14

The most persuasive writing neutrally surveys both sides of an argument

A piece of advice for people posting here and elsewhere: what you write will be more convincing and higher quality if you set out to survey the considerations on both sides of a disagreement.

This is because readers will be able to weigh the arguments on either side against one another in a single place. It also means you yourself will have to consider a wider range of angles in reaching your conclusion, rather than making a one-sided search for arguments in favour of whatever you believe at the start.

A example of the problem with the alternative is Peter Hurford's post on 'EA Falling into a Meta-Trap' which is one of the most up-voted posts ever written here. I don't mean to pick on Peter in particular because most people naturally write 'the case for conclusion X', including me. Fortunately, as Peter is one of the most popular EA writers I don't feel like a jerk using him as an example.

Conveniently I disagree with Peter's conclusion and believe that EA has been, and is likely to continue, to under-invest in meta-charity. I don't intend to convince you that I'm right about that here - instead simply imagine the voice in my head as I'm reading that post:

  1. Here are some arguments against spending too much on meta-charity.
  2. Hmmmm, I've already heard most of these considerations before, but think they face very strong considerations on the other side.
  3. Oh, the blog post ended without considering the overall weight of the arguments on either side.
  4. And it didn't try to measure what fraction of our resources go to meta-charity, what fraction might go to meta-charity in the future, and what would be an appropriate fraction all things considered. It's completely consistent with everything in this post that the primary risk is actually spending too little.

Unfortunately, this means I didn't update my views that much in either direction, despite it being a very important issue to me. Which is a shame, because everything Peter wrote was sound in and of itself.

Here's an alternative structure for a post:

  1. Currently many people believe something like X (including me).
  2. Here are the best arguments that people offer in favour of that belief.
  3. Here are the best arguments / counter-arguments I can think of pointing in the other direction.
  4. Overall I think points A, B and C should be given most weight, which means my overall judgement is now Y.
If the goal is to help others form sound views, putting all of this in one place makes the whole more than a sum of its parts. Everyone will feel like they have been heard so they are more open to updating in one direction or another.

Furthermore, if you write this way you may well change your mind in the process. Phillip Tetlock's research has found that people who use something like this approach, instead of the 'write a strong case for conclusion X', produce more accurate forecasts about the future.

What are the best arguments against writing in this way?
  • It's more boring to read because you usually won't offer a strident view that people disagree with, and it takes longer to read.
  • It's at least twice as much work.
  • Commenters will offer the counter-considerations anyway.
These are true to some extent, but I think they are outweighed by the benefits.

Though it's more boring and creates less emotional investment, that's a pro as far as getting people to form accurate beliefs. Our goal here isn't primarily our own entertainment. People who care and make decisions based on a question will usually still read a comprehensive post about it. However, if your goal is just entertainment and attention, then you should aim to write things that are as biased and inaccurate as possible.

Posts like this take twice as long to write, but only because they contain twice as much information, in a package that makes that information more than twice as valuable. You are saving someone else the hassle of writing a response with everything you left out.

Commenters will often provide counterarguments if you don't. However, this is less satisfactory than putting them in the original post. Firstly, many readers will not dig into comments and will only read the original post, causing them to get a biased view. Secondly, comment threads tend to rapidly become messy and hard for observers, and even participants, to follow. They rarely extend to more than 2 responses from each side so we want each one to count. Thirdly, if you have already provided the main considerations in favour and against something in the original post, then the comments jump to how they should be weighed against one another, which is more interesting and typically the real source of disagreement.

In conclusion, anyone who makes the extra effort to write up an all-things-considered judgement gets an up-vote from me.

Update: People are mostly only taking issue with whether single-sided posts are more persuasive. I disagree, at least for myself. But if single-sided posts are more persuasive, that strengthens the argument for not using them here. Being convincing in the service of less reliable views is no virtue.

Comments (23)

Comment author: Pablo_Stafforini 18 February 2016 11:40:21AM 4 points [-]

What are the best arguments against writing in this way?

Kudos for acting on your own advice!

