If our blogging is to be more than shouting into the void, we have to write good comments. But that's hard to do. [LW · GW] Too much perfectionism, enforced positivity, criticism, status smackdowns, or vagueness in our commenting standards can all be problematic.
I'm using a framework that I think has been helpful. It's simple enough.
P = Prickly
O = Opaque
N = Nitpicky
D = Disengaged
S = Shallow
Let me define each term.
Prickly means that your comment has a chilly, disapproving, mean, or unappreciative tone. It could hurt feelings, and make somebody feel dumb for opening their virtual mouth.
Opaque means that your comment makes assertions without backing them up. You're saying stuff without giving any reason at all for it. This can also include some types of lazy "questions" that are really meant as cheap shots. Even giving a partial reason or motivation for your comment or question means that it is not opaque.
Nitpicky means that your comment is expressing criticism of one particular part of an argument, without stating whether or how this local disagreement informs your view of the original argument as a whole. Even saying "this is just a local nitpick; I don't know if it means much for the argument as a whole" is enough to make your comment not a nitpick.
Disengaged means that your comment doesn't give the impression that you'd be likely to carry the conversation further if you received a response. It's a "drive-by commenting."
Shallow means that you didn't read the entire original post or comment thread to which you're responding.
Each category is meant to be a very low bar. Express even mild warmth, underlying reasoning, attempt at synthesis, display of engagement, or depth of consideration -- not even all five, just one! -- and it's not PONDS. This term is only for the very worst comments.
Comments that are even slightly better than PONDS are totally acceptable. This is a way of speaking and responding that gives almost total freedom for Socratic grilling, while creating some minimal self-enforced safeguards to promote good conversation.
I've build a habit [LW · GW] of checking my own comments to ensure they're not PONDS. It's not that hard to improve a comment to do better, or else to delete it. I also have a policy to never respond to PONDS. Instead, I heavily downvote them. In fact, they're the only type of comment that I heavily downvote.
See Willa's comment below for an even more in-depth treatment of the five anti-PONDS virtues. Thanks to Raemon, Richard Kenneway, and Christian KI for helpful discussion and suggestions.
I don’t agree that a one-off or “drive-by” comment is a bad thing. Yes, it might be nice sometimes to have a lengthy discussion, but, you know, this is just a website, commenting is not my job, and people have other shit to do. I’d rather get a single good comment than get nothing at all from a person who has time to write one comment but not time to be on LW every day checking to see if their comment got a response.
Agreed! That's why, as long as your comment has at least one non-PONDS characteristic, it would pass muster. It can be warm, give a reason or motivation for the criticism, offer some synthetic reflection on the original arguments as a whole, or show that you've read and considered the whole post/comment thread that you're responding to.
I enjoy meta discussions like these, thanks for opening it up with your post.
Note: I read the above linked Stratechery articles last night and found them quite impactful + thought provoking, I've been wanting to share them. Doing so in the context of this comment might have been a stretch, but "changing defaults" seems relevant both to this post and my comment since the post is asking that readers comment with a certain set of defaults underlying their comment-style / is providing suggested commenting rules.
I think that, ideally, a comment should avoid those 5 failure modes and be written with an eye to "what quality-bar makes sense given my current skill level, time availability, knowledge, and the context". I'll demonstrate what that might look like below.
Suggestions on countering each failure mode and doing the other things I mentioned:
Prickly: I agree with your description of this failure mode. If someone wishes to improve their commenting "warmth" and/or avoid prickliness, they might try the following:
saying Thank you for writing this
saying "I like what you did here" or "This looks like a good effort" prior to "but, I have these specific criticisms."
if a post makes you mad and/or you think it is bad, take a moment to think about whether you really want to jump in and respond to it. If you decide to respond to it, then be mindful of the poster's experience level, their background, the inferential distance between you and what the poster wrote, and/or the intent of the poster (if you can't explicitly tell their intent from their post, use your comment to ask them what their intent was). Being kind but firm seems, anecdotally at least, to be a good approach in most cases.
Opaque: I agree with what you said about combating this failure mode, and I think it can be combated with varying levels of time and effort investment by someone. Specificity [LW · GW] is great and seems quite powerful, so providing specific examples or sources to support one's statements, presuppositions, assertions, and so on is usually a great way to combat opaqueness. Additionally, I think that being explicit about one's motivations, source and quality of one's evidence, etc. when making an assertion is valuable (I find it helpful to know my own motivations when doing something, and similarly find it helpful to know another person's motivations when they do something). An individual might be specific at different levels relative to their time availability, experience, skill level / expertise, the context, and so on, e.g.:
low time availability: "I assert x because it maps well to mine and other's anecdotal experiences, but don't have non-intuitive or non-experiential evidence to support this assertion." (then quickly share at least one personal experience example) Note that one can avoid opaqueness even if the assertion they made is supported by very low epistemic confidence evidence, it's via providing the evidence itself, source of that evidence, and confidence in that evidence whereby one moves from opaque to specific and explicit.
Nitpicky: I agree with what you said about this failure mode, and don't have much to add other than: if an individual has trouble combating their own nitpicky-ness I recommend writing out the main points, presuppositions, important assertions, sources of evidence, etc. from a post so that you can view the post's constituent elements. I do this when unpacking complicated posts and it makes evaluating a post much more doable for me, leading to much deeper analysis, counterpoints, rebuttals, suggestions, etc.
