Basics of Handling Disagreements with People

Epistemic Status: This is a collection of useful heuristics I’ve gathered from a wide range of books and workshops, all rather evidence-based (robustness varies). These techniques are designed to supplement basics of rationalist discourse, helping facilitate interactions—mostly with those unfamiliar with rationalist thought, especially on entry-level arguments. They may also be useful in conversations between rationalists on occasion. This is also a minimal viable product for an upcoming sequence that will dive into the analysis of well-managed disagreements. Details are intentionally left out.

Tl;dr: Rephrase, ask questions, do not presume your conversation partner shares your epistemology (i.e way of coming to conclusions, in general), ask them for real-world counter-examples, share personal experiences both ways as a means to get clearer, check what kinds of blindspots their own motivation presumes, dovetail interests with brainstorming, and also, all of what I’ve just said is merely pointing to a specific state of mind. You can get to this state of mind only with some form of introspection.

Arguing is sometimes wonderful. Yet sometimes it derails, or flat-out fails. Circumstances in which arguing fails tend to involve people who are not actively displaying rationality. LessWrong has done a lot to teach how to mutually progress on such disagreement. Yet this is only a very small community -the Rest of the World, aka People, still hasn’t read the Sequences.

Unproductive disagreement with people can lead to poor impression, pig-headedness, stress, anger, and sometimes worse. There has already been numerous discussions, on this forum, of ways to avoid getting there: there is already a book review on How Minds Change, then I spent too much time breaking down good disagreements to teach how to do it. But there wasn’t a short document that was summarizing the key takeaways. This serves as that document.

Ethical Caveat:
This post presumes that you follow ethical advice such as:
1-Being earnestly truth-seeking
2-Getting the consent of your partner beforehand to question their beliefs
3-Not harassing people who don’t want to talk
4-Choosing the right context (most of the times, 1:1 conversations)
5-Choosing the right person (not a hierarchical subordinate, expect conversations to be harder with a family member)
6-Choosing the right topic (probably not things that are subject to trigger warnings, such as the gender identity of the person, nor things displayed during the conversation itself, such as thoughts you interpret from their non-verbal cues)
7-Not Being a D*ck, in general.

Attention to the reader: Reading about tennis does not teach enough to actually fluently play tennis. Practice is key. In the same way, reading about Effectively Handling Disagreements will be less effective than training yourself at it. Workshops in the comments (feel free to suggest some).

0-Actually, maybe, don’t argue.

Arguing is a choice. It can be fitting or unfitting. It can be a good choice, or a bad choice. Argument is a virtue of rationalists, but it is a virtue because it coheres with all the other ones. When you discuss with a stranger, surrounding virtues such as evenness or curiosity might slip away. A good way to bring it back is to refrain from counter-arguing, and start with listening. The argument will still be there -but in a form that will make it softer and more pedagogical.

1-Rephrase, Rephrase, Rephrase

By Default, You Don’t Understand Your Partner. Understanding your partner does not take a large amount of pondering and ostentatious thinking as a first step: It requires at least repeating back, with your own words, then genuinely asking your interlocutor if this is what they mean, and if not, offering them to make a correction (“If I understood you right, and if this is not what you meant, feel free to correct me, you meant that.… is this right?). This is fairly basic, but it is worth practicing if you’re not accustomed to it yet. Remember that you probably don’t understand your conversation partner if you didn’t rephrase what they said.

Of interest: Smart Politics.

2-Ask More Questions

By Default, You Don’t Understand Your Partner. Worse than this -You Don’t Know You Don’t Understand Your Partner. Your partner, you might think, came to their conclusion because of claim X, or person Y, or argument Z. You might follow-up on those reasons without having pre-emptively checked they were even relevant to the discussion at hand. Cached as they are, your conversation partner will say counter-arguments. But they will not bother about whether said counter-arguments figure in their crux at all. This is a massive waste of time and rapport.

If anything, ask for a working definition^[1] of the things you’re talking about. Ask for their reasons to believe, rather than presuming what those reasons are.

3-The Typical Method Fallacy

Your partner is not necessarily an empiricist. If they tell you that God exists, and that they do so because of personal experience, this does not mean they think (like I would personally do) that their experience is statistically significant. Their method is relying on a claim, and the claim is Personal experience is reliable (as understood in, “more reliable than science on topics where science challenges it”). You might think that this departs so much from sanity that the only dignified move is to impatiently frown your eyebrows and go talk to someone worth your time. But you might as well question the claim.

