Socratic CODES to help others unlock the solutions to their problems

By Peter4444 @ 2022-09-17T17:28 (+21)

I often enjoy helping others solve problems - moving from messy thoughts or confusion to concrete next steps and action plans. Happily, I've recently had several people in the EA community tell me they think I'm unusually good at this. With some help, I've decomposed the things I think I do quite well, then found a framework that captures most of it and is hopefully easy to remember.

Most pieces of this framework come from other people. Some I intentionally adopted; others come naturally to me or emerged in some kind of accidental, perhaps subconscious, way. I’m especially grateful to Neel Nanda, Nate Soares, the folk at CFAR, and Ben Kuhn[1].

This project will have been worthwhile if the framework helps someone else have more productive conversations, or leads to a discussion that results in more productive conversations (e.g. by clarifying unclear parts of this post or pointing out missing pieces - I would love to hear where you disagree or what you think I’m missing). It's also a way of clarifying what I think and remembering things that seem to work for me.

The Socratic CODES acronym highlights the characteristics that (I think) make problem-solving conversations likely to be successful. I explore each in a fair amount of detail, so feel free to skip around.

For easy reference, let's say I'm trying to help you.

Summary

In problem-solving conversations, I think it’s usually beneficial to be:

Socratic: avoid making statements; instead, ask questions.

Neel Nanda explains this method best:

[M]y favourite technique by far is the Socratic method. Rather than explaining my idea for a solution, I’ll ask leading questions that can lead them through the thought process I went through. If this goes well, then they’ll generate thoughts such that my solution makes sense. And if this goes even better, they’ll give an unexpected answer to one of my questions! And this is awesome, because it shows I misunderstood something, and we can dig into that, with no harm done. So, either we get to the same point, but in a way that feels more intrinsically motivating, or it fails gracefully and I discover that I’m wrong.

When I say Socratic, I mean to refer to the combination of three qualities:

Concrete: be extremely specific and use examples.

Being very specific and using plenty of examples, real world or hypothetical, can hugely improve a conversation. I’ll call this pair of virtues ‘concreteness’. Concreteness can help to clarify a person’s existing claims, ideas, intuitions, or concerns, as well as supporting them in generating new ideas. Here are some concrete examples (oh, so meta!) of concreteness in action:

 

  1. Clarifying problem and causes: You say you want to speak confidently to large groups of people but notice that you sometimes feel distressed and stumble. I might ask you:
    • what exactly happens when you stumble;
    • for some examples of times when you stumbled and some examples of times you didn't;
    • if, when you feel distressed, you always stumble, or if sometimes other things happen;
    • what specific reasons you've had for wanting to speak to large groups of people.
  2. Distinguishing between possible views: You say you think x might be a good idea but are concerned because it seems very socially costly. I might ask if you're concerned because (a) it's actually very socially costly or (b) it's not actually very costly but feels that way (i.e. feels awkward).
  3. Suggesting a hypothesis: I think perhaps you're frustrated about a repeatedly procrastinated task, even though you could easily get everything done, so I ask, "I sometimes find[2] that I keep procrastinating a task I have pushed back and back, then feel slightly awkward after I finally do it and it takes all of 5 minutes! Could something like that be going on here?"
  4. Strengthening a plan: You say you'll definitely get some task done by the end of the week; I am unsure about this. I might ask, "Suppose it's Sunday afternoon and you're sitting at home, looking over your todo list. Suddenly you realise you didn't get that task done. Vividly imagine this situation. How unlikely or shocking does it seem to you?"
  5. Visualising costs and benefits: I think a plan has little downside, but I think you think there's a big downside to it. I might ask, "Suppose you did this and it went horribly wrong. If you don't mind, could I ask what the worst thing is that you imagine happening? The more detail, the better. … And what if it went exceptionally well? What would that look like?"

Open: be warm and honest

Without meaning to blow my own trumpet, it seems I sometimes demonstrate a kind of openness that helps people in turn be willing to open up to me about uncertainties, bottlenecks, mistakes, and so on. But what does openness mean? My guess is that it has two main components.

