Since I moved into the Boston rationalist house, I've found myself having an overwhelming amount of conversation compared to my previous baseline. The conversations at Citadel tend to be fairly intellectual and interesting, but there is a lot of topic drift and tendency for entertainment over depth, which seems to be a fairly common pitfall. How can we optimize conversations and direct them towards areas of usefulness and insight?

There have been some previous discussions on this topic on LW, e.g. on useful ways to avoid low-value conversations or steer out of them. I would like to focus on the complementary skill of stimulating high-value directions in a conversation.

First of all, what makes a conversation high-value? There are several possible metrics:

  • people learning from each other’s expertise and experience
  • people getting to know each other better
  • exchange of advice and feedback
  • generating ideas and insights

All of these involve increasing the total amount of information available to the participants, either through revealing information that is already there, or through creating new information. This is more likely to happen in a topic area where someone has strong opinions or expertise, or, on the other hand, an area that someone finds challenging where they stand to learn a lot.

One effective way to steer a conversation is through asking purposeful questions. The questions should have sufficient depth to lead to interesting answers, but not be vague or put the other person on the spot. In that sense, a question like “What have you been thinking about lately?” is better than “What do you care about?” or “What are you terminal goals?”. It is better if the question leaves a line of retreat and doesn’t make the person feel low status if they don’t have an answer.

The types of questions that are productive and comfortable are generally different for group and one-on-one conversations. Two-person conversations are more conducive to openness, so one would be able to ask personal questions like

  • what memes have affected you strongly in the past or shaped your beliefs?
  • what has been important to you lately?
  • what has been difficult for you lately?
  • what eccentric things have you done?

Some questions are likely to lead to interesting topics in an N-person conversation for any N:

  • what have you learned recently?
  • what surprised you about experience X?
  • what have you been reading?
  • who are your role models?
  • I have been confused about X, does anyone have advice?

It is generally harder to steer a group conversation in productive directions than a two-person conversation, but the payoff is higher as well, since more people’s time is at stake. Since a single person has less influence in a group conversation, it’s important to use it well. Sometimes the most useful thing to do in a group conversation is to split it into smaller conversations. Asking someone about a subject that only they are likely to be interested in might be considered impolite to the others, but often leads to better separate conversations for everyone involved.

Questions do have limitations as a conversation tactic, and can sometimes result in awkward silence or a string of brief uninformative replies. If this happens, it’s handy to be prepared to answer your own question, which might inspire others to answer it as well. It is generally a good idea to have something that you’d like to talk about, perhaps something you've been working on or a concept that puzzles you, that you can bring up independently of whether and how people respond to your questions. Thinking in advance of topics to discuss with specific people is especially useful, e.g. relating to their past experiences or skill areas.

Do people have advice or good examples of directing conversations? Recalling the best conversations you've ever had, what made them happen?

New to LessWrong?

New Comment
35 comments, sorted by Click to highlight new comments since: Today at 8:50 PM

I find the best conversations happen in a room whose average I lowered by entering.

Recalling the best conversations you've ever had, what made them happen?

One of the participants putting in serious work ahead of time to collect their thoughts and present them.

Why should conversations be steered towards "insight"? What's wrong with entertainment?

There is nothing wrong with entertainment if entertainment is what you want. But people often switch to entertainment automatically just because it's easy and immediately socially rewarding.

It's like a difference between reading internet once in a while, and spending your whole day online. There should be an "internet" time, but also a "non-internet" time. Similarly, in social situations we want to have both "entertainment" time and "non-entertainment" time.

In the spirit of purchasing fuzzies and utilons separately, perhaps we should try to make separate time slots for maximum entertainment (fun and group bonding) and time slots for maximum becoming stronger. As opposed to having mediocre insights interrupted by jokes and board games.

(As an extreme example, two hours of group sex followed by two hours of high-quality quantum physics lessons would give you both more pleasure and more knowledge than four hours of playing Monopoly and having small talk.)

