1 min read30th Oct 201425 comments

11

In line with my continuing  self eduction...

What are the most important or personally influential academic papers you've ever read? Which ones are essential (or just good) for an informed person to have read?

Is there any body of research of which you found the original papers much more valuable than than the popularizations or secondary sources (Wikipedia articles, textbook write ups, ect.), for any reason? What was that reason? Does anyone have a good heuristic for when it is important to "go to the source" and when someone else's summation will do? I have theoretical preference for reading the original research, since if I need to evaluate an idea's merit, reading what others in that field read (instead of the simplified versions) seems like a good idea, but it has the downside of being harder and more time-consuming.

I have wondered if the only reason to bother with technical sounding papers that are hard to understand is that you have to read them (or pretend to read them) in order to cite them.

 

New to LessWrong?

New Comment
25 comments, sorted by Click to highlight new comments since: Today at 11:13 PM

In Chemistry in particular, and the natural sciences in general, I find that reading textbooks is a much more efficient way to digest knowledge than reading papers. The largest advantage which reading papers confers relative to reading textbooks is that textbooks rarely cover the newest of the new advances in any field. I rarely find that I need to read a paper to learn something that I can't find in a textbook-- this is probably because, in the natural sciences at the undergraduate level, people don't often need to find information which was discovered within the last five years. The major exception to this trend is people who specialize heavily within a particular field, such as PhD students, postdocs, professors, and the like.

There are other reasons why reading individual journal articles can be helpful, but since you asked this question from the perspective of someone hoping to continue their efforts at self-education, I would advise you to stick with textbooks, for the most part.

Also, reading meta-analyses of papers, which will themselves be published in journals, is often better (in terms of efficiency and knowledge gathering power) than reading individual studies.

Papers gives you empiric arguments for why something is true. Textbooks often only assert facts. That makes reading textbooks easier but sometimes you are interested in the underlying arguments for a claim.

Maybe he doesn't want to absorb knowledge but see how it looks at the frontier?

I personally find lecture notes superior to textbooks. Text authors frequently fail to edit their work down to the essentially material. When I am first learning a subject I want the key results. Once I understand the core ideas I can look up more details when I need them. Textbooks can be great for reference but as an introduction I like notes.

I am mostly familiar with studying math, economics and computer science. So I am not sure how my experience compares with people studying other subjects.

Is there any body of research of which you found the original papers much more valuable than than the popularizations or secondary sources

Medicine, in particular nutrition. My prior is that mass-media reporting is just nonsense on stilts, pretty much always, and you have to look at the original paper to see what the authors tried and what they found (which, often enough, is junk, anyway).

You might be asking the wrong question. For example, the set of papers satisfying your first question:

What are the most important or personally influential academic papers you've ever read? (call this set A)

has almost no overlap with what I would consider the set of papers satisfying:

Which ones are essential (or just good) for an informed person to have read? (call this set B)

And this is for a couple of reasons. Scientific papers are written to communicate, "We have evidence of a result--here is our evidence, here is our result." with fairly minimal grounding of where that result stands within the broader scientific literature. Yes, there's an introduction section usually filled with a bunch of citations, and yes there's a conclusion section, but papers are (at least in my field) usually directed at people that are already experts in what the paper is being written about (unless that paper is a review article).

And this is okay. Scientific papers are essentially rapid communications. They're a condensed, "I did this!". Sometimes they're particularly well written and land in category A above. But I can't think of a single paper in my A column that I'd want a layman to read. None of them would make any sense to an "informed" layman.

My B column would probably have really good popular books written by experts--something like Quantum Computing Since Democritus, or, like others have said, introductory level textbooks.

I acknowledge that they are separate questions.

I hope asking the wrong questions leads me to the right ones. Thank you.

rapid

LOL.

"I didn't have time to write you a short letter, so I wrote you a long one" I think the rapid part is in terms of the writer's time, not the readers'.

I think the rapid part is in terms of the writer's time

Do you have an idea how long it takes to write and publish a paper in a peer-reviewed journal?

I didn't mean the readers' time: probably in average it takes me less to read an academic paper than to read a Slate Star Codex post, at least if the latter is tagged as “long post is long”. :-)

In a world of publish or perish and a lot of articles getting rejected I don't thing the problem is that researchers don't invest enough time in writing papers. It's rather that there are incentives for writing in a way that signals sophistication.

Peer review also adds extra time for the communication process.

Pure unhelpful nitpickery: Your title should say "Academic", not "Accadmic". It doesn't really matter, but since it's sitting there in big letters on the main Discussion page it might be worth fixing up.

Nope. I should care about the most basic signaling at least, and I've come to rely on those little red lines to tell me I've got a speelign error.

Fixed.

(you also have "soundign" in your article)

The more technical and abstruse a paper, and the less you are an expert in the area yourself, the more you should rely on secondary sources that may be able to present it in a more user-friendly way. There is, after all, no point reading the original if you can't truly understand it. However, some academic papers are written in a sufficiently comprehensible style that almost anyone can be enlightened by them.

Some that have had a particular effect on me:

The last is the most cited law review article of all time, in no small part because of its accessibility.

On the topic of which: an economics paper which made a big impression on me is Hahnel & Sheeran's "Misinterpreting the Coase Theorem" (Journal of Economic Issues, 43, pp. 215-237). Unfortunately there appears to be no freely available copy of the published version online, but there is a preprint without the figures. It's a bit less accessible than Coase's paper, but I imagine pretty well anyone who's taken a microeconomics class could follow it.

Is the basic argument that same as the one in this blog post?

It doesn't look like it. (Assuming the post you're linking is this; I think your URL omits the last "l".)

Nick Szabo's argument is that the Coase theorem only works by assuming away prior allocations of rights that allow one party to overtly coerce or inflict violence on another.

Robin Hahnel & Kristen Sheeran's core argument is that parties engaging in Coasian bargaining often do so under incomplete information, which opens the door to Pareto inefficient bargains. [Edit: or indeed to no bargain being reached at all, a possibility I once illustrated with someone else's example.]

More in the 'personally influential' than the 'essential for an informed person':

The Long Term Evolution Experiment is one of my very favorite in the biological and physical sciences. It's decades old by now, and using that time to actually test the behavior of bacteria on true evolutionary timescales. They have seen multi-part ecosystems evolve from monocultures, and used the lab environment to 'roll back the tape' and see if evolutionary patterns can be made to jump through the same hoop twice.

I have alot of fun exploring Academia.edu (I have papers posted there myself). I think going to the source is important when you want to understand an idea at a deep level and you wish to really digest something. Research, of course, is best done at this level.

I'm currently going through this paper on attempts to explain the individuality of subjective experience, and I'm feeling relieved that this guy has made it his life's work to research questions that have obsessed me since childhood.

Carnap's Empiricism, Semantics, and Ontology. Posting as a response here because I have a feelings it may be vaguely relevant.

I would mainly look for review papers. These papers summarise a large deal of knowledge and can give you a sense of what is and isn't known within a field.