First, I like to think I am quite a warm person. That is, I am often able to be friendly, empathetic, and caring. I am far from the warmest person I know, and I can sometimes not be friendly, empathetic, or caring at all, but at my best I think I do a decent job. At my best, I hope that my friendliness makes me approachable, my empathy helps others feel safe, and really caring about solving a person’s problem helps that person feel supported and optimistic.

I think the ‘caring’ part can be broken down into being driven and being supportive; more on each soon.

If you would like to be warmer, I’m not sure what to advise. It probably depends on your situation. But here are some off-the-cuff suggestions:

  1. Find reasons to genuinely care about your conversation partner.
  2. Practise gratitude. It is well-documented that thinking of things to be grateful for can make you more happy. Happier people are often also warmer in how they interact with others.
  3. Take a step back and be excited! Out of all the possible ways the world could be, you and your conversation partner exist. Not only that, you’re in the same room together and can communicate in a rich way. Your partner has unique and fascinating experiences to share, and you’ve been given this unique moment to get to know them - and, perhaps help them through a tough problem. This is extraordinary, crazy, and cool!

Second, I tend to be honest and direct (like many in the Effective Altruism community). This is not always convenient, and probably is not always optimal in terms of helping others feel better. And I can’t say that I’ve carefully weighed the costs and benefits of directness over a more indirect or cautious approach; I just much prefer being direct. But one potential benefit is that directness can build trust. Suppose I tell you something that seems true but unflattering, like, ‘Yes, it looks to me like that was a poor choice. But you can fix this situation.’ I now provide you with evidence that I am willing to say things that are true but unflattering. To some kinds of people at least, this is a positive thing.

Another form of honesty is self-disclosure: mentioning something private, such as something that I personally struggled with, big or small.

Self-disclosure can be bad. If, upon learning a person is grieving the loss of their grandfather, you tell them you know how it feels because you remember the enormous difficulty you went through when your father died, you may unwittingly shift the focus onto you and engage in a competition for who has dealt with the toughest situation. (This is, of course, not the only case in which self-disclosure may be inappropriate.)

However, it can also be a useful tool, because it can show three things:

  1. A willingness to disclose. By making yourself vulnerable in this way, you may both earn the trust of your conversation partner and grant her permission to be vulnerable too.
  2. A shared context. While no two people’s feelings are exactly the same, a former chain smoker can better understand a current chain smoker than can a person who has never smoked
  3. A problem-solving approach and reason for hope, if you solved the problem that you disclose. If you could solve a problem that seemed difficult, perhaps your conversation partner could too, and perhaps they can trust you to help them through.

Driven: we are going to solve this problem

Assume a solution exists

Sometimes, a person does not want to solve a particular problem. Perhaps they want instead to vent about it, because it has been frustrating them. In appropriate amounts, venting can be very useful and I have nothing against it.

But if I’m talking to a person who is interested in solving a problem, I like to throw myself into the challenge. There is a solution, and we are going to find it together! Then we’re going to do our best to make it work. 

This is not mere wishful thinking. I have noticed that many of my and others’ default behaviours and approaches are far from optimal. Especially when we’re specifically talking about problems - things that seem wrong - it’s very likely that there’s some big improvement to be found, and often the improvement is surprisingly straightforward. When I say ‘There is a solution,’ I do not necessarily mean that there’s a perfect solution, just that there’s a way of resolving the problem that is significantly better than what you would do by default. Solutions aren’t always beautiful; they are often just a set of next steps to gather information, e.g. ‘Cold message 10 people in ___ field and ask ___ and ___. If no-one replies, then ___.’ But that could be a lot better than what your partner would otherwise do. All that is left is for us to find it and for you to implement it.

You might say that there’s not always even a significant improvement. Suppose Bob has carefully considered all his options in a difficult situation and decided on some least-unsatisfactory option. In that case, I might not be able to help him find something much better. This is true in theory, but in practice I have found these cases to be rare. In practice, I find it is best to doggedly assume the existence of a solution (i.e. significant improvement over default) and set out to find it.

Why is this helpful? It seems that you are more likely to succeed at most goals when success is a background assumption than when the question, ‘Will I succeed?’ is explicit in your mind. If a solution is assumed to exist but you can’t help your partner find it, it’s on you and you have failed. (This happens sometimes, and it’s OK, but you own it.) If instead you’re unsure whether the solution exists, you can always say you tried but there just wasn’t a solution; then it’s the world’s fault, and not yours, and you might not be quite so persistent. Persistence helps you find creative solutions and expand your partner’s perceived option set - more on this in the next section.[3]

I think I made the shift from asking if there was a solution to asking where it was without even realising it, some time after I noticed that, when I really looked for a solution there almost always was one.

