Fiction writers used to have to submit their manuscripts to 6 or 7 big corporations. Only these corporations could seriously publish a book. Room on library stacks has always been scarce. Only a tiny number of authors could ever become independently wealthy under this system. Amazon.com is changing the game by getting rid of the scarcity. Hugh Howey put his books for sale on Amazon (receiving 70% of the sales instead of the 12.5% offered by publishing houses) and he made a killing. He is a millionnaire by now. He did not have to convince a highly selective editor to accept his books. He simply wrote some great fiction. He created value and was rewarded for his work. He did not have to displace other authors on the bookshelves.
I am sure that many people do not consider him to be a real author because he self-published. Our society is very driven by this fight for status, for elitism. I think it is probably responsible for a lot of unnecessary stress. People overwork, pollute and die younger than they should because of it. And it affects science as well.
There is a common view in computer science research that the only publications that matter are those appearing at selective conferences. Conferences in computer science are characterized by a low acceptance rate (top conferences reject 90% of all papers). The more selective the conference, the better. Ideally you must prove that you belong to the top 1%. A journal article or a workshop paper is “wasted effort” in this respect.
Does it follow that journal articles are inefficient or even wasteful? According to DBLP, Donald Knuth (a living legend) published 120 journal articles, and 12 conference papers. The Turing Award recipient Peter Naur has 25 journal articles and 7 conference papers in DBLP. (The Turing Award is the Nobel prize of computer science.) Robert E. Kahn, another Turing Award recipient, has 12 journal articles and 2 conference papers. Even if we assume that they participated in many conferences that are not indexed by DBLP, it is undeniable that some influential computer scientists like to publish in journals.
But maybe all of the important work appears first at conferences? According to Fortnow, leading conferences regularly refuse to accept such work:
(…) nearly half of the Gödel Prize winners (given to the best CS theory papers after they’ve appeared in journals) were initially rejected or didn’t appear at all in the top theoretical computer science conferences. (Fortnow, 2009)
To land a paper in a very selective conference, you still had to beat incredible odds… acceptance rates are routinely under 10%… surely this says something about your work? Maybe being accepted by the most selective conferences proves your worth. Maybe not:
The view that conference rejection rates are a good proxy for conference quality did not hold up to scrutiny (Freyne, 2010)
I believe that people like to tell themselves simple stories about how one should succeed. Many of these simple stories are based on half-truths. Just like how fiction authors believe that they must land a competitive book deal to be a writer whereas none of us care about any of that. This status ranking game that you play… is probably much less important than you make it out to be on the long run.
Franceschet wrote an interesting survey where he identified the 10 most prolific computer science authors, the 10 most cited authors and the most prestigious computer scientists (i.e., Turing Award recipients). There was no overlap between the 3 lists. It is worth taking time to reflect on this fact.
- If you could find a way, somehow, to become the most prolific computer scientist in the world, you are not likely to figure in the 10 most cited authors or win a Turing Award.
- If you could become the most cited computer scientist in the world, you may not receive a Turing Award.
It means that if you are an extremely successful computer scientist, you are still likely to be left out from someone’s top-10 list. Maybe it means that you should not worry about how you are ranked.
There are also significant differences on how the 3 types of authors published.
- To get a Turing Award, your publications may be unimportant. Out of the 16 Turing Award recipients considered by Franceschet, two thirds had fewer than 50 papers on DBLP and five had less than 30 papers. Alan Kay published less than 20 papers according to DBLP.
- “(…) high impact scholars publish significantly less than prolific ones, and more frequently in journals.”
So while telling the world about what you do matters… you have a lot of freedom about it.
What should a sane computer scientist do then? His main focus should be on producing lasting contributions to his field. He should then publish them where they are likely to be noticed.
If all you have ever done is fight for scarce spots at a selective venues, you have achieved nothing of importance. Really important work creates tangible value that is self-evident.
Note: To be fair, I have never achieved anything of importance and probably never will. But I am having a lot of fun.
Further reading: Mentoring Advice on “Conferences Versus Journals” for CSE Faculty by Kevin W. Bowyer and How Are the Mighty Fallen: Rejected Classic Articles by Leading Economists by Gans and Shepherd.