No one reads your paper either

Simon Goring wrote an interesting post a few days ago arguing that no one reads your blog. In his post, he discussed reasons for why you might want to blog anyway. This post prompted me to tackle a topic I’ve been thinking about for a while: Why should you do science? Why should you publish? No one is going to read your paper either.

In particular, while I’m sure everybody agrees on the value of publishing groundbreaking, highly influential papers, most papers are neither groundbreaking nor highly influential. I have written over 100 scientific articles, and by my own modest guess at least 80 of those are completely irrelevant in the grand scheme of things, and probably more. My own measure for acceptable performance in a given year is that I publish about one paper to which I can point and say “probably more than 10 people will read this and actually care.” Over a 15-year career in science, that comes out to about 15-20 such papers, so I’m more or less on track [1]. Nevertheless, the vast majority of the papers I write, on average about 6 per year, are not that special and aren’t really read by anybody. So is there any reason to write them at all? Why not just write one big hit a year, or whenever it’s ready, and leave all the chaff unpublished?

Quality tends to correlate with quantity. In your field, how many people do you know who publish one Nature or Science paper a year and not much else? I can think of maybe one or two people who have made that strategy work for them, but I’d argue that they are successful despite their publishing approach not because of it. For most people, productivity and impact tend to be correlated. The most prolific scientists also write the most widely-read papers and vice versa. It’s just like in the music business, you need to write a lot of songs to write a hit, and the people that write many hits also tend to write many many songs that aren’t hits. And, more importantly, it’s difficult to predict which song will be a hit and which will flop.

Let me illustrate this idea with an example from my own career. In 2001 I wrote seven papers. Five of the seven have been quite successful, and have been cited over 50 times each. One of them, a Nature paper [2], is still one of my most cited papers of all times, with about 400 citations according to Google Scholar. When we wrote that paper, I was pretty sure we had a big story, so the success of the paper was somewhat expected [3]. The second-most cited paper of that year, however, which at present has about 100 citations and is my 13th most cited paper of all time, was a total surprise. Notably, this paper was also one of my fastest ever from original idea to submission. I had a cute idea one afternoon, did a few calculations for a day or two, wrote a simulation in a few more days, and after a little over a week the entire paper was done and out the door. When I submitted this paper, I had no idea whether it was even publishable. Yet it got printed, and people really liked it. For the other papers I wrote in 2001, I also was skeptical whether anybody would ever care about them, and in the end five of the seven did really well. It would have been easy for me to be more selective in 2001 and write only 2 or 3 papers, but if I had done that I’d have missed out on writing a couple of well-received papers.

This leads me to lesson 1:

You don’t know which of your papers other people will like. The more you publish, the higher the chance that something you write is useful to somebody.

Frequent publishing builds an audience. My next point applies to scientific publishing just as it does to blogging: You need to build an audience. If you regularly publish on a topic, people start taking notice. They will make it a habit to keep an eye on your work, even if they don’t spend a lot of time actually reading your papers. Then, when you finally publish something really good, you have an audience ready and waiting. The exact same paper, published by somebody nobody has never heard of, might not get the same reception. In fact (and I’m not saying this is right, just stating how things are), it might not even make it through review the same way.

Thus here is lesson 2:

You need to keep publishing regularly so people are aware of you and take you more seriously.

There’s a limit to this, of course. If all you ever publish is useless drivel then you’ll eventually gain a reputation for publishing useless drivel. Stick to my guideline from the beginning of this post, publish at least one paper a year that is actually good.

Frequent publishing develops skills. My next point is also just like in blogging: You need to write a lot to get good at writing. With every paper you write, you get a little better. Therefore, even the papers that don’t really have any content of consequence, where you know only three people will ever read them [4], are worth writing just for the exercise. Can you imagine you make a major discovery, a real breakthrough, and then you don’t have the skill to turn your results into an excellent paper? That’d be a disaster.

Thus, we arrive at lesson 3:

You need to publish as much as possible so you develop the writing and responding-to-reviewers skills that you will need to write your major hits.

Lesson 3 applies particularly to students and young postdocs. They need all the practice they can get and hence should publish a lot, even if some of their papers turn out thin and inconsequential.

Publishing clears the mind. My arguments so far can be summarized by saying that more publication quantity equals more publication quality. Some might raise the following objection: By writing fewer papers, they will say, they can put more effort into each one and hence raise the overall quality of their output. In other words, quantity can be detrimental to quality. I am not convinced by this argument. As I said earlier, one of my top-cited papers of all time took me all of two weeks to complete. At the same time, I could point you to papers that I put a huge amount of effort into and they never went anywhere. Impact doesn’t necessarily correlate with effort. More importantly, though, I believe that if one tries too hard to control quality both quality and quantity suffer. We need to publish papers so we are mentally ready to move on and write more, hopefully better, papers.

In fact, ideas for papers can hurt your productivity if you get stuck on them. The most insidious ideas are the marginally good ones, the ones that aren’t so bad that they are easily dismissed out of hand but also aren’t good enough that you want to rush to publication. Those kinds of ideas can drag down your productivity for years, because they divert your attention and energy while not going anywhere. The best thing you can do with those ideas is to just publish them and move on. It’s so much easier to dismiss an idea as bad once the paper is out. I have plenty of papers on my cv where I’m the first to say they are pretty pointless [5]. But I don’t regret publishing them. If I hadn’t published these papers, maybe I’d still be thinking about the central ideas expressed in them.

Thus, the final lesson:

The best way to get a weak idea out of your head is to publish a paper on it. Don’t dwell on weak ideas, publish them and move on.

There you have it, my reasons to keep publishing even though nobody reads (most) of my papers anyway. Now I’d like to hear from you: What are your reasons to publish or not to publish?

Notes

[1] Until you realize that the argument I made was circular, because I arrived at ~80 irrelevant papers by assuming that I’m writing about 1 relevant paper a year, give or take.

[2] C. O. Wilke, J. L. Wang, C. Ofria, R. E. Lenski, and C. Adami (2001). Evolution of digital organisms at high mutation rate leads to survival of the flattest. Nature 412:331-333.

[3] To be honest, though, 400 citations and counting still blows me away. When we submitted the paper, I hoped that it would garner maybe 50 citations or so.

[4] Those three being the two reviewers and the editor of the journal.

[5] For example, everything I published in 1998 and 1999.