Diminishing Returns and Life

I think Sturgeon's Law is applicable here. If 90% of authors write nothing but crap (a conservative estimate), then even if most authors who do write something valuable only write one valuable book, you'd still have better odds reading an author you've found valuable in the past than one you haven't so long as more than 10% of authors who write one valuable book also write a second one. Based on my own experience, I'd say that's the case.

I posted that question. Thanks for responding! You and Blackadder both make good points.

How about shifting the advice to avoid becoming overread in any one niche area? Read lots of work by really smart people in a wide variety of disciplines and skip the followers (the "normal" scientists in a Kuhnian sense). You seem to encourage interdisciplinary thinking...

You didn't ask for requests, but I'll put in one for more on why you hate macro. I am learning a lot about macro--and its flaws--from those posts!

I disagree. Anyone who has written one really good book, has probably another good book in him. I mean, if you have a colleague you find very insightful, do you, after a year, say, he has nothing left of interest? No, you infer his insight's quality is autocorrelated, and so, if he wrote a good book on WW2, his book on WW1 will probably be a good read.

Isn't that why you check out the same blogs, day after day?

I'd change "The folks at lower-tier schools think they belong at top-20 schools" to the folks at lower tiered schools want blinder editing at the top journals, and a better set of referees for their papers at those journals.

Arnold,

Congratulations! Tyler Cowen has declared you to be "wise," even though you are not a lunch regular like Bryan.

If you become bitter once you get on the ego treadmill, do you get a gun and a bible and start hating foreigners?

Would you also apply the law of diminishing returns and life to sports and other forms of entertainment? How much weekly TV shows that seem to get better each week (i.e. if I watch the first episode of the Soprano's, then it follows that I will want to watch the next one even more so than the previous one).

From one hard-core bleeding-heart libertarian to another:

I think that fiction -- or at least genre fiction -- provides a clear counterexample to your assertion about lightning not striking twice. If I like something by Heinlein, or Stross, or Westlake, then I should keep reading more of their books.

I believe this also applies to non-fiction.

To be frank, your assertion sounds an awful lot like saying "That restaurant served a good meal the last time we went, so it probably will not be able to repeat the performance."

Am I overlooking or misunderstanding something here?

I think an exception to the "one book worth reading" rule can be made for writers in certain clearly-identified genres -- mysteries, thrillers, science fiction, historical fiction, for example. Within the oeuvre the quality may vary, but later books often approach or even match the quality of the first one.

I think the proposed book rule works if modified for writers who are of 'historical' note. (I.e., dead for years.) If you read their 'famous' books, you probably needn't read their obscure ones unless you really like the famous stuff - the 'market' of fame generally retains the good stuff and sheds the bad.

In general, though, author quality has a random element with mean 0 and variance q and an element F (which can be considered a kind of fixed effect in our panel data analysis of literature), so the quality of a book by a given author is drawn from a variable mean F, variance q. Arnold is saying the variance swamps the mean, and so book quality is primarily a crapshoot. I disagree, for two reasons.

One - it is apparent that people have different objective writing abilities - bad shakespeare generally exceeds good most everyone else. F may be drawn from a distribution, as well - Shakespeare would be an outlier draw, but it retains its value within an author's output. Some authors 'get lucky' and get a big, positive q but then don't replicate it, but others have good F. Expected value of the second book of an author with a good first book is less than the first book, but better than average.

Two - writing STYLE is correlated across books for most authors, so your first resort for finding books of a style you favor would be authors whose style you like.

In reality, "Quality" isn't an objectively measurable thing, it is a, well, subjective and qualitative variable. And "style" is a separate, also qualitative variable. Beauty in the eye of the beholder and all that.

I think the problem in the model is in taking account of individual differences. Even with diminishing returns, there are "talented" writers who start at a better chance to produce a good book (say, 25% for the first , 20% for the second, and so on diminishing, while the average for all writers are e.g. 5% for the first, 4% for the second, etc.). These differences don't even out in the average, because once you got a good book, it's disproportionately more probable that it comes from a "talented" writer, than from an average one. So it pays to stick to book of the same writer (at least for a while).