I saw this tweet from New Scientist about the Kepler mission. Kepler of course was an amazing mission that completely changed our view of the universe. But the numbers got me thinking a bit about the cost of a scientific publication.

It says that 2946 scientific papers were published, I guess using Kepler data. Since the mission cost about 700 million USD, that gives about 230 kUSD per published paper.¹
Is that much? Well, I mean I’d personally rather have a Tesla two Teslas than one publication… but in the grand scheme of things? For the US it counts to about 2-3 times the salary of a senior researcher. So if every senior scientist publishes a paper every two years, with no other expenses, we’ll end up somewhere there. But that’s obviously not realistic. I vaguely remember reading, many years ago, some estimate of the total cost for publishing a paper but I cannot find it now. But we can make our own crude estimate.

Looking at my own recently finished projects I see that they tick in at around 50-55 kUSD per paper – a bargain! But that is far from all the costs that go into the making of a paper. It doesn’t include minor items like my salary, but also not things like lab infrastructure and major equipment. To account for that we might look at the entire institute. My institute is one the top institutions in Poland. We had a yearly budget of about 22 MUSD (PPP) in 2018 and we publish about 200 papers that year. That gives us some 110 kUSD per paper. But we’re still counting a bit low here. We do collaborate and those collaborators are also paid, but not from our budget. So what if we zoom out a bit.

What is the cost of a paper if we look at a national level? In the table are the R&D expenditure for some selected countries, together with the number of publications from Scimago⁴ We then get a MUCH larger number, and one that varies wildly from country to country! An American paper clocks in at almost 800k, while a Chinese one is still more expensive. China is doing a lot of research infrastructure investment, which is not going to pay off for several years or decades. Researchers in the UK seem to super cost-efficient, but still… that’s a lot of Teslas!

But speaking of Teslas: A lot of the R&D spending is not intended to result in publications. The numbers include industrial R&D spending as well, which probably does not end up in the publications record. But the internet comes to our rescue. Unesco has a list of the “SHARE OF R&D EXPENDITURE BY THE BUSINESS SECTOR“, and using data from there we can create a new table. If we compare only public spending on research we should get a more realistic measure of how much a country has to spend on each publication.

Now the numbers are comletely different. In the US, China and Germany a paper seems to cost somewhere north of two hundred thousand dollars. British and Swedish publications seem to come very cheaply, but the Wikipedia data is from a different source (World Bank) than the others (Unesco). The Polish data comes from the Statistics Poland, the government statistics agency, and agrees with the Unesco data.
That Polish publications are cheap might not be so surprising considering how little researchers are paid here compared to the other countries in the list. It’s also kind of agree with the data from just our institute. There is a reputation, in my opinion mostly true, that parts of the Polish university world is very unproductive. Something that cannot be seen from this data. The caveat of course is that we say nothing here about the quality of those publications…

But to conclude, it seems like a typical scientific paper costs somewhere between 100,000 and a bit over 200,000 dollars. We need to remember that this is an average – I guess a publication in the humanities⁶ is less expensive than a typical medical publications, with chemistry landing somewhere in between. In any case, given the monumental shift in perception we’ve gotten from the Kepler mission, not only was it remarkably successful, it was also quite the bargain.

1. Of course this is a serious underestimate, since many researchers using Kepler data are not funded by the Kepler budget.
2. https://en.wikipedia.org/wiki/List_of_countries_by_research_and_development_spending
3. According to https://www.scimagojr.com/countryrank.php
4. I don’t know how good their data are, but at least they’re accessible and easy to find.
5. Data from GUS, with PPP factor from OECD.
6. But this might just be my prejudice!