I enjoyed this, as a professor; I must publish in journals regularly and teach students to do the same. Students must always do a literature review; they think the reason is a typical classroom exercise of "prove that you know X" but it really is, as Hanson says, towards the accumulation of knowledge. I get tired of constantly telling every student that the first commandment as soon as an idea has entered their head: find out what is already known about X.
I have so many metaphors for this, e.g., sudents have an idea that is like an island disconnected from the mainland; there must at least be a peninsula; do the work to build the peninsula if you must... Now with AI, I may even send students an AI overview of whatever their question is before we have our first meeting, to keep us from tediously laboring over that territory and having to rediscover the wheel. Students don't understand how important it will be to their progress to understand whether their question is in virgin territory, sparsely settled, or crammed with high-rises. Sigh.
The problem is norm enforcement. In academia you have peer review and an informed readership that each serve to maintain the norms.
It would be nice if the academic quality mechanism (peer review) were decoupled from the distribution channel (publishing in a specific journal). With electronic distribution there's no longer any reason for the two to be tightly coupled. If that decoupling happened then anybody could pay to get their content quality certified to a certain standard of accuracy/novelty – including public intellectuals who otherwise don't have an incentive to navigate the journal submission process.
This is a very good point, but an opposing argument is that the obligation to be an expert on a topic before publishing on it has the devastating impact of preventing contributions from non-experts.
Often, an entire field could be salvaged by the injection of a little expertise from some other field.
- Linguistics wasted decades under the reign of Noam Chomsky because of his personal ignorance and dislike of information and probability theories, which led to his flawed objections to behaviorism and probabilistic grammars, to his refusal to admit that grammaticality is fuzzy, and to his mythical "poverty of the stimulus", which can be easily disproved on one side of a napkin by anyone who's read Shannon's 1948 book on information theory.
- Philosophy today is in an even worse state. Any mathematician could resolve either Zeno's paradox or the "raven paradox" in one minute (and let's not even get into the ridiculous "Newton's laws are circular" and "Newton's laws count forces twice" controversies). Anyone familiar with deep neural networks and LLMs could solve all the remaining mysteries of epistemology, though probably not in a single paper. Anyone who understands the normal distribution could annihilate Buddhism and post-modernism.
- Any historian could tell evolutionary biologists that group selection has been a powerful force in human evolution.
The problem is that public intellectuals are dumber than most academics, and cater to a dumber audience (the public). They also tend to have personality disorders, so they won't obey norms other than making follower count go up. Academic thought can be done in public, but it will always underperform publicly compared to much worse (by academic measuring sticks) writings meant to be sensational and tell the public what it wants to hear at the current moment.
Having been an academic researcher in economics and business (albeit a very mediocre one) for 18 years, the "situate" sounds good until one realizes it takes the extreme form of not only distracting the readers but actually obsession about referencing every trivial statement ad absurdum. E.g., "Competition with China has increased over the past 30 years" (seven references.) "incentives drive behavior" (10 references including one from 1978). Etc. Yes, I know, my examples are exaggeration - but not by much...
It's a good point. It just can't be in main essays, tho. I almost think sights like Substack would need to build some other sort of format to do it. The average reader doesn't want to see it. It's for writers to communicate to writers.
Tho, honestly, I despair of anything like this happening.
The New York Times copies stories from other journalists every day and doesn't credit them. The norm of linking other writers is totally wrecked.
And now you want someone to do a literature review?
It seems hard man.
Might be something for AI tools to try to do? Like rather than the writer do it, AI does it for them. Of course this doesn't force the original writer to actually do the reading, which is important.
Don't really know what to say. This seems a bit hopeless.
After, and I’d add that this has become 10-100x easier to do with modern AI. You take your draft, feed it into Claude, and ask:
“If this were an academic paper, who might it be citing? Who writes about these ideas in this essay, and might they have to say about them? Which of my points are more novel and which ones are covering ground well-trodden by others in the past, and specifically who?”.
This simple approach is likely to yield 90% of the value of a full-fledged academic “literature review”. It’s so easy nowadays that there’s no reason not to bother.
I think that situating your essay would have quite a negative impact on reach and engagement. Of course if everyone did it that wouldn't matter, but I'm just skeptical about the degree of compliance you would need to achieve given social media dynamics etc.
Great insight! I wonder if the medium of a public digital forum has anything to do with this. While it is miraculous that the Internet enables learning on anything and everything, there is no longer the same level of guidance one would normally receive in an academic setting. And with the added emphasis on digital anonymity, how are we supposed to even cite some authors accurately if we don’t know who they are?
Fair! But how would we reference this in such a way as to establish a sufficient scholarly trail if an address/link/URL isn't enough? A citation system needs some amount of identifying information to be useful. Aliases aren't always reliable, especially on the Internet.
I enjoyed this, as a professor; I must publish in journals regularly and teach students to do the same. Students must always do a literature review; they think the reason is a typical classroom exercise of "prove that you know X" but it really is, as Hanson says, towards the accumulation of knowledge. I get tired of constantly telling every student that the first commandment as soon as an idea has entered their head: find out what is already known about X.
