck debugging involves someone explaining their problem to you, but people aren’t actually very good at explaining things (see How to explain anything to anyone), and a duck doesn’t give them any feedback on the quality of their explanation. A human can, by asking questions.

The two most useful types of question you can ask are for clarification where you don’t understand, and for elaboration where you feel like something is missing. e.g. “What are you actually trying to do?”, “What have you tried?”, “Why doesn’t (thing that you tried) work?”, “I didn’t understand XYZ, what did you mean by that?”.

If you can’t think of any questions to ask, try repeating their description of the problem back to them to check your understanding. Leave plenty of space for them to say “No, that’s not quite right” or “Actually it’s a bit more complicated than that”.

Also, feel free to push back on things that seem factually wrong to you - it’s better not to start an argument, but asking “Wait is that right?” or “Why do you think that?” style questions is usually helpful.

This back and forth draws on their knowledge of the problem they are trying to solve, and gets them to properly flesh out and articulate what the difficult parts are.

Rubber ducks don’t actually give advice

Actually none of the above was about how to give advice, that’s all coaching. Sorry, I fibbed. Rubber ducks can’t give advice, because giving advice requires making suggestions and, well, they’re rubber ducks.

I do think you should probably learn to coach, and I think learning to coach will automatically improve your ability to give advice. Also understanding the basics of coaching is important for what I have to say next, even if you never do it.

Agnes Callard’s “Against Advice” makes the following distinctions (she also makes these in the above video), which I will also follow:

Let me make a three-way terminological distinction between “advice,” “instructions” and “coaching.” You give someone instructions as to how achieve a goal that is itself instrumental to some (unspecified) further goal—here is how someone might get to the library, if for some reason she wanted to go there; this is the way to put toner in a printer, etc. Coaching, by contrast, effects in someone a transformative orientation towards something of intrinsic value: an athletic or intellectual or even social triumph.


As I’m using the word “advice,” it aims to combine the impersonal and the transformative. [It tries to provide] the kind of value she would get from the second, but she wants it given to her in the manner of the first. But there is no there there. Hence the advice-giver is reduced to repeating reasonable-sounding things she has heard others say—thoughts that are watered down so far that there’s really no thought left, just water.

I think this is a good set of distinctions, and very accurately describes a pitfall of giving advice, but is much too strong. It’s perfectly possible to give useful advice, many people do it all the time, and Callard is herself doing it right in this very piece with her advice of “Don’t offer advice”.2

For example, here’s a piece of advice: If you’re stuck on a problem, you could try explaining it to a rubber duck or journaling about it. If that doesn’t work, you could try hiring a coach, if you can afford it, or asking a trusted friend if they’re willing to talk it out with you. If none of those work then uh… sorry, I’m out of ideas I can give you without more specifics.

Call this the default duck advice. The default duck advice has a couple of key features:

  1. It’s pretty general.

  2. It uses the word could instead of should.

  3. It suggests a number of things to try.

  4. It makes no promises that it will work.

  5. It acknowledges that it will not work for everyone.

Here’s another example of me offering advice the other day:

This advice is for a more specific problem than the default duck advice and is based on things that have worked well for me and that I’ve seen work well in the past.

These all exhibit what I think is the fundamental characteristic of good advice: They are offering a set of ideas of things to try that may or may not work, but have a pretty decent chance of working and being worth it.

Have you tried giving better advice?

I claimed you have to be better than a rubber duck. This is only somewhat true. Being better than a rubber duck is a lot of time consuming effort. Really what you have to be better than is the default duck advice. You have a certain amount of time and trust to make use of when giving advice, and if you don’t have anything better to say you might as well give them the default duck advice.

In order to do that, it’s useful to understand how advice works.

A useful rule of thumb is that all progress happens through trial and error. You try something new and either it works, and you’ve made progress, or it fails, and you learn from its failure in a way that improves your future trials.

Callard’s distinctions relate to trial and error in different ways:

  1. Instruction provides people with something to try which has a low chance of error.

  2. Coaching (at least, of the sort a duck or superior ducklike entity can do) improves the recipient’s ability to learn from error, by giving them space and support in which to do so

  3. Advice can do several things.

    • negative advice (“Don’t give advice”, “Don’t eat the yellow snow”) advises you on things not to try, improving your future success rate.