Maybe people switch to entertainment because it's what they'd prefer.

two hours of group sex followed by two hours of high-quality quantum physics lessons would give you both more pleasure and more knowledge than four hours of playing Monopoly and having small talk

Definitely not. That would be like eating a hundred tomatoes, then a hundred leaves of lettuce, and then saying that it's better than a salad. A mix of the two is more enjoyable, at least for me.

I guess that personal preferrence differs a lot in this.

Especially for a group that meets recurrently, it seems worthwhile to experience many different conversational contexts, some more structured and some less so. More structured contexts seem to be hard to sustain - the weekly sessions at my house focused on goal analysis often "devolve" into semi-related conversation.

Completely separating fuzzies and utilons doesn't seem possible, e.g. the quantum physics lessons are likely to be at least somewhat entertaining, but it makes sense to try to focus on one or the other for a particular activity.

I'm not saying that all conversations should be steered towards insight, but more of them should be, especially in the rationalist community, where the potential value of conversations is high. There is nothing wrong with entertainment per se, but conversations gravitate towards entertainment by themselves more often than not.

I would say that given the high correlation between rationality and sharing common interests and ways of communication, the potential entertainment value of conversations is high as well.

Recalling the best conversations you've ever had, what made them happen?

Alcohol.

X-D

You should have recorded them, might change your opinion. The effect of discussing subjectively deep shit is even worse with weed and LSD.

Actually, my criterion wasn't how deep it felt at that moment, my criterion was what I thought the next morning and in the days after.

Alcohol, in proper amounts, loosens people up. That helps with good conversations.

Many of the best conversations came from those I've developed strong social bonds with through alcohol.

I generally agree with you here, just wanted to point out that particular failure mode.

I just found out my liver enzymes are over the reference range (not because of alcohol), so the temptation to rationalize away good aspects of boozing is especially high at this moment.

ALT/AST?

And you can also have excellent conversations by plying the other party with alcohol while remaining sober yourself. :-) Though it is less fun.

I've been completely sober for the past five months, so that ratio wouldn't tell us much :)

Besides, I only know my ALT for now. Have no idea what's wrong yet.

The best conversations I had, measured along the "people getting to know each other better" metric, took place when I was using CouchSurfing to host people at my flat, a few years ago. Having a new person every week for a few days made me conscious of recurring social patterns, which in turn stimulated me to be more mindful and proactive about how I spent my time with these guests. My basic approach was to gradually move from impersonal questions, to questions about what the person is or does, to questions about what the person thinks and feels. There is research in social psychology that suggests that humans develop closeness and intimacy through a process of reciprocal and escalating self-disclosure. (See e.g. Aron et al, The experimental generation of interpersonal closeness, Pers Soc Psychol Bull, vol. 23, no. 4 (April, 1997), pp. 363-377.) More generally, these two blog posts on conversation skills contain some valuable advice.

Thanks for the great links! I agree that getting people to feel comfortable is really important for having good conversations. The list of questions in my post is intended to find personal topics that would be of common interest, but I think it's much more likely to work if the people involved are already feeling comfortable with each other.

Your CouchSurfing experience reminds me of a class at Burning Man where people were split up into pairs, and tried to take the shortest conversational path to getting to know each other. There was a built-in affordance to ask personal questions about feelings though, since everyone was in the class with that purpose. Did you end up developing patterned ways to ask strangers about they thought and felt without it feeling awkward?

It didn't feel awkward, probably because my approach was sufficiently coarse-grained: I didn't rely on any specific "lines" or "routines" to artificially generate closeness, but only on the general principle that the conversations should become progressively more personal and intimate.

I wish I could remind myself of the value of this principle more often, however. I often end up having superficial interactions with people I'd like to know better simply because that's the default way I relate to others. Recently, I started experimenting with Anki, using one side of the card to describe the relevant situation (e.g. "I find myself in the company of someone I'd like to get to know better"), and the other side to describe what I should do in that situation ("I ask questions that are progressively more personal and intimate").

Obligatory non-standard font grumble.