I have so many metaphors for this, e.g., sudents have an idea that is like an island disconnected from the mainland; there must at least be a peninsula; do the work to build the peninsula if you must... Now with AI, I may even send students an AI overview of whatever their question is before we have our first meeting, to keep us from tediously laboring over that territory and having to rediscover the wheel. Students don't understand how important it will be to their progress to understand whether their question is in virgin territory, sparsely settled, or crammed with high-rises. Sigh.
The problem is norm enforcement. In academia you have peer review and an informed readership that each serve to maintain the norms.
It would be nice if the academic quality mechanism (peer review) were decoupled from the distribution channel (publishing in a specific journal). With electronic distribution there's no longer any reason for the two to be tightly coupled. If that decoupling happened then anybody could pay to get their content quality certified to a certain standard of accuracy/novelty – including public intellectuals who otherwise don't have an incentive to navigate the journal submission process.
This is a very good point, but an opposing argument is that the obligation to be an expert on a topic before publishing on it has the devastating impact of preventing contributions from non-experts.
Often, an entire field could be salvaged by the injection of a little expertise from some other field.
- Linguistics wasted decades under the reign of Noam Chomsky because of his personal ignorance and dislike of information and probability theories, which led to his flawed objections to behaviorism and probabilistic grammars, to his refusal to admit that grammaticality is fuzzy, and to his mythical "poverty of the stimulus", which can be easily disproved on one side of a napkin by anyone who's read Shannon's 1948 book on information theory.
- Philosophy today is in an even worse state. Any mathematician could resolve either Zeno's paradox or the "raven paradox" in one minute (and let's not even get into the ridiculous "Newton's laws are circular" and "Newton's laws count forces twice" controversies). Anyone familiar with deep neural networks and LLMs could solve all the remaining mysteries of epistemology, though probably not in a single paper. Anyone who understands the normal distribution could annihilate Buddhism and post-modernism.
- Any historian could tell evolutionary biologists that group selection has been a powerful force in human evolution.
The problem is that public intellectuals are dumber than most academics, and cater to a dumber audience (the public). They also tend to have personality disorders, so they won't obey norms other than making follower count go up. Academic thought can be done in public, but it will always underperform publicly compared to much worse (by academic measuring sticks) writings meant to be sensational and tell the public what it wants to hear at the current moment.
you may be the dumbest dude in the history of dumb dudes
Having been an academic researcher in economics and business (albeit a very mediocre one) for 18 years, the "situate" sounds good until one realizes it takes the extreme form of not only distracting the readers but actually obsession about referencing every trivial statement ad absurdum. E.g., "Competition with China has increased over the past 30 years" (seven references.) "incentives drive behavior" (10 references including one from 1978). Etc. Yes, I know, my examples are exaggeration - but not by much...
It's a good point. It just can't be in main essays, tho. I almost think sights like Substack would need to build some other sort of format to do it. The average reader doesn't want to see it. It's for writers to communicate to writers.
Tho, honestly, I despair of anything like this happening.
The New York Times copies stories from other journalists every day and doesn't credit them. The norm of linking other writers is totally wrecked.
And now you want someone to do a literature review?
It seems hard man.
Might be something for AI tools to try to do? Like rather than the writer do it, AI does it for them. Of course this doesn't force the original writer to actually do the reading, which is important.
Don't really know what to say. This seems a bit hopeless.
After, and I’d add that this has become 10-100x easier to do with modern AI. You take your draft, feed it into Claude, and ask:
“If this were an academic paper, who might it be citing? Who writes about these ideas in this essay, and might they have to say about them? Which of my points are more novel and which ones are covering ground well-trodden by others in the past, and specifically who?”.
This simple approach is likely to yield 90% of the value of a full-fledged academic “literature review”. It’s so easy nowadays that there’s no reason not to bother.
I think that situating your essay would have quite a negative impact on reach and engagement. Of course if everyone did it that wouldn't matter, but I'm just skeptical about the degree of compliance you would need to achieve given social media dynamics etc.
Are you giving general advice, referring to at least one essay that's been submitted to you, or both?
Great insight! I wonder if the medium of a public digital forum has anything to do with this. While it is miraculous that the Internet enables learning on anything and everything, there is no longer the same level of guidance one would normally receive in an academic setting. And with the added emphasis on digital anonymity, how are we supposed to even cite some authors accurately if we don’t know who they are?
You cite the work, not the true name of the author.
Fair! But how would we reference this in such a way as to establish a sufficient scholarly trail if an address/link/URL isn't enough? A citation system needs some amount of identifying information to be useful. Aliases aren't always reliable, especially on the Internet.
A URL + archive system is enough.
Why would a URL not be enough?
This references a previous comment left on this post (this question exacerbates the problem described in the essay above)
That's what hyperlinks are for, yes?
Mere links aren't enough, and even they won't happen sufficiently without a norm to encourage them.