Really care about finding a solution 

Apparently, it can feel like I ‘adopt’ my conversation partner’s problem for the duration of the conversation. There are two ways in which really caring about solving the problem is useful for helping your partner reach a solution. First, it makes me more likely to ask good questions, probe when it makes sense, and think better in general. Second, at my best, I think I give off energy, optimism, and excitement, which may be contagious and help bring that solution to fruition.

It’s hard to explain what it feels like internally to really care about finding a solution, though one concrete example is given in the Expansiveness section when I introduce the ‘resolve cycle’ technique. At my best, I see a problem someone is having, see the suboptimal way things are likely to go by default[4] and feel an urge to find a better way. My guess is that it’s some combination of wanting to solve a puzzle and defy a challenge, having a taste for solutions and efficiency, empathy and desperation. My exact motivation will be different in every case. Of course, this relies on believing there is probably a solution: a better way (above).

Make it happen

Often, I know that I have a problem, I know roughly what to do about it, yet the problem remains unresolved for some time. This is OK if I’ve deliberately deprioritised implementing the solution - there is only so much effort a person can exert every day - but usually that’s not the case and it’s just a massive waste.

I’m keen to prevent my conversation partner from suffering in the same way. My aim is not to tell you what to want, or even prescribe what to do; but I care that, IF you want something AND think that some solution is a good and worthwhile way of getting it, you implement that solution. In general, this will not happen automatically. I repeat: this will not happen automatically.

So I tend to follow something like the following three-step process:

  1. Double-check that the solution is one my partner actually likes and wants to go for. This means that they have a clear view of the possible costs and benefits, believe the benefits outweigh the costs, and believe that it is a better approach to solving the problem than what they might otherwise do. While I do my best to help my partner come up with their own solution, occasionally I end up overconfident in my own guesses at solutions or assuming that they want the same things I might want. So I want to be sure I haven’t missed something.

    “Do you really want to go with this solution and make it work? You're allowed not to want that! But if you want to do this, I want to help you figure out how to make it as likely as possible[5] to succeed.”
  2. Figure out failure modes together. This often involves doing a pre-mortem - asking my partner to assume that, at a particular future time, this plan has failed, then figure out why, and sometimes suggesting reasons I’m concerned it might fail.
  3. Ask if they want accountability or to find ways to make those failure modes less likely.

Expansive: there are a lot of possible options

Many people artificially constrain their options on what they can want and how they can get there. That is, they think there are fewer options than there actually are. (Spoiler: there definitely aren’t only two options! But the third, or thirteenth, option can sometimes be hard to see.) To be expansive is to introduce a wide range of possible options for consideration - options for what to aim for, options for potential solutions, and options for ways to make sure a solution happens (e.g. different types of accountability system).

Expansiveness goes hand in hand with drive, since you are much more likely to do this if you believe there is a solution (i.e. big improvement on default approach to problem) and really care about finding it. Let’s see how they interact.

One part of being driven is to assume that there is a solution. Being expansive involves assuming that there are a lot of possible solutions. So the expansive person will then suggest - or, better still, help his partner generate - a wide range of possible solutions. Then, being driven to find a solution, he will help his partner figure out which of these solutions is best for her.[6]
 

Concrete examples of expansiveness

Expansiveness in finding a new solution

(My caffeine improvement was clearly less transformative than Reggie’s door-knocking; I include both to show that expansiveness can help solve problems big and small.)

Expansiveness in different viewpoints from which to see things

Expansiveness in ways to implement a solution

The ‘resolve cycle’ - a 5-minute expansiveness technique

Your friend has lost a crucial presentation on his laptop that he is due to give in 1 hour. Elon Musk tells you that he wants 15 distinct possible solutions in the next 5 minutes. To incentivise you to generate these, he offers you $1 billion dollars if and only if you succeed.[7] Picture the scene. You rush to find pen and paper or something to type on, then scribble down as many ideas as pop into your head. You’re not too worried about the quality of your possible solutions, at least at the start, because what matters is that you have some passable ideas; you might later filter out ones that seem especially high-effort or unlikely to succeed, if you come up with more than 15.