Something about being up late at night leads to good conversations, in my experience. So does revealing something personal about yourself (it pressures others to reciprocate).

I find that there is a self-signaling aspect to this: "if I'm staying up late to have this conversation, it must be interesting / important".

I find that the signaling aspect maybe accounts for 20-30% of the phenomenon.

I think that about 20-30% of the time (off the top of my head), my good conversations happen late at night because we stay up late because we're in a good conversation. But more often, we stay up late because we just don't want to go to sleep, and then it is like 3 in the morning, and something about it being 3 in the morning triggers good conversation.

It would be interesting to test what actually about "it being late" triggers the good conversation. For example, you could test to see if it is tiredness, or time of the day, or how many hours it's been since you woke up.

I think reduced inhibitions that come with tiredness might help here.

[-][anonymous]10y20

I suspect I've experienced a similar phenomenon on occasion. But to make sure, how would you define "good conversation" in this context?

Intellectual, deep, serious, personal, revealing, engaging. Something along those lines.

Same with me.

I think it could be coupled to our circadian rhythm. Maybe there is something that makes deep thoughts enjoyable in the night. Usually we don't get there with our unnatural unsegmented sleep, but when skipping the first segment maybe we accidentally get in the sweet spot.

Do people have advice or good examples of directing conversations? Recalling the best conversations you've ever had, what made them happen?

I think most good conversations happen because you react to the other person. You listen to what they have to say and relate to it instead of trying to push the conversation into a particular direction.

This sounds like crap a teacher would ask their class or a psychiatrist would ask their patient to elicit responses. Robotic, unfun, and vague. Have you actually tried asking these questions of people in conversation, and what were the results?

This sounds like crap a teacher would ask their class or a psychiatrist would ask their patient to elicit responses. Robotic, unfun, and vague.

I suspect the bad feelings come from the experience that the teacher usually does not care about things that I care about. So I am asked to give a response of a very specific type, or I will be punished (socially).

Contrast this with an environment where speaking about unusual things may be socially rewarded, and admitting your weakness may lead to a rational advice how to fix it.

Sometimes people learn that the best way to avoid being hurt is to avoid being open. That's a good advice for a harmful environment. It's not an optimal behavior in helpful environment.

I have tried around half of these questions. I have asked "what have you been thinking about lately?" a couple of times, and found out about a project a friend was working on, for example. Asking what people have been reading led to learning about what people have been reading, which was interesting. Some of the questions were posed to me, like "what memes have affected you?" and "what surprised you?", and I found them to affect the conversation positively (in the former case, we ended up discussing the effect of the diligence meme on my early life).

I agree that a lot of these require some leadup. For example, the question "what eccentric things have you done?" might come naturally after telling people about your skydiving trip last week. Vagueness is partly addressed by the "recently/lately" specification, but I agree that the questions could use some further narrowing down. Do you have any suggestions to that end?

If you need to have form questions for conversation I think having them with built-in blanks to tune them to the existing conversation/people is important. Most of your conversations are probably not going to be with random strangers, but with people you already know a lot about (because they're your friends) or a little (because you meet them in a LW meetup or at school or work).

Even with strangers, people will generally respond a lot better to more specific questions than more vague ones. Consider how complicated the real question behind "How's it going?" is and how lame the answers usually are.

My intention is indeed to improve conversations with people I know well or semi-well. Some good questions with built-in blanks in this context are "How is your project X going?" or "What did you think of book X?". Do you have examples of other such questions? I think the kinds of questions you would use for starting a new topic and for deepening an existing topic are likely to be different, and the latter are much more context-dependent.

It does not seem difficult to avoid being robotic/unfun with these questions, if you ask them with actual caring and curiosity, and if the motivation is not to fill silence but to learn about the other person.

I think that the real determinant of conversation quality is the people participating (not the techniques). They have to want to analyze rationally, and they have to know how to. Most people don't want to, and don't know how to. I think the techniques you talked about might help a little bit though.

WIth this hypothesis of mine in mind, I think a better question is, "how do you find the right people to have conversations with?".