Why I don’t write as much as I used to (Brian Tomasik, 2022)

[niplav]

Link post

Brian Tomasik is well-known in Effective Altruism for his prolific writing on reducing suffering, but in the last years he has been updating his website less and less. He recently wrote a post about the reasons for doing so (emphasis all mine):

Now that I already have a large body of online writings, the marginal value of additional writings is plausibly lower. If I were to write more, it would probably be about increasingly niche issues. But I think the main value of my writings is to get people interested via the big ideas, like wild-animal suffering, insect sentience, suffering-focused ethics, and the concept of s-risks. From that starting point, people can “graduate” to the more advanced research being done by organizations like the Center on Long-Term Risk and the Center for Reducing Suffering. […]

As the EA and suffering-reduction communities matured, they also had more money to spend on grants, as well as more people to connect with. I often felt that it was more useful to give some high-level thoughts on grantmaking and organization strategy than to write a new public-facing article for my website. […]

In the old days, many of my discussions about EA had been in public forums with other amateurs. This meant it was possible to consolidate insights from those discussions into public website articles. As EA became professionalized, more discussions became private, which made publishing insights from those discussions trickier. […]

These days, I tend to feel like you can only write “state of the art” and relatively error-free articles if (1) you’re an expert in a particular field or (2) you’re writing about something you know personally, such as an organization you work with or your personal life (which are special cases of being an expert in a particular field). People who for their jobs have to write about lots of different topics often get things wrong—sometimes small details but sometimes crucial points. […]

There’s a saying that “young people think they know everything”, and while that was never literally true for me by any means, I think a much weaker statement in that direction was true. When I was younger, I felt like I had important insights that needed to be expressed to a greater degree than I do now. These days, I’ve been exposed to enough intellectual viewpoints that any given one of them appears less special. Most new ideas feel like the same old kind of stuff I’ve been hearing for a long time. […]

I tend to become interested in more and more areas over time, as I get exposed to new things. This means the amount of time I have to spend on any given topic area is smaller, and it becomes infeasible to continue collecting information on each of those topics. I also realize that people can often just do a web search or consult Wikipedia to find a lot of relevant information, and I personally don’t have to try to collect it. […]

Eventually I concluded that there were just too many new (and old) articles on the topic, and I didn’t have time to keep adding links to them.

[Mo Putera]

I’m reminded of what Brian Tomasik previously wrote in I prefer life in the slow lane—Too much literature to keep up with, which resonated with me as well:

In the modern world, scientists have to specialize into smaller and smaller sub-fields because there are so many papers being published all the time that it’s impossible to keep up with any more than a tiny slice of them. I feel the same way about various causes in effective altruism. I used to follow work on AI safety to a moderate degree of depth until 2015, when the field seemed to grow faster and faster, and I felt unable to contribute anything useful without reading much more material than I had time for. Toward the end of the 2010s, the fields of wild-animal suffering, invertebrate welfare, and s-risks also began to explode (at a smaller scale than AI safety), and I likewise had trouble keeping up with the work being done in those areas.

I’ve heard some people say that it’s important to have a critical mass of researchers working in some area in order to sustain people’s motivation. That’s an interesting viewpoint, though I happen to work in roughly the opposite way. If I’m one of the few people exploring some topic, the work feels useful and manageable. Once too many other people join in, I find it harder to contribute anything unique, and most of my time is spent reading the work of others rather than saying anything of my own.

Of course, it’s a good “problem” to have when there’s too much exciting work being done in an area to keep up with, but it does exacerbate the feeling of being overwhelmed.

One way out might be to do what Sarah Constantin describes in her essay How To Be An Educated Layman. I was curious about this, since her background is in math but she’s written hundreds of extremely fact-dense essays on a wide variety of non-math topics, so she’s as good an example as anyone. Quoting some parts:

[An] “expert” or “specialist” is someone who has spent their whole life studying or working on a specific narrow topic. If it’s an academic topic, they’ll typically have a PhD in that topic. An “educated layman” is someone who is not an expert, but who, in a discussion with experts about their field of specialization, can contribute useful ideas. An “educated layman” will typically know less stuff than any expert does about their field of expertise, but can still contribute good ideas that no expert has thought of.

How is that possible? Usually, because the “educated layman” brings a different perspective, or a toolkit from another field, or has an unusual set of priorities that the expert community hasn’t been optimizing for.

There are a couple standard examples I’ve encountered where it’s useful to be an “educated layman” — making decisions about whether/​how to apply a new technology, making medical decisions as a patient, managing experts, and importing a highly general “toolkit” or technology to a field where it hasn’t penetrated yet.

Really, there’s no such thing as reaching “educated layman” status in something as broad as “biology” as a whole. What you can do is tool up to the point of being able to have discussions with experts in a super-specific question.

What you’re missing, as an “educated layman”, is apprenticeship with a “master.” You will not know how to actually do lab procedures, so you will not know how they tend to fail in ways that aren’t captured in the published papers. You will not necessarily be able to replicate the results in papers if the “secret sauce” is unwritten and passed from lab tech to lab tech. You will not know the field’s gossip about who is worth listening to and who isn’t.

Your strength, as an “educated layman”, is the ability to drill down and learn everything about the one question within the field that you care about, which will often be different from what most experts in the field care about. Your lack of embeddedness in the community can actually be helpful, because by aggregating everything in the one slice of data you care about, you can generate ideas that aren’t correlated with the field’s biases.

Gwern also strikes me as someone who does this well; I think his “longtermist” approach to creating content is well-suited for pursuing as many super-specific questions as you want. Holden Karnofsky’s learning by writing via minimal-trust investigations fits this mold too.

[Brian_Tomasik]

Thanks. :) Yeah, I think I sometimes can offer a different perspective than other people working on a problem (such as s-risks). A difficulty I have is that I worry that what I’m saying might be obvious or wrong, and therefore it’s a waste of people’s time to talk with me if I haven’t read more of the background literature. But I still sometimes take the risk of wasting people’s time just in case what I’m saying is actually useful.