You might start like this (though with each option written in shorthand):

  1. Check file isn’t backed up anywhere by using multiple search tools and asking friend if he used any backup services.
  2. Check trash
  3. Recreate presentation from scratch with him in remaining time

This is where some people would stop. But $1 billion dollars is on the line, so you keep going:

  1. Find out if he emailed draft versions of the presentation to anyone; if so, find newest version by contacting person emailed, then modify if time
  2. Find an expert in data recovery (probably won’t have enough time)
  3. Message ten friends living nearby who might know about the subject to see if they could give the presentation instead
  4. Find out if the presentation could be given remotely, then message people with relevant knowledge around the world

Some people would stop there, but you go on:

  1. Offer a bounty to anyone who can give the presentation; publish it online.
  2. Find out if other presentations are to be given alongside your friend’s. If so, ask to rearrange schedule order, leaving time for friend to find or recreate presentation.
  3. See if friend knows content well enough to do presentation without slides
  4. Find out if friend or someone else can do an alternate presentation that would be at least approximately as helpful for the audience.

‘Ugh! What more?’

  1. Find a document on the presentation’s contents. Ask 5 different people to make slides simultaneously for each of the 5 sections, then put together.
  2. Turn the presentation into a discussion session or Q&A, requiring less preparation.
  3. Learn how to back files up better, then deliver a self-deprecating tutorial on why and how to make backups in your presentation time
  4. Deliver a much shorter presentation, then announce a much-needed break

Many of these solutions probably wouldn’t work or would be undesirable, but by doing this exercise you have forced yourself to come up with 15 starting places. Maybe you discover that the eighth option - the online bounty - is the best option; most people wouldn’t even have considered it, but you did because you pushed yourself to expand your option set. Maybe none of these solutions is useful and your friend just has to apologise and skip the presentation. In that case, you wasted 5 minutes, a small cost compared to the significant possible benefit.

Spoilers:

  1. You don’t actually need $1 billion dollars on the line to do this. 
  2. You don’t actually need to sit down for 5 minutes and generate 15 options to get into the habit of entertaining unusual ideas and aiming to find a lot of possible solutions.
  3. By entertaining unusual ideas, you encourage your partner to do the same, and they may come up with better ones than you. It is, after all, their problem.

Non-exhaustive list of ways to encourage others to consider several new perspectives

Experiences that seem to have made me more expansive

  1. Being very suspicious of the word ‘impossible’. Using the heuristic, ‘There is always another option.’ When the world seems to have presented me with two undesirable options, I do my best to defy that perception and find a third, better workaround.
  2. Practising ‘resolve cycles’ like the above. Enjoying the game of coming up with unusual solutions that might be rubbish, rather than worrying about the quality of each solution I generate.
  3. Trying weird things and seeing them sometimes work. This includes ‘upside bargains’: low-cost solutions that are unlikely to work but have little downside and are great if they do succeed.
  4. Gaining more familiarity with the EA community, and especially its willingness to spend money and time on unusual but high-upside bets.

Your experiences may vary.

Supportive: this is hard and we are in it together

It is hard to write about being supportive without sounding cliché or arrogant, and it’s an area in which I’m certainly not an expert, but let me try anyway.

At my best, on the occasions when I have a long, problem-solving conversation with someone, I think the person quite often comes away with the impression that I deeply care about them. This may be true especially, but not only, if I deeply care about their goals. Whether EA-related or otherwise, the attitude I often want to convey is, ‘You’re on my team. I want you to win.’ If this is felt, it can help, alongside other factors like shared context and problem-solving drive, to create a high-trust environment.

Many goals, let alone EA goals, are hard to achieve. It’s tough to have a conversation with someone in which you admit to something that is holding you back but don’t know exactly what to do about it. It’s tough to think carefully about root causes; about what solutions will and won’t work. You might have earnestly tried a lot of things and had none of them work for you. You might have worried about this problem but felt powerless to change it. The feeling I get is something like, 'It’s not just that I want your problem to be solved, but that I want you to solve it and your life to become better.'

