Do scholars produce too much and revise too little?
Art Carden asks me:
…do you think scholars spend too much time producing new content and too little time revising and refining their arguments? I’m thinking about the Scholars of Old (e.g. Smith) taking their work through multiple editions. Thomas Sowell is good about producing revised versions of some of his books, but a glance at my shelf makes me wonder if Eminent Scholar should’ve revised his/her first book instead of writing a second or third or fourth.Do you think books go through the optimal number of revisions? Is the editing process so good today that the first edition usually should be the last?
In music, Brahms is notorious for having spent a lot of time revising his drafts. Pierre Boulez explicitly revised and improved many of his pieces, after adding in years of extra work. Stravinsky’s later 1947 version of Petrouchka is much sharper and cleaner than his 1911 release. In all of these cases the revisions are worthwhile, because these works were very special and worth improving. Brahms’s first symphony might have done better with some further revisions yet.
When it comes to economics, individual works are less and less special all the time, unlike in the days of Adam Smith. Smith waited, more or less perfected his book, and in the meantime he was not really scooped. Today it is the collective body of work which carries the force. We also live in the age of the working paper, where it is the first released draft which matters most. There is an institutional failure that first released drafts are released too soon, for the purposes of receiving attention and credit. Yet I don’t think there is a corresponding problem of too few editions of the working paper. Ideally there should be no more than two. A first “here I am” version, to stimulate discussion and feedback, and eventually a final version which maximizes accuracy.
In fact if scholars had to commit to only two versions of the paper, they might be induced to make the early release more accurate in advance, knowing they cannot magically revise away errors a week later with the magic of word processing and web re-posting. This move toward fewer editions would offset some of the costs of premature release. Again, we do not see a case where a greater number of revisions would be better.
The scarcity of attention is the key reason why a very small number of accuracy-enhancing revisions is more or less optimal.