    • positive advice (“Get a coach”, “Write every day”) suggests things to try that you might not otherwise have tried.

    • diagnostic advice (“You might be undervaluing card draw”, “Check if this is a side effect of any of your medications”) suggests ways of learning from errors that have already occurred.

Generally negative advice is better replaced with diagnostic advice, or requires heavy explanation of the logic behind it, because if someone follows it religiously and it’s wrong (or they misunderstand!) they’ve lost a valuable source of things to try. For example, in How to do everything I talked about my rather embarrassing realisation that you had to vacuum or sweep before you mopped. I eventually traced the root of my belief there back to a cleaner once telling me that I shouldn’t vacuum the bathroom! In retrospect I think she was actually telling me the much more reasonable advice that I shouldn’t do so while wet, but because I internalised “Don’t vacuum the bathroom” I never tried the thing that would have worked.

Both diagnostic advice and positive advice have the potential to be very useful as long as you signal that they are limited things that won’t always work. My diagnostic advice is that if people are getting cranky at your advice or, worse, trying to follow your advice long past the point where they should have given up, you might need to start caveating it better.

My positive advice is that when giving advice you should think specifically about things they are not likely to try or understand on their own, and err heavily towards suggesting things like that. Novel advice expands their toolkit of things to try in a way that hearing the same thing over and over again will not. If they already know the answer, advice is not likely to be very helpful, and they’re more likely to need coaching than advice.

Sometimes the best advice…

Is to shut up and say nothing.

There are a lot of cases where the default duck advice is not that helpful, and offering no advice at all is the best thing you can do. The main ones are:

  • They’ve already done it.

  • Someone else has already told them to do it.

  • They haven’t asked you for your advice.

If you’re following my advice on asking “Would you like some advice?” before offering advice, you probably shouldn’t bother offering at all unless you’re going to give something better than the default duck advice.

Checklist advice

There is a specific category of advice-giving that is useful but requires a little bit of careful handling, which is roughly the advice of “obvious things you should have tried”. One might call it “checklist advice”, because it’s the checklist of things that anyone with this problem should really have tried before getting stuck. It’s easy to miss things, and different things are obvious to different people, so it’s often worth going through the list.

My main piece of negative advice on this is that you probably shouldn’t give it to strangers on Twitter, because too many other people will be doing the same, and it’s rarely worth it. At the very least, check their replies first.

My other main piece of advice is to make it really very clear that you are doing this. Starting the checklist advice with something like “OK, so you’ve probably already thought of some or all of these, but just to double check…”. If you’re only going to provide checklist level advice, roll it into the initial question: “I don’t have any deep insight into this problem, but would you like me to go through some basic things to help you confirm you’ve not missed anything?” works well.

Also, checklist advice is more useful in coaching than impersonal advice giving, because their explanation of why the obvious thing doesn’t work will reveal interesting facets of the problem that you hadn’t been able to elicit from them beforehand. Once you’ve gotten through the duck phase of coaching, this may be your best bet.

Or do something else, I’m not the boss of you

Don’t give advice if you’re going to be upset if it’s not followed, it doesn’t do anyone any favours.


If you got to the end of this issue and thought “Man, I really want David to give me some advice about a problem”, I’m still available for one off coaching sessions and you can book one using calendly. I’m also in principle available for recurring clients, but I only want to take on another one or two at the moment, so I’ll likely say no unless I think I’m a really good fit for what you want.

This newsletter also has a Discord community attached to it, where we both talk about newsletter issues, chat about unrelated stuff, and generally try to help each other out figure life out. You can read the community guide here if you want to get a feel for it, or if you were instantly sold on the idea of such a discord server, you can just join us there by clicking this invitation link.

Also, if you liked this issue, you are welcome and encouraged to forward it to your friends. If you’re one of those friends it’s been forwarded to and are reading this, I hope you enjoyed it, and encourage you to subscribe to get more of it!

Finally, this is the preview image for this post in full:

It comes from this flickr post by IQRemix.


No, I don’t know how to pronounce it either.


Probably she would argue that it’s instruction rather than advice, but I don’t think the argument holds because the injunction she is providing is a values change rather than teaching someone how to solve a well defined problem.

Loading more posts…