Comments
Papers can seem more prestigious and don't get annoying negative comments.
Reviewer #2 would like a word
Papers can seem more prestigious and don't get annoying negative comments.
Reviewer #2 would like a word
yeah, as an academic, I find writing blog posts to be much more pleasant and enjoyable than writing scientific papers. Part of the reason for that is that scientific papers are much more technical and rigorous, and you are expected to back up all your claims in order to pass peer review.
The flipside, of course, is that I trust blog posts a lot less than scientific papers, my own posts included. I still think more academics should have a go at blogging though, it can help bridge the gap between the technical work and the public.
Yea, sorry, this was short.
Both blog posts and papers clearly have a lot of different tradeoffs.
I think that many academics feel much more comfortable with the specific peer review processes that take place in their field, than they would many online discussion forums.
At the same time, some of the reverse is true as well.
Happy to see this post just come out, which is relevant here!
Great post and I agree! Curious about one point:
> 6. Academics often prefer writing papers to blog posts. Papers can seem more prestigious and don't get annoying negative comments. To the degree that prestige is directly valuable this is useful, but for most things I prefer blog posts / Facebook posts. I think there are a bunch of "mid-tier" LessWrong/ EA Forum writers who I value much dramatically than many (far more prestigious) academics.
What are examples of comparisons between far more prestigious academics and mid-tier LW/EAF writers? Curious about what the baselines here are because it's definitely a bit harder for me to make this comparison.
As a very simple example, I'd compare Gwern to Yann LeCun. LeCun has done some great work historically, but when it comes to overall judgement on most things, I far prefer Gwern.[1]
Or, in philosophy, I think it's understood among some of our crowd that lot of modern philosophy work really isn't that interesting or great. Being a professor in philosophy is very prestigious, but that doesn't mean they're doing great work. I think it's telling just how few modern philosophy (and really, humanities) talks or papers people here pay attention to. Some of that is ignorance, but I think a fair bit is justified.
[1] My point isn't that Gwern is "mid-tier", but he's well known here. So take less successful/prominent versions of both Gwern and Yann LeCun for a more relevant example.
But then I guess you end up with something like Kat Woods vs. Uri Hasson or something like that, and that's not a comparison I'd necessarily make. And separately, what Yann lacks in holistic reasoning, he does make up for with the technical work he's done (though he definitely peaked in '96).
The same for philosophy. What are some examples where theory of philosophy on the forums is significantly better than e.g. the best book on the topic of that year? I can totally buy this, but my philosophy studies during cognitive science were definitely extremely high quality and much better than most work on the forums.
Then of course add that EAs are also present in academia and maybe the picture gets more muddled.
Quickly:
"What are some examples where theory of philosophy on the forums is significantly better than e.g. the best book on the topic of that year?"
-> I think the word "better" here is oversimplifying. I'd expect that published work in formal philosophy will represent more labor and often more skills of certain kinds than blog posts. But I'd also expect this to come with certain assumptions and focus areas that I disagree with.
I'd point people to this previous post:
https://www.lesswrong.com/posts/FwiPfF8Woe5JrzqEu/philosophy-a-diseased-discipline
And maybe:
https://www.lesswrong.com/posts/LcEzxX2FNTKbB6KXS/train-philosophers-with-pearl-and-kahneman-not-plato-and
I'd notice that I'm finding it awkward to be specific here, because then I feel like I'd be calling out some writers as "mid-tier", and also calling out specific academics that I know.
I didn't mean to make a very controversial or strong point here.
Yeah, makes a lot of sense! I think of mid-tier not as offensive since it's also just about Gwern spending all his time on writing vs. Kat Woods running an organization as well - huge respect to both of course for what they do.
Great post, hadn't seen that one before.
I'll also mention that I don't think SoTA philosophy happens in any way within any of the areas that Luke mentions. If this is classified as academic philosophy, then that's definitely fair. But if you look at where philosophy is developed the most (outside of imaginary parallel worlds) in my eyes, it's the summaries of academic work on consciousness (The Conscious Mind), computer science (Gödel, Escher, Bach), AI (Superintelligence), genetic foundations for morals (Blueprint for Civilization), empirical studies of human behavior in moral scenarios (Thinking, Fast and Slow), politics (Expert Political Judgment), cognitive enhancement (Tools for Thought), and neuroscience (The Brain from Inside Out), all of which have academic centres of excellence that are very inspiring.
Like, the place philosophers who truly want to understand the philosophical underpinnings of reality go today looks very very different than it did during the renaissance, in the sense that we now have instruments and mathematics that can measure ethics, morals, and the fundamental properties of reality.
Yep, I agree with most of that.
My point wasn't at all that bloggers are better than the best academics. More I was highlighting that there are situations like those in what's known as contemporary "Philosophy", which features people who are well-respected within certain niches, but whom most of us wouldn't find that exciting.
Yep, probably agree with this. Then it's definitely good to lead a promising researcher away from the bad nichés and into the better ones!
I have an observation but may be noise; it is something like "you have to be known to the community somehow, to get more interactions from the community, unless it is negative interactions (debates of some sort)". I am slightly critical but not very critical about this.
This seems right to me - personally I am more likely to read a post if it is by someone I know (in person or by reputation). I think selfishly this is the right choice as those posts are more likely to be interesting/valuable to me. But it is also perhaps a bad norm as we want new writers to have an easy route in, even if no-one recognises their name. So I try to not index too heavily on whether I know the person.