Trying to comment about the entire gist of a post at once seems quite taxing on the working memory, and I usually can't do that so I have to deconstruct posts into their constituent elements to actively engage with them and think critically about them let alone write good comments in response.
Disengaged: Mostly agreed. I will add that it's possible to write a long, detailed comment and still come across as disengaged if one does not ask good questions, explicitly state they are open to further discussion, and so on. I agree with remizidae though and believe that short, drive-by-comments can be engaging, depending on what the commenter says and how they say it: well targeted words stated succinctly can be powerful. Additionally, offering someone encouragement, or saying "Good job", saying "I liked this and want to see more", or some other nice thing in response to a post can be helpful for a poster to receive, especially if they have low confidence, are just getting started, are exploring something new, and/or might brighten their day and make them feel better about posting here on LW.
Shallow: I agree, though similar to Disengaged, I think it's possible for long detailed comments to be shallow if the commenter doesn't actually address what is said in the post and instead rambles on about other things. I think shallowness looks like disengagement, just as disengagement may look like shallowness. Are these really two separate things? They can each be found through the presence of one of them and look similar is why I ask.
I definitely demonstrated having decent time availability with this comment, did I sufficiently address the other things I mentioned that comprise the "quality-bar" I asserted?
General question: What are everyone's defaults with regards to commenting here on LessWrong? @AllAmericanBreakfast, what other defaults do you have besides PONDS?
Thanks for your very in-depth response. I edited my post with a note to point people to it.
While I tried to set the bar about an inch high, as Ericf points out, I generally try to edit my posts to be fully anti-PONDS. I'll edit a top-level post, like I did here, to appreciatively note a particularly long and substantive comment.
Sort of meta, but... I'm thinking about the use of acronyms in order to get more density into your title.
If you only get about 5 words, but one of those words is PONDS, have you successfully gotten 9 words? Maybe, and it's worth checking. If you can spend one of your words on unfolding complexity, that's great.
But my experience is "not really" – I tend to totally forget acronyms, unless they are hammered into me a lot over time, which requires other infrastructure. And I could go out of my way to set up a spaced-repetition thingy for this particular post, to get it to stick into my head. But, that wouldn't be scalable. I can only space-repeat so many things.
This is not a strong claim that this acronym won't work or that this post isn't useful. (I think this post works sort of as a space-repeat for the collective LW commentariat about "things to keep in mind for commenting", even if people don't end up remembering this particular acronym). But, I dunno, something to keep in mind.
I made a second attempt: Avoid cold, lazy nitpicks.
Cold: The comment has both a prickly tone and fails to show a clear willingness to continue the conversation. (Combines "prickly" and "disengaged").
Lazy: The comment fails to show any underlying reasoning and fails to demonstrate that the commenter even read the whole original post. (Combines "opaque" and "shallow").
Nitpicks: The comment is addressing a small, local issue without addressing how the disagreement impacts the argument as a whole.
"Cold, lazy nitpicks" is easier to remember as a phrase than "prickly, opaque, nitpicky, disengaged, shallow," since it's shorter and uses more common words. But it's harder to unfold the concepts, due to the inconsistent, imprecise lumping.
And of course, "PONDS" is a single syllable that is also a vivid noun, and is kind of related to the words it contains ("opaque, disengaged, shallow," "prickly" like prickly bog plants, "nitpicky" like the parasites that grow in a bog).
In general, I think there can be two different goals:
Create a phrase that's easy to remember. "Cold, lazy nitpicks" wins here.
Create a mnemonic that's easy to use. PONDS is better for this purpose.
In this case, I was trying to create a mnemonic that's easy to use. If most readers don't care to try, that's fine. But my hope is that at least one reader will actually use PONDS to change their long-term behavior. That would be a great success for less than 500 words :)
I think it's useful to have opaque as an extra category. In a realtime conversation it often makes sense to ask one sentence questions even if the motivation for the question is opaque if the question progresses the conversation.
My takeaway from reading your post is that it would be benefitial to sometimes be less opaque when writing a comment that contains a single question and adding content to make the motivation of the question more clear.
Oh, one thing that totally didn't occur to me was that PONDS was meant to illustratively evoke the sub-words. I think if the post did more work to make that actively salient, that'd help a bit. But, not perfectly
I think an important piece of mnemonics is to evoke both the initial-trigger state, and the desired end state (and perhaps the obstacle to avoid/route-around as you get to the end state). So in this case, somehow I want to have the experience:
notice I'm about to write internet comments
notice what sort of comments I'm about to write, and in particular if they are PONDS.
change what I write
In order for PONDS to be memorable, I need an illustrative bridge from commenting (Without necessarily noticing that I'm writing PONDS) to to actively reflecting on my communication goals.
"See, yonder is Llyn-dhu, garlanded with mosses and mean dwellings.”
Colin and Susan looked where Fenodyree was pointing, and some two or three miles out on the plain they could see the glint of grey water through trees.
“Men thought to drain that land and live there, but the spirit of the place entered them, and their houses were built drab and desolate, and without cheer; and all around the bog still sprawls, from out the drear lake come soulless thoughts and drift into the hearts of the people, and they are one with their surroundings."
— Alan Garner, The Weirdstone of Brisingamen.
Although the real Lindow Common, with its Black Lake, is nowadays a place for nice people to go with their dogs for a nice walk.