Comment author: Robert_Wiblin 18 February 2016 11:55:47AM 8 points [-]

I've been in the internet debating game for long enough to know to cover my own butt. :P

Comment author: Stefan_Schubert 18 February 2016 09:36:11AM *  7 points [-]

I very much agree that one should neutrally survey both sides of an argument. However, one could argue that the fact that Peter Hurford's post is one of the most upvoted articles indicates that writing "one-sided" articles are in fact more persuasive, even though it shouldn't. I therefore think your title is slightly misleading. But generally I think this is a very good and important post.

One-sidedness is generally a sign of confirmation bias, the halo effect, or some other bias. This has got quite a lot of attention in EA circles. See, e.g. Yudkowsky's Policy-debates should not appear one-sided, and Gregory Lewis's Beware Suspicious and Surprising Convergence. I myself have made a political bias test based on this idea and written an academic paper explicating its logic in Bayesian terms. I have also argued that one-sided multiple-factor explanations are suspicious (see also Scott Alexander' post on this issue ).

In the multiple factor post, I argue that illuminaries like Jared Diamond and Steven Pinker are guilty of arguing in this one-sided way, which is very pervasive indeed. Possible arguments in favour of writing in this "one-sided" way (a question that Rob discusses above) are discussed in the comments section (see the comments by DavidAgain and my responses).

Comment author: Robert_Wiblin 18 February 2016 09:40:04AM 2 points [-]

It's probably persuasive to people with no existing view or knowledge, but not to people who disagree or know a lot already.

Comment author: Peter_Hurford  (EA Profile) 18 February 2016 03:20:14PM 4 points [-]

I agree with Stefan that it's more persuasive to write one-sided, and I'd point to the fact that the most popular articles out there (both here on the EA Forum and definitely elsewhere) are presented one-sidedly. I think by "persuasive" you meant "best for helping readers form accurate beliefs", which are different things ;)

I write one-sidedly from the perspective of "offering additional considerations people haven't thought of to the considerations everyone already knows" and I don't spend much time talking about the considerations everyone already knows. This is mainly to save time as you said, because blogging here is definitely a very side project to me and for nearly all my pieces, I don't have much longer than 3-4 hours to write them.

Comment author: AGB 18 February 2016 09:06:45PM 4 points [-]

As Rob pointed out, balanced articles are normally more boring to read. So we shouldn't be surprised that they are relatively less popular. Popularity is a (very) poor indicator of persuasiveness, because the popularity of an article giving a viewpoint is mostly a function of people who hold the same view holding it up and saying 'Look at this. It argues for my view!'. Confirmation is not persuasion.

This is especially true if we're going to measure popularity by 'number of shares' or 'number of upvotes'.

Comment author: tyrael 18 February 2016 04:17:37PM 3 points [-]

This seems like a good opportunity for collaboration! Perhaps one-sided posts could include a disclaimer at the end of, "In this post, I've covered the most compelling arguments for X because I think the considerations on the other side are things most people already know. However, I invite someone with more time/interest to compile those in a comment, or message me and I can add them to the post itself."

Of course, this assumes we want to be "persuasive" in the way Rob means rather than the common definition of, "most likely to get people to agree with you."

Comment author: Peter_Hurford  (EA Profile) 18 February 2016 04:45:48PM *  0 points [-]

I'd be interested in doing that.

I'll also likely invest a lot more time (and look at both sides) in posts I care a lot more about, e.g., about where to donate.

Comment author: Robert_Wiblin 19 February 2016 06:25:59AM *  0 points [-]

For the purposes of the title I was thinking more like 'persuasive to me', or 'should change people's minds'.

Dan Keys chased up some studies suggesting that a more comprehensive treatment might be more convincing after all.

But if I can get people to think that blog posts which only present arguments on one side are more persuasive, but should actually only be slightly persuasive.

If something is one-sided, then the more persuasive it is the worse!

Comment author: Dan_Keys 19 February 2016 02:09:45AM 4 points [-]

A quick search into the academic research on this topic roughly matches the claims in this post.