I think it does help to have a username that people recognize, but IMO it's not too hard to get to that point. There's not a ton of activity on the EA Forum and it's not that hard to read most comments, so if you write a few good comments people will probably start to recognize your username.
Even if it just gets a few upvotes, it is still likely to show up when people search the forum or read all of the posts under a particular tag. Arguably, those are the readers you care most about, since they are most likely to be working in the area you're writing about.
There tends to be a winner-take-all effect for post upvotes. Perhaps it makes sense to assess quality according to log(upvotes)? A post with 100 upvotes is probably not 10x as good as a post with 10 upvotes. It just happened to hit the frontpage at a time when there was little competition or something like that.
This is a more general problem with social media. Many users are either getting more attention than they want, or less attention than they want. It should really be called "DIY broadcast media", not "social media".
I think it's easy to get disenchanted when writing posts or comments on blogs. I've particularly noticed this on the EA Forum and LessWrong.
Engagement can be highly random, and often rewards behaviors not correlated with quality. For example, posts like "10 more reasons to dislike Sam Altman" will often do a lot better than "My theoretical exploration of a new idea regarding AI policy."[1]
It's incredibly easy to have excellent work be completely ignored. I've experienced extreme frustration from this with some of my own posts, and have seen a lot of anger from others who feel unappreciated. I know many strong thinkers who quickly give up on writing anywhere.
But while it's tough to stand out in these crowds, I think it's also true that even a low-percentile post can have more value than many other things one can do with their time. I think it's easy to compare one's work to the most popular writers on the platform and be discouraged - but arguably this is a very competitive and high bar. A better bar would be to compare oneself to all the people who don't even bother to write - it's just that you don't see that when posting.
A part-time writer who gets 10 karma per month can easily be an intellectual heavyweight as a percentile of all people. I think that incredibly few people in the world (maybe 1%?) can produce semi-reasonable takes on these platforms. There is a high learning curve, and some of it asks that the author really be well-versed in a wide variety of topics.
Other various points:
1. If you're looking for interesting intellectual challenges, "Writing posts that even get 1-2 likes/upvotes" is often a higher bar than things like crosswords, debate, or many other things one can do.
2. Writing can be used as a way to keep oneself sharp and educated in a variety of topics. Again, it's not needed for others to read your work.
(Generated with ChatGPT)
3. There generally are few ways now that many people can remotely contribute to areas like existential risk. Median blog participation, while clearly worse than the top work, is still a lot better than most things one can do.
4. I generally find it much easier to talk to people who have some experience writing/commenting (or at least reading such work). I like talking about intellectual topics, and people who engage with these sources often have a high floor of knowledge.
5. In the future, I really hope to have better estimates of impact of all the things. I expect that small blog posts and comments will greatly under-perform some key areas (in terms of estimated impact), but over-perform the majority of actions people typically do to be useful.
6. Academics often prefer writing papers to blog posts. Papers can seem more prestigious and don't get annoying negative comments. To the degree that prestige is directly valuable this is useful, but for most things I prefer blog posts / Facebook posts. I think there are a bunch of "mid-tier" LessWrong/ EA Forum writers who I value dramatically more than many (far more prestigious) academics.
7. If you're curious to get into writing, I suggest it! Don't expect comments/upvotes, especially in the beginning. If you can stay with it, it can be fairly rewarding and valuable. Half of the challenge is developing a stubbornness so that you won't give up, when you compare yourself to others or see a lack of engagement.
8. A bit more evidence - a lot of top EA researchers basically can't blog in a way that they both feel good about and gets decent activity (maybe averaging 5+ karma, and/or 1 comment/post). Many of these people just don't bother. Doing this kind of writing can be really tough, even for much of the best intellectual talent.
9. I find it easy to get into unnecessary cycles of polish on my posts. The smaller your community/audience is, the more okay it might be to focus on quantity instead of polish. Plus, in terms of having AIs read the content (if you want that), polish probably matters less than one would expect. For example, you might notice that this post really isn’t organized. I’d expect if I tried to organize it, it would take me another hour or two, and I’m really not sure if it’s worth that.
(See a few other comments on the older Facebook post)
[1] I've previously discussed this as an issue of "Convenience". Some posts will directly help members of the community via things like "give this community higher status."
Thank you for praising my new hobby, Ozzie.
As an expansion on point 7: consistent, legible, self-motivated output on topics that matter is just a huge signal of value in intellectual work. A problem with basically all hiring is that the vast majority of people just want to "get through the day" in their work rather than push for excellence (or even just improvement). Naturally, in hiring, you're trying to select for people who will care intrinsically about the quality*quantity of their outputs. There are few stronger signals of that than someone consistently doing something that looks like the work in their personal time for ~no (direct) reward.
Also, if you're worried you're not good enough, you're probably right, but the only way to get good is to start writing bad stuff and make it better. I wrote the first post of my meh blog on this topic to keep me going. It's sort of helped.
Agree with the above—fwiw I and hiring managers I work with have solicited applications and hired candidates based on seeing their personal writing here and elsewhere. That's definitely not a sure-fire path to getting hired and you shouldn't Goodhart it, but if you have the urge to write and are holding yourself back for any reason, I'd endorse you pushing through it. As @Mjreard notes, this sort of writing is a hard-to-fake signal of intellectual investment and productivity, and hiring managers are hungry for such signal.