For those tougher times, it’s nice to know that there’s someone on your side who cares about you deeply, believes in you, and wants you to succeed.

Debugging, not blaming 

I aim to stick firmly to a ‘debugging’ framing, both with myself and with others. I find the analogy with debugging code useful. Upon noticing an error in a software program, you could:

  1. Angrily delete all the code and start from scratch. The coder must be stupid. Why else would there be a bug?
  2. Find the cause of the bug and fix it.

Now, (a) sounds – and is – harebrained. Clearly, (b) is the better option. The fix needed and cause identified in (b) might be deep, but the commitment is to focusing on finding the best, i.e. most appropriate, cause to target, rather than being emotional and assuming some deep moral failing. When talking to you about your problems, I refrain from blaming you, not because you shouldn’t be responsible but because responsibility isn’t the point; we're simply not in the realm of fault/blame, but in the business of seeing what's wrong and how to fix it.

Concrete ways to be more supportive

I’ll end this section by quoting from Nate Soares, who finds that thinking of people as striving monkeys improves his empathy. I have not tried this technique, but nonetheless find this passage inspiring:

We're monkeys, and we're fairly fragile monkeys at that.

So you don't need to beat yourself up when you miss your targets. You don't need to berate yourself when you fail to act exactly as you wish to act. …

Don't berate the monkey. Help it, whenever you can. It wants the same things you want — it's you. Assist, don't badger. Figure out how to make it easy to act as you wish. Retrain the monkey. Experiment. Try things.

And be kind to it. It's trying pretty hard. The monkey doesn't know exactly how to get what it wants yet, because it's embedded in a really big complicated world and it doesn't get to see most of it, and because a lot of what it does is due to a dozen different levels of subconscious cause-response patterns that it has very little control over. It's trying.

Don't berate the monkey just because it stumbles. … The things we're trying to do are hard. So when the monkey runs into an obstacle and falls, help it to its feet. Help it practice, or help it train, or help it execute the next clever plan on your list of ways to overcome the obstacles before you.

One day, we may be able to choose our cognitive patterns at will, and effortlessly act as we wish. … But we aren't there yet. We're not gods. We're still monkeys.

How personal problem-solving can sometimes feel
 

  1. ^

    Caveats:

    1. I've only recently put these pieces together, so I have mainly tried it only with students or recent graduates in the EA community, mainly asking questions about their career or productivity. It might not work well elsewhere.

    2. I'm not a psychologist and this is not the result of detailed scientific study. The sample size is still small (perhaps 7 conversations that seem to have been impactful) and I still don't know how all of them have turned out.

    3. This is mainly my (and some friends') interpretation of what causes me to have productive conversations. It could easily be inaccurate or misleading.
    Sometimes, people don't want to solve a problem. Just because you think they have some problem or they express some issue, it does not follow that they want to solve it then and there!

  2. ^

    Another great suggestion from Neel Nanda is to suggest hypotheses, wherever possible, in the first person, talking of your own experience instead of jumping straight into what you think might work from them.

  3. ^

    The careful reader might worry that simply trying to find ‘a significant improvement’ would lead me to stop pushing forward once we found a decent solution, instead of hunting for an even better solution. But often a decent solution is all that is needed to get started; and besides, usually we end up finding multiple solutions and I find in myself a strong urge to discover an excellent solution, not just a good one.

  4. ^

    suboptimal, that is, given their goals

  5. ^

    This is clearly not to be taken entirely literally, since you will have other goals and don’t want to increase the probability of solving this one problem at the cost of everything else.

  6. ^

    For those curious about the distinction between being driven and being expansive, consider these unusual but possible cases. Driven but not expansive: Alice has a problem; Bob, caring a lot about solving it but only thinking of two possible solutions, neither very compelling, helps Alice figure out which of the two is least bad and implement that solution. Expansive but not driven: Alice has a problem; Bob casually thinks up 10 possible solutions, then heads out because he’s bored of the conversation and doesn’t care very much which is chosen.

  7. ^

    For more information on this technique, please see CFAR handbook, ‘Resolve Cycles’, where I first read this analogy. Asking, 'Have you tried just solving it in 5 minutes?' may sound silly, but the technique has had a surprising success rate for me and several friends, and can be used for more than just generating options.