Meta-analyses by Allen (1991) (pdf, blog post summary) and O'Keefe (1999) (pdf, blog post summary) defined "refutational two-sided arguments" as arguments that include 1) arguments in favor of the preferred conclusion, 2) arguments against the preferred conclusion, and 3) arguments which attempt to refute the arguments against the preferred conclusion. Both meta-analyses found that refutational two-sided arguments were more persuasive than one-sided arguments (which include only the first of those 3 types of arguments), which in turn were more persuasive than nonrefutational two-sided arguments (which include the first 2 of those 3 types of arguments).

So: surveying both sides of the argument, and making the case for why one side holds more weight than the other, does seem to lead to more convincing writing.

These results are at a fairly broad level of generality. I don't know if any research has looked at questions like whether it matters if you include the strongest arguments against the preferred conclusion (vs. only including straw man arguments) or if it matters if you act as if the arguments against the preferred conclusion have been completely refuted (vs. somewhat outweighed by the arguments in favor of the preferred conclusion).

A quick skim through the list of articles citing Allen and O'Keefe's papers turned up some studies which look for additional sources of variability which might moderate this effect, but I didn't notice any that challenge the general pattern or which get into really good detail on whether normatively good arguments (e.g., non-straw-man, measured conclusions) are more convincing.

Comment author: Robert_Wiblin 21 February 2016 04:27:46PM 1 point [-]
Comment author: saulius  (EA Profile) 19 February 2016 12:45:13AM *  2 points [-]

If you are writing a summary of existing arguments, then yes. But if you have a new argument, then there is no reason to drown it in old arguments.

Comment author: Robert_Wiblin 19 February 2016 06:27:04AM -1 points [-]

That's true, but then you can't claim to say where the balance lies - you're just adding one new brick to the wall.

Comment author: Peter_Hurford  (EA Profile) 20 February 2016 03:06:15AM *  1 point [-]

Does anyone know why this article has been so heavily downvoted?

Comment author: Robert_Wiblin 20 February 2016 04:36:44AM 0 points [-]

I presume because people think unbalanced arguments are actually more persuasive (even if they shouldn't be).

Comment author: Peter_Hurford  (EA Profile) 20 February 2016 04:38:45AM 0 points [-]

Even so, downvoting shouldn't be used to indicate disagreement.

Comment author: MichaelDickens  (EA Profile) 20 February 2016 06:36:05PM 0 points [-]

You are correct, but some people do it anyway.

Comment author: Robert_Wiblin 20 February 2016 04:55:09AM 0 points [-]
Comment author: Peter_Hurford  (EA Profile) 19 February 2016 12:00:47AM *  1 point [-]

I wonder if my articles would be even more popular if I added images... (joke)

Comment author: Robert_Wiblin 19 February 2016 06:32:08AM 1 point [-]

They will be shared vastly more on social media if they have an image!

I think it also helps people recall the information because it will be associated with an image.

Comment author: Gleb_T  (EA Profile) 19 February 2016 04:39:37PM 0 points [-]

While I appreciate the sentiment expressed here, I agree with most commenters that the most persuasive writing style, and the most popular writing style, is to present one side of the argument.

If we intentionally want to change this dynamics, we have to change the incentive structure. For example, we can get a group of people to make a commitment to upvote EA Forum articles that present all sides of an argument. Other options are available as well.

However, until the incentives are changed, people will do what they naturally get rewarded for, and what takes less time and efforts.

Comment author: Robert_Wiblin 19 February 2016 06:11:39PM 2 points [-]

"If we intentionally want to change this dynamics, we have to change the incentive structure."

That's what I'm doing here by changing what is perceived as virtuous.

"the most persuasive writing style ... is to present one side of the argument"

If it's persuasive that's a further argument against doing it, because you're biasing what people believe. At least if it was unpersuasive you wouldn't be doing any harm.

Comment author: Gleb_T  (EA Profile) 20 February 2016 05:18:21PM 1 point [-]

If it's persuasive that's a further argument against doing it, because you're biasing what people believe. At least if it was unpersuasive you wouldn't be doing any harm.

Regarding the persuasiveness, it depends on what you're optimizing for. Many people - including on the EA Forum - optimize for less effort and winning, as opposed to truth. This is a problem, as you rightly point out. I support your effort, and am suggesting a further specific incentive of people making a commitment to upvote posts that present both sides.