In Defense of Debate
TristanTrim wrote on LessWrong (link added):
I dislike "argument" and "debate" as contexts for discussing discourse. As implied in Centola's book, more minds are changed by discourse than by debate. I know it's obtuse to use a phrase like "collaborative truth seeking" but it would be nice if there were more people standing up for the "collaborative truth seeking" vibes.
And wrote:
the things that "debate" normally points to in natural language doesn't seem worth trying to salvage.
And (quoting me):
Finding a point of contradiction, and knowing that at least one of you is wrong, and talking critically about it using arguments and evidence
I'd say you have to do a lot of work to have enough shared context to know that at least one of you is wrong. I like the "at least one" sentiment, since both people can be wrong, but both people can also be right and just using language differently and failing to really connect with one another. The parts doing all the work are sharing world models, synchronizing terminology, locating differences in world modelling, and looking at evidence. The debate framing doesn't seem to add much to that.
I have nothing against collaborative truth seeking. I like both the phrase and the activity. But I also like debate. They're different and I'm going to talk about the pros and cons of debate and why I think debate has value.
I acknowledge that a lot of debate is low quality. Some good comes out of it anyway, but also some bad. I'm going to defend debate as a concept, not the common practice. I too think most debates on social media and on stages should change. My preferred form of debate takes place over multiple days in writing (e.g. using an internet forum like mine, LessWrong, letters or email), so people have time to think over and edit their responses, engage with exact quotes, and read citations.
Debates can be adversarial but can also be fairly collaborative. Would it be better if all debates were friendly and collaborative? If so, what makes debate different than collaborative truth seeking besides the existence of bad, adversarial debates?
Debate has a specific, narrower goal. It's a more focused activity. It's less open-ended and typically shorter. It requires less sharing world models, synchronizing terminology, etc. Those are valuable activities, but not everyone should do them with everyone, and they aren't always necessary. They take a lot of time and effort, and they're difficult to do very thoroughly across large cultural gaps, which limits how many people you do them with, so having other positive ways to engage with people is important.
The goal of debate is to address a specific point of disagreement. When two people believe contradictory things, at least one of them is wrong. Unless they're overconfident, they can mutually agree that it'd be good to know which belief is wrong.
If you've put significant effort into your belief and would be surprised to be mistaken, and you're somewhat rational, then finding out you're mistaken would be highly valuable. Correcting others who are mistaken about your area of expertise (especially people who have also put in effort) also has value: teaching and sharing knowledge is good (as long as people listen some instead of ignoring you) even if it isn't as wonderful as learning something new yourself.
Debates (like collaborative discussions) can involve more than two people, but for simplicity I'll assume two debaters.
Because debate is limited in scope to resolving a specific contradiction, less collaboration, discussion, sharing ideas, synchronizing terminology, etc., is necessary. This can save time and enable engaging with more people and especially with more people who are very different than yourself. It's fine if debaters engage in some collaborative truth seeking if they want to. It's fine to mix some of that in. But it's not necessary.
Debate can be productive even when some people have adversarial attitudes. The requirements for debate to be productive are more modest than for collaboration to work. Debate is more accessible. It requires less integrity, rationality, goodwill, tolerance, appreciation of diversity, non-tribalism, etc. And even if someone is good at all of those things, there's nothing inherently wrong with debate. Sometimes it makes sense to focus a discussion on a specific disagreement even if truth seeking collaboration is an available option. People have busy schedules and figuring out which contradicting belief is wrong is an especially high value thing to prioritize.
If both people are highly rational, you can collaborate or debate. Both work well. If neither person is very rational, then neither debate nor collaboration will be very productive, but I'd consider collaboration the better option if it's available. But if one person is highly rational and knows how to approach debate well, then debate can work even if the other person is somewhat adversarial and irrational. A skilled public intellectual ought to be able to approach debate productively with audience members and critics who are less rational and charitable than (s)he is. They can partially guide the debate using reasonable, neutral, common sense points like asking what the other person's position is, asking about relevant premises, asking to stay on topic, setting a good example with their own claims, and working to find root causes of disagreements in fair ways. Some people will be fully uncooperative and unreasonable, in which case debate won't be productive and you can end there, but this can work with a lot of people who would not collaborate, such as people who strongly disagree with you.
When I was younger, I used to want to talk through any and all disagreements, and share ideas, and learn from each other. I still like that, but I also better appreciate that people are busy, there are many other people to talk to, and focusing on the most productive areas is often best. We have limited time and can't get to know everyone really well. I prioritize more now, and more often look for ways to skip discussing some mistakes or disagreements instead of trying to resolve everything. The most helpful author for me, for this topic, was Eli Goldratt (he has a lot of good books, but start with The Goal: A Process of Ongoing Improvement).
To debate well, it's important to be familiar with and use standard terminology (and well known literature references). This can't be done perfectly and some clarifying questions will still be needed, but it can help speed things up. In debate, we usually try to say things that are understandable to an audience, not just to the person we're talking with. I don't like "debates" where people focus on talking to the audience. I think debaters should engage with each other but keep the audience in mind secondarily. (It's OK to debate with no audience, but I do like public debate that allows an audience to learn from the debate and even allows the audience to comment and share ideas.)
In debate, we should clarify where we disagree and try to identify a specific contradiction, then share our reasoning (explanations, arguments and evidence). Then we'll often need to trace things back to premises and the reasoning for those, possibly for several layers. We can do this for each person's position, though sometimes just focusing on one person's position is mutually agreeable. A tree diagram for each person's position, or one merged tree, can help keep things organized.
