I'm wondering how much Covid-19 affected scientific productivity, if it affected productivity negatively at all (c.f. Newton actually did a lot of his most important work while the University of Cambridge was closed due to plague).

I'm using the rate at which papers are published on arXiv as a proxy for "scientific productivity", fully aware that it is not a perfect proxy and probably lags actual productivity by quite a bit. If country-specific statistics are available, e.g. "The number of publications by authors from Italy changed by __% on 9 March 2020, the date the lockdown started", that's even better.

  • 1
    Do you assume all researchers are at home or still at work? Can they / did they take material home? Is it possible for them to work effectively at home? Are you assuming they will get more work done at home? no students to distract them, based on your point about Newton...
    – Solar Mike
    Mar 19, 2020 at 8:37
  • 1
    @SolarMike are you assuming all these things about the question that aren't actually in it?
    – Allure
    Mar 19, 2020 at 8:43
  • 1
    No I clearly asked if you were making those assumptions...
    – Solar Mike
    Mar 19, 2020 at 8:52
  • 4
    @SolarMike my interpretation is Allure is asking an empirical question that does not require assuming anything. It sounds like you’re trying to come up with a theoretical model to answer the question by pure reason/theoretical analysis, and want to know what assumptions you should make, but I don’t think that’s what’s being asked for.
    – Dan Romik
    Mar 19, 2020 at 9:12
  • 3
    I think too short time has passed to have some statistics. Speaking for myself, I had to put on hold at least one ongoing manuscript because I had to stop the experimental activity. For another manuscript, nothing has changed because the experimental results were already completed. For third manuscript, which I planned to start writing at the beginning of March, I instead postponed the start at the beginning of April because I'm now too busy organizing the online lectures, and I'll probably submit it with fewer experimental results. Mar 19, 2020 at 9:20

3 Answers 3


ArXiv posts its monthly submission statistics here. It's noisy data even in the absence of viruses, so I can't tell if there's an effect yet or not. Then again, I'm no statistician. Extrapolating the March data as of today (March 19th), it's at least set to be larger than the February number of submissions, as is typical for March submissions.

However, even if the number of monthly arXiv submissions is to be considered a good proxy, it will likely lag by more time than we have seen so far. Yes, closures, adapting to online classes, and changed schedules will delay new experiments and other activities. However, finishing writing up manuscripts already in the pipeline, or analyzing already collected data are still productive tasks that will lead to new submissions for quite a while. Not to mention people such as myself, who can continue our theoretical/numerical work from home.

  • 1
    Rough estimation using arXiv statistics: 8938 articles in March (as of March, 22) so about 12,594 for the whole month. About the same as March 2019 (12,654). Of course, March submissions are the results of months of research, so it doesn't mean anything.
    – Taladris
    Mar 22, 2020 at 2:59

I think it will be interesting to study this 5 years from now, but right now is way too soon. At least in math, people don’t post papers at a steady, constant rate. I personally post around 3 papers per year, and most of those represent work that is at least a year old (often much longer than that) by the time I get around to writing it up. Given what I’ve got partially written right now, I suspect that I will post about the same number (or even a little more) this year, bit that’s just part of the ebb and flow of research. What is harder to predict is the effect of all this chaos on the production of new math, which won’t be visible for several years.

  • More than math, many experimental fields will be affected, all those in which labs have been shut down. Mar 19, 2020 at 20:52
  • 1
    That’s absolutely the case. I feel terrible for my colleagues who can’t event comfort themselves with their work in these dark days. Mar 19, 2020 at 22:31

I think your assumption that scientific productivity can be measured by arXiv uploads is flawed. Surely it can measure that the usual workflow has been changed - but especially for short periods of time, productivity cannot be measured by number of papers.

Taking myself as an example. I am using my time at home to do research tasks I have put off for too long. Cleaning up a large code base. Revisiting some old results. These are not tasks that results directly in arXiv uploads, but tasks which are needed in the long run. What I hear from my colleagues, is that they are doing similar things - if not completely worn down by teaching and supervising remotely, of course.

Not the answer you're looking for? Browse other questions tagged .