Of interest: Street Epistemology.

4-How to generate a good question, fast

Note: This one is a personal observation. Although it took reading scientific litterature to notice it, there aren’t publications on it, to my knowledge. Addendum: I’ve replaced “personal experience is reliable” as an example with “Karm exists”. See comments on LW.

Let’s take the argument “Karma exists. For example, if I throw garbage out of the window of my car, then I’ll break a nail within 24 hours”.

Questioning productively such a claim might sound like it leaves a lot of options open, but there is a rough-and-quick way to do it.
Step 1: Identify the property that makes the inference valid in the eyes of your partner (here, the fact that the nail was broken after throwing garbage out of the window)

Step 2: Ask for an example of the same (super)class that has the same property, but does not lead to the conclusion. (here, “Could be there times where you break your nail, yet you haven’t done anything bad prior to that?’’)

Step 3: If you get an answer (e.g, “Yes, that’s an accident’), ask your partner how they distinguish said answer (“accident”) from their initial answer (“Karma”).

This is a rough outline of the process, which I’ll elaborate on in a future post. In short, ontological relationships form a socratic artillery. A true socratic move is one that helps your interlocutor have more than one hypothesis and apply an approximation of Bayes’ rule.

5-Personal experiences help clear out confusions.

If there is one thing I’d like you to remember and that is the most evidence-backed, and the most impressively efficient at solving disagreement, it is that stories help people understand what you’re actually talking about. This is mostly valid of short and to-the-point stories, so keep anything you refer to clear and concise.
When offered to share a story, then hearing a story, people develop more trust, which helps with paying attention to the interpretation you have of it, the actual information you’re trying to convey. They get a lot more details and a fleshed-out example of what you are talking about. They actually get what you’re trying to convey in a way that theoretical arguments completely fail at.

This does not mean that you should use the emotional force of stories to sway your conversation partners around. It rather means that a story -and the emotions it generated in you- are crucial background information to understand what you mean. These two situations can be hard to discriminate, yet the telltale sign of being in an epistemically honest case is contrast: “You see, what I’ve just shared, this is what I mean when I say X’’.

Of Interest: Deep Canvassing.

6-Care about their underlying values.

Whatever your proposition is, it might well fit perfectly within your conversation partner’s values in some instrumental way, contrary to their own beliefs. Try to spot and bring up your partner’s motivations -say, an e/acc who cares about innovation- then, from there, point at whether the topic at hand fits with it -typically, AI Safety can be expected to contribute to innovation. Of course, do not say lies about how the topic at hand fits with it (“lies” here is understood broadly and refers to epistemic obfuscation in general).

Of interest: Motivational Interviewing.

7-Negotiate through Brainstorming

In the spirit of Ask More Questions, focus on your partner’s interests (or “needs” if you’re the CNV type), not positions (“Why do you want that ?” instead of “What do you want?”). You’ll get building blocks for brainstorming creative win-win solutions together. Note that in real-world situations, you’ll still need to spend a lot of time building positive rapport and getting your partner to think about the solution with you.

Of Interest: Getting To Yes

8-Heal yourself to get in the right Mindset

As she stirred and opened her eyes, I saw her differently. Her freckles were more obvious now, the colors of her face more vibrant. It was like I was seeing her in high definition for the first time.

-Chris Lakin, Learning to do *real* empathy.

Empathy isn’t just a series of scripted responses. I’ve been nudging you to imitate the ways in which the right mindset manifests -through questioning, rephrasing, narrating. Yet the mindset itself is the key. Getting in a mindset is a therapeutic act, it requires both practice, but also and mainly introspection, insight, and acknowledging denial. Getting in the right mindset does not only change your actions -it changes your perception.

Of interest:
-Chris Lakin, How Unconscious predictions update
-VIEW Mindset
-Compassion Focused Therapy
-Focusing

Finally, a word of caution:

What is shared here is not necessarily suited to all people and contexts. Of course, relying on conversation patterns also has a potential to fall within the Dark Arts -police yourself. My belief is that there is a subset of conversational attitudes that are in line with virtuous rationality. These attitudes have to be mastered in order to manage conversations with the rest of the world. Such conversations will happen regardless—so it’s best to be prepared.

Many thanks to Neil for suggesting me to write this post.

^
Contested, see comments for discussion. My position is that people tend to focus on platonic definitions (as opposed to “working” definitions) way too much, even if it can be good in some instances.