Dan Luu
patreon
Concision vs. pre-emptive nitpicking defenses
🕑 Added 2017-11-24 15:54:21 +0000 UTC There’s a spectrum where, on one end, you have clean writing that sounds like it’s making generalizations that might not hold, and on the other end, you have writing so packed with disclaimers that every topic turns into a book. I think Paul Graham is on the clean end of the scale. My guess is that he spends roughly 20% of the space I do to express an idea of similar complexity. Eliezer Yudkowsky is at the other end of the scale. My guess is that he spends about 10x the space I do for a given amount of complexity. An idea that would be single, moderately long, blog post for me is a series of long blog posts for him. Maybe it doesn’t matter and this is all a matter of taste, or maybe there’s some optimum I should be searching for.As part of my writing process, I typically write up the core idea I’m trying to convey and then do another pass to address nitpicking. This usually expands the post by at least a factor of two. I’ll then do a third pass and trim down the refutations for the nitpicking that’s mostly obviously wrong. This shrinks the post a bit, but usually still leaves it at more than twice its original length.
When I do this, I’m not only focusing on nitpicking. I also re-write sections so they flow better and re-order or remove sections that don’t fit. On average, the non-nitpicking re-work shrinks the post because I’m much more likely to remove something that doesn’t fit than try to add transitions to make the material fit. This implies that the anti-nitpick content is more than the half of the content by length.
I’m not sure what to do about this since this is both too much defense against nitpicking as well as too little. It’s too much because all of these little defenses bog down the text with details that are irrelevant to the main point. It’s not nearly enough because there are an unbounded number of incorrect “corrections” people could make. It’s usually the case that, when I remove an anti-nitpicking clause from a draft, I will get comments on the exact thing I removed and it’s not uncommon that I’ll see five or ten copies of pretty much the same comment, though the copies are usually phrased differently. Of course, if I don’t point out that I’ve also tried not removing anything and adding even more nitpick-defense, someone will inevitably email me or tweet at me saying that I should obviously just leave in the stuff I remove because, what am I, some kind of idiot? The previous sentence is an example of the kind of nitpick-defense that will save me a couple emails or tweets and a handful of internet comments, but is otherwise just bloat. I actually had to stop myself from writing the refutation to the next-level of nitpick on the sentence I just wrote about nitpicking because there’s no end to this. Every sentence adds more nitpick surface area.
One funny thing about this kind of comment is that comments that are angry or condescending almost always fall into this category of “comment I thought about refuting and maybe even did in a draft but then didn’t because it seemed too obvious to waste space on refuting”. I also get a lot of useful comments and corrections, but they’re almost always polite. For a while, I wondered if it only looked this way because of my own bias so I ran comments by a third party and they came to the same conclusion I did. My theory on this has been that people don’t reason very well when they’re angry, so angry and mean spirited comments are relatively likely to be poorly reasoned. Peter Bhat Harkins has the theory that this is about theory of mind: people with enough ToM to avoid writing angry comments will also tend to have better reasoning and be more likely to respond with interesting comments.
Whatever the reasons for nitpicking, I often wonder if I should just ignore all of the possible well actuallies and adopt a cleaner writing style. Since there’s widely read writing that’s all over the spectrum, it seems like this is mostly a matter of taste? Maybe some people prefer more streamlined writing and others prefer to read a long list of pre-emptive corrections and there will be an audience regardless of where I fall on the spectrum.
Comments
I think one level of detail for the main potential objections is good because oftentimes seeing the authors response to those things is illuminating in itself. Sometimes when the writing is too streamlined it gives the impression of not being thought through, and I'm having to stop and think it all through myself which counter-intuitively has the property of making it less streamlined for the reader (although, of course, it's nice to be intellectually engaged). When it comes to bad-faith, or mean-spirited, comments, I would try to avoid even considering them. As for poorly reasoned comments, they are only useful to contemplate if there would be some improvement to the writing that could actually avoid or reduced readers experiencing those problems. And when it comes to that, you have no direct way to measure that because, as you've identified, it's hard to discern them from comments which are in bad faith. And I think if you expend effort trying to optimize an objective that you can't measure or directly perceive, you're almost guaranteed to be on a wild-goose chase.
Personally, I love Dan's long and detailed posts and wouldn't want to see them trimmed down any more than they are.