This should quickly lead to some understanding of the cause of the disagreement: e.g. someone was unfamiliar with a piece of evidence, disputes a piece of evidence, or believes a particular argument has a logical error. Many disagreements may surface in which case people should focus on something they mutually want to or on root causes. When there are many disagreements, most of them don't have to be resolved in order for the debate to reach a conclusion. They aren't all independent issues. And many disagreements are quantitative and small and can fit in margins of error instead of resulting in different conclusions.
I won't be able to explain in this essay in detail how to resolve a disagreement once you narrow it down to one or a few key issues. This difficulty also comes up in collaboration and it's a complex issue. Available tools include math, logic, grammar, explanations, looking up facts, increasing precision, breaking issues down into smaller chunks, tree diagrams, pro and con lists, evaporating clouds, and double crux methods.
A common reason people favor collaboration over debate, which isn't usually stated, is because they believe in indecisive arguments. E.g. Bayesians believe in incrementally updating credences based on arguments and evidence. In this model of learning, it makes sense to work together to gather evidence and update. However, if you believe a decisive epistemology which focuses on pass/fail evaluations of ideas and qualitative issues, instead of quantitative evaluations, then it becomes much more realistic to reach decisive conclusions in debates, and to do it fairly quickly. I know that indecisive epistemology is far more popular, but (building on Karl Popper) I've developed a decisive epistemology. I believe it avoids some difficulties like the problem of induction and the various known flaws of multi-criterion decision making and other weighted factor approaches. Decisive epistemologies better support productive debate (this benefit does not tell us which epistemology is true), so one's opinion about epistemology is relevant to one's attitude to debate.
What are downsides of debate? The toxicity of many actual debates isn't a random accident. Collaboration does more to avoid toxicity than debate. As I see it, unstructured, unfocused collaboration is mildly good and not very dangerous, while many unstructured, unfocused debate have negative value. Debate without goodwill needs to be organized a lot more than collaboration does. Adversarial debates need a skilled moderator or a debate methodology plus the discipline to follow it (and the methodology needs to include strategies for dealing with situations when you or your debate partner don't follow the methodology). I'm a fan of unmoderated debates, but it requires significant skill to keep them organized and well-managed. The more people are guided by a flowchart for what to do during a debate (or by a moderator), the less their hostilities and biases influence what happens. (Debate structures do not prevent all bad behavior, e.g. they don't prevent cherrypicking evidence. But they can help provide the opportunity to point out cherrypicking.)
Another major potential downside of debate is wasting time unproductively. This is also a risk with collaboration, though the ways time is wasted tend to be different. I think the basic solution is to have a good debate methodology which is designed to address this. In short, staying organized and on-topic, and focusing on figuring out the cause of the disagreement, helps avoid wasting time.
So why debate? It focuses on key issues more than generic collaboration. A collaborative attitude isn't required for a productive outcome (which matters in a world where collaborative attitudes are too rare and fragile). Debate can still function productively in cases where cooperation and goodwill break down or were never present. And I'll explain another reason. Consider a public intellectual who is wrong about some important issues. If he's open to collaboration but not debate, he can easily stay wrong for his whole career while engaging in lots of truth seeking collaborations. If he's open to debate, a critic can come along and correct him. Debate is better suited for actually getting errors corrected and facing challenging issues instead of letting people dodge key issues.
In general, I place high value on having a good answer to the question, "If I'm wrong, and someone else knows it, how will I be corrected?" I do my best to correct my own errors, but I also want to be correctable by others. I don't want to stay wrong about something where the right answer has already been figured out and people are willing to share it with me. Being open to debate is my main answer for how to do this, and I don't think being open to collaborations satisfies this. I have publicly guaranteed debates to anyone on the internet who meets minimal conditions, who claims to know I'm mistaken about something important, even if I don't like them and I don't think they have valuable ideas and I wouldn't want to collaborate with them. I think this is important so people aren't ignored due to poor culture fit, low social status, poor ability to get attention from crowds, lack of credentials, and other heuristics that are different than actually being wrong. I think that believing someone's idea "sounds dumb and not worth my time" is too often related to the social hierarchy and is a high bias risk, and I don't think people (including me) should trust themselves to make that judgment. I think people can protect their time in less risky ways. I think actually debating and pointing out why an idea is wrong (in reusable writing – or just linking to the explanation if it's already been written is fine as long as it genuinely covers what the critic is saying) is important to building up human knowledge. In general, if someone's claim is dumb it should either be already addressed in writing somewhere or quick and easy to address (if the explanation is complicated and no one has written it down, then they aren't dumb for not knowing it).
Because debate works in low trust, low collaboration, low goodwill contexts, it enables more engagement with people one sees as low status or outgroup (but this is only effective when the debate is well-structured; hostility without structure leads to toxicity). I think good intellectual communities should have some collaboration and some debate. If they focus too much on just collaboration, that ends up excluding some people with less shared context. Debate can address reasons someone thinks the community is mistaken before asking them to invest a lot of energy into learning its ideas.
PS I'm open to collaborative truth seeking, debate, or moderating a debate. I have written a lot about debate methodology but am also open to trying other methodologies if you believe you have a better idea. Contact me if interested. Finding a point of disagreement won't be difficult, so you don't need to already have a topic in mind. My main condition is I prefer medium or long interactions. I've already had a ton of discussions and debates, and I find the early parts can be repetitive before branching to be more unique, so I try to avoid doing the early parts then stopping. I also prefer to talk in public. And I like to have the shared goal of trying to reach a conclusion, which is rarely achieved in short interactions. I'd typically consider 5 messages per person in 1 day short (most interactions in online comments don't make it to day 2) while 15 over 3 days would be medium (of course it varies by message length and other factors).