Here’s why I think that’s the case: In philosophy and economics, progress often comes from diagnosing and correcting the discipline’s own bad habits and cancerous mistakes — some of the most significant contemporary work in either field is work that exposes hidden biasses, implicit assumptions, unwarranted intuitions, and so on. And in both philosophy and economics, the biasses that need to be corrected are often ones that are specific to the discipline, and have a historical origin, rather than general human biasses. (I doubt that most people would intuit that perfect information is an innocuous assumption for an economic model that’s meant to predict the way markets behave, or intuit that qualia is bound to reduce into physical/functional facts under the right analysis.) While it surely is possible to spot, diagnose, and overcome present disciplinary biasses simply by being an incredibly good economist or philosopher, one thing that’s very helpful is to look at the inception of currently prevalent methods, assumptions, intuitions, and so on. Reading the original gives you the best view of the web of beliefs from which a position or an idea first originated, and of the reasons for which it became popular within the discipline. While a notion’s having a problematic history (e.g. the arguments that made it commonplace are nothing like the arguments used to defend it today, or it’s a claim that was originally inferred from a now-discredited premise but nowadays people cite it as raw intuition or common sense) isn’t gonna matter if you’re absolutely certain that the modern usage of the notion is good philosophy/economics, and also isn’t gonna matter if you’re absolutely certain that the modern usage of the notion is bad philosophy/economics, it is a seriously useful discovery if you are anywhere in the middle. Not only does such a discovery give you a good reason to take whatever suspicions you presently have about the notion’s soundness more seriously — and to pursue developing them into definite arguments –, it also gives you useful hints about what bad implicit assumptions to look for or what intuition pumps to construct in attacking the notion.

I’d point out that this does, I think, explain why “in all those disciplines where we have reliable quantatative measurements of progress (with the obvious exception of history) returning to the original works of past great thinkers is decidedly unhelpful.” Once a discipline has an idiot-proof way of seeing what works, it doesn’t have to spend so much time on spotting, diagnosing and uprooting its own bad habits.

Physics doesn’t care if the great physicist Pauling said something was true. The court cares a great deal if the prior esteemed members of their court (or better yet a superior court) have said something.

I don’t doubt that reading Smith is a better way to learn about what the historical person thought or what it was like to invent these economic concepts than reading a modern economist. However, that’s not why people think Smith’s work is great. Few people would think it important to read Smith if he’d been a crackpot with totally crazy theories even if he’d had the same subjective experience of developing them and was equally good at capturing the sense of the times. What I’m arguing against is the idea that it’s useful to read Adam Smith to be good at economics.

Again I don’t doubt that Smith’s work inspires all sorts of interesting side thoughts. However, so too do modern economic papers. The question is whether reading Smith actually inspires so many more interesting thoughts than the same time spent reading modern economic texts that it makes up for all the misguided notions, false starts, and incomplete presentation that always plague the first description of novel work.

One reason I think many people falsely believe the great originals have these superior properties is that in school they are exposed only to great originals and modern textbooks so the great originals are the only things they read that present original research, and grapple with unsolved questions. But that’s a consequence of this fetishization of the originals not a necessary situation.

But there are a few instances where returning to the originals has some utility. I went to the U of C as an undergrad and experienced the whole great books thing. The value of reading, say, Kant in a general education context is not learning what Kant thought–learning particular facts has limited utility–but in learning how to think, how to understand difficult material, and to be skeptical and critical of what you read.

Of course it has value. Reading the little comments some undergrad left in the margins of the critique has some utility as well. Maybe reading “Kant’s so gay… I hate this fag,” will spark an important philosophical thought about the role of sexuality in metaphysics. It’s surely more likely to produce useful thought than if you literally did nothing (ceased to think) during that time.

That’s not the question. The question is whether it’s a better use of your time than reading something else. Personally I don’t think offensive margin notes by an undergrad make the cut and I would continue to think so even if they happened to give me a great idea once.

The problem here is you directly observe the ideas you got reading Kant while the ideas you might have had reading some modern philosopher during the same time are much less salient to you since you can’t point to them.

Looking at your argument as a piece of argumentation, I would say you have presented a successfully organized essay. But it seems to me that you really do understand the importance of reading original writers.

To belabor the point, you argued well and I am sure you will revisit the classics, or at least check them out of the library and scan them as needed.

One of the really fun things about going back to the hoary old original sources is the fascinating side thoughts that have been overlooked by the “distillers”.

I had occasion recently to read “The Theory of Moral Sentiments” by Adam Smith. This wildly popular work predates his famous book by 20 years. The time span between the two books covers the Watt steam engine and the American Revolution. Reading the older book is like a visit to the “modern mind” while the secondary science of Economics was solidifying out of the churning components of the Enlightenment.

So I say, feel free to ignore the distillers anytime.

But I definitely wouldn’t recommend reading him to learn economics. You can gain a better knowledge of the subject more quickly by picking up any modern economics textbook.

But there are a few instances where returning to the originals has some utility. I went to the U of C as an undergrad and experienced the whole great books thing. The value of reading, say, Kant in a general education context is not learning what Kant thought–learning particular facts has limited utility–but in learning how to think, how to understand difficult material, and to be skeptical and critical of what you read.

Second, the picture is quite different when your goal is not to learn what’s already known but to do new research. I’m a grad student in math, clearly a field that benefits enormously from distillation and simplification of work, but when doing research it’s almost always necessary to return to the original papers except for the oldest, best-understood topics. Understanding the entire history of thought on some topic is quite useful in this case. For example, the original proof of a theorem may be long and convoluted, but the paper may contain useful nuggets about how the authors actually came up with their proof or half-baked conjectures and ideas that are needed to move forward.

progress is impossible because it’s always better to learn from the mistakes of past great thinkers

It’s useful for a while but brings diminishing returns.

Reading great thinkers might also be useful if it gives you a glimpse at not just what their thoughts were but also how they had their thoughts. It’s not useful to know that some greek philosopher thought men had more teeth than women, but it’s useful to know that he was dumb enough to think that.

BTW if you had a traceback section this would probably be in it: http://lesswrong.com/lw/1ul/for_progress_to_be_by_accumulation_and_not_by/

If this was the most read blog post of all time, I’d be a happy man.

I would add that in law school, many classes require you to waste a lot of time reading old cases in order to learn one little point of law.

Fortunately, most law students learn pretty quickly that you can buy commercial outlines which summarize the important cases in easy-to-read English.