Basic Conversational Coordination: Micro-coordination of Intention

post by elityre · 2020-07-27T22:41:53.236Z · score: 40 (13 votes) · LW · GW · 7 comments

Follow up to: Something simple to try in conversations

In this essay, I want to describe a simple, but important, abstraction for how to think about what’s happening in a conversation, and how to make use of that abstraction for helping a conversation go more smoothly. This idea applies at multiple time-scales: in this post, I’ll describe the simpler case of second-to-second interactions. In some future post, I’ll explain the broader (and more complicated) minute-to-minute case.

Let’s consider some two-person epistemic conversation, say two people attempting to converge on some topic they disagree about.

There are two roles (moves/ strategies) that one can play in a two-person epistemic conversation. 

Both participants will almost certainly do both over the course of the conversation, and each may switch very rapidly between explaining and listening, but for any discrete time-slice of the conversation, each person will be holding an intention either to convey, or to come to understand.

It’s useful to track conversations in terms of which participant is implementing which role, second to second. Thinking about conversations this way, and being conscious of what intention you are implementing, as a participant, is useful because it enables conversational coordination.

At any given time, no more than one person should be explaining, and at least one person should be trying to understand what is being said. Specifically, this means that it should never be the case that both participants are in the explaining mode at the same time.

At any given moment, one person should be explaining, and the other person should be listening and trying to understand.

This may seem like an obvious point, but conversations frequently fail to meet this standard. A common failure mode: both parties are excited about the point they have to make, and have a visceral compulsion to make that point clear to the other person. And so both people are explaining at each other. When both people are in the explaining role at the same time, communication usually fails.

(The way Double Crux has been traditionally described aggravates this issue. The concept of a Double Crux can leave one with the impression that one is supposed to try and do both roles at once, both making their point and attempting to understand the point of their partner in one swoop. I, at least, don’t hold that Double Crux conversations should be conducted this way. [Instead, I think they should usually look something like this [LW · GW].])

From the inside, this situation often feels like seeing that your partner is just missing this one point, and if you just convey it to them, they’ll realize that you’re right. It feels so easy to tell them that reason why they’re wrong. Or seems that what they’re saying is so absurd that there’s almost a compulsion to point out exactly how absurd it is.

When I notice this happening in a conversation I’m participating in, I step back to the listening role and try and understand the point that my interlocutor is making. Only after I’ve paraphrased and they feel like I’ve understood them, do I return to making my point. (If it’s still relevant. Often, once I’ve understood what they were trying to tell me, my objection is obviated.)

TAP: notice that I’m pushing to make a point over my interlocutor -> mentally step back, and paraphrase what they're saying

This is super simple, but it is also an important foundation for making conversations go well.


Comments sorted by top scores.

comment by Yoav Ravid · 2020-07-28T10:00:42.843Z · score: 10 (3 votes) · LW(p) · GW(p)

This reminds me of an improv game, where in each moment one player needs to be standing, one sitting, and one laying down, whenever someone switches a position the others need to rearrange to restore the order.

The analogy here is that whenever a partner switches position you should switch position so that both positions are always occupied.

comment by romeostevensit · 2020-07-28T00:07:44.925Z · score: 7 (4 votes) · LW(p) · GW(p)

I've noticed that ADHD exacerbates the failure mode because one or both people are concerned about losing the trace on the idea they have.

comment by elityre · 2020-07-28T01:08:09.564Z · score: 2 (1 votes) · LW(p) · GW(p)

Not a perfect solution, but a skilled facilitator can pick up some of the slack here:

But yeah, learning to put your point aside for a moment, without loosing the thread of it, is an important subskill.

comment by romeostevensit · 2020-07-28T16:34:28.903Z · score: 2 (1 votes) · LW(p) · GW(p)

Yes, and I think having a facilitator manage branch control is basically the only good solution for now until we get an AI assistant that can add structure to speech transcription.

comment by Pongo · 2020-07-28T00:47:01.329Z · score: 1 (1 votes) · LW(p) · GW(p)

Yeah this can make it very hard not to be in “formulating“ mode as opposed to “listening” mode when not speaking

comment by romeostevensit · 2020-07-28T16:35:13.687Z · score: 2 (1 votes) · LW(p) · GW(p)

having a note taking program open helps a lot in my experience.

comment by Dagon · 2020-07-28T19:14:25.073Z · score: 2 (1 votes) · LW(p) · GW(p)

For moment-to-moment roles, it's probably worth a few additional states - these take time (often more than listening or explaining) and serve important purposes.

  • Processing. A participant is integrating information into their models, or considering how to explain something better that failed in the last iteration. During this time, they can neither listen nor explain well.
  • Switching. Either for error recovery (both tried to explain) or as a natural flow between explain/listen/process, this is non-instantaneous.
  • Socially reinforcing. This is a lot like explaining or listening, but the intended information is very different, and the transition to other states will be different.