We all know about higher education being squeezed financially. Government support is falling and tuition is going up. We see academic jobs getting scarcer, and more temporary. The pressure for research to focus on the short term is going up. Some of these changes may be fair, since society always has to balance its immediate priorities against its long-term progress. At other times, like when comparing the NSF’s $7.6 billion FY2014 budget request to the ongoing travesty that is military procurement, it does feel as though we are eating our seed corn for not very wise reasons.
Against this backdrop, the travesty that is scientific publishing may feel like small potatoes. But now we are starting to find out just how many potatoes. Tim Gowers has been doing an impressive job of digging up exactly how much various British universities pay for their Elsevier subscriptions. Here is his current list. Just to pick one random example, the University of Bristol (my former employer), currently pays Elsevier a little over 800,000 pounds (currently $1.35M) for a year’s access to their journals. Presumably almost all research universities pay comparable amounts.
To put this number in perspective, let’s compare it not to the F-35, but to something that delivers similar value: arxiv.org. Its total budget for 2014 is about 750,000 US dollars (depending on how you count overhead), and of course this includes access for the entire world, not only the University of Bristol. To be fair, ScienceDirect has about 12 times as many articles and the median quality is probably higher. But overall it is clearly vastly more expensive for society to have its researchers communicate in this way.
Another way to view the £800,000 price tag is in terms of the salaries of about 40 lecturers ($latex \approx$ assistant professors), or some equivalent mix of administrators, lecturers and full professors. The problem is that these are not substitutes. If Bristol hired 40 lecturers, they would not each spend one month per year building nearly-free open-access platforms and convincing the world to use them; they would go about getting grants, recruiting grad students and publishing in the usual venues. There are problems of collective action, of the path dependence that comes with a reputation economy and of the diffuse costs and concentrated benefits of the current system.
I wish I could end with some more positive things to say. I think at least for now it is worth getting across the idea that there is a crisis, and that we should all do what we can to help with it, especially when we can do so without personal cost. In this way, we can hopefully create new social norms. For example, it is happily unconventional now to not post work on arxiv.org, and I hope that it comes to be seen also as unethical. In the past, it was common to debate whether QIP should have published proceedings. Now major CS conferences are cutting themselves loose from parasitic professional societies (see in particular the 3% vote in favor of the status quo) and QIP has begun debating whether to require all submissions be accompanied by arxiv posts (although this is of course not at all clear-cut). If we cannot have a revolution, hopefully we can at least figure out an evolutionary change towards a better scientific publishing system. And then we can try to improve military procurement.
O brave new quantum world!
In that case it’s simple:
Fuck the establishment and publish your work only on open access journals e.g. PLOS
Fuck Elsevier they’re known Mfs.
Once thing that I learned from Gowers’s new post is that there is a nice project by Scott Morrison who has set up a website called The Mathematics Literature Project where arXiv versions of papers in recent mathematical journals are collected. http://tqft.net/mlp/wiki/The_Mathematics_Literature_Project
It would be good to see more Quantum Pontiff essays upon this general theme. After all, if there were no Elsevier, wouldn’t all these problems remain?
A template is Alberts, Kirschner, Tilghman, and Varmus’ recent essay Rescuing US biomedical research from its systemic flaws (PNAS, 2014), which concludes
If this is true for biomedical research, how much *more* true is it of QIT research? Especially because (as it seems to me and many) that QC/QIT/QSE and biomedical research are destined to become inseparably linked? With all the tremendous opportunities and sobering risks that this linkage conveys?
NB Quantum Pontiff readers may wish to reflect that in every developed nation of the world (including the USA) military expenditures are a small fraction of healthcare expenditures (invariably 20% or less). To the degree that QC/QIT/QSE research can lead in the future — as such research has in the past — to healthcare economies that are more effective and/or more efficient, then perhaps here can be found some eminently practical opportunities to help end academia’s present-day “QIT/QC Winter.”
Inspired by Aram’s economic analysis, and cautioned by Clay Shirky’s much-discussed essay The End of Higher Education’s Golden Age (January 29, 2014), and informed by Wikipedia’s “List of countries by military expenditures per capita” and “List of countries by total health expenditure per capita”, the following statistic is given for the wealthiest 20 countries:
$latex (military)/(military+healthcare)$
• 46% Israel
• 34% Greece
• 22% United Kingdom
• 20% United States
• 19% France
• 19% Australia
• 18% South Korea
• 18% Norway
• 17% Finland
• 16% Luxembourg
• 16% Italy
• 15% Denmark
• 14% Sweden
• 13% Netherlands
• 11% Spain
• 11% Japan
• 11% Germany
• 11% Canada
• 11% Belgium
• 9% Switzerland
The relative homogeneity of this ratio was (as it seemed to me) surprising, and the median ratio of 16% supports the view that a 21st century “Golden Age” of research can be funded more securely by improving the health-care sector of the global economy, than by cutting the military sector.
——-
Israel’s outlying ratio of 46% is notable too, as (perhaps?) helping to explain the depressed status of biomedical research in that nation, as exemplified by the distressing (to me personally) economic travails of Hadassah Hospital in Jerusalem, regarding which Gil Kalai and his colleagues is likely better informed than me.
Conclusion The economic foundations of STEM research in general, and quantum research in particular, deserve a more thorough, more integrative, more historically informed consideration than they presently are receiving.
Summary Ample economic margins exist each way; either for ongoing STEM deterioration, or for a 21st century Golden Era.
John, you make a lot of good points. Indeed creating a utopian academic publishing system will not save us trillions of dollars. But there is a reason to fix the problems in an area where we are experts and have a lot of leverage, before wading into territory where lots of other people have strong opinions and vested financial interests. The impact, although small on a “eventually the sun will burn out” scale, is large relative to the size of our research community.
Aram, you and I (and most researchers) absolutely agree regarding the desirability of open-access scientific literature. And we further agree that this policy (1) saves money, and (2) augments equity.
Which is the more significant advantage: economy or equity? Wrangling over this point is scarcely the best use of our time!
And I think we agree too, that a foreseeable dystopian outcome, to be avoided by all means, is a more-and-more open literature, that enables a larger-and-larger pool of students, to compete with greater-and-greater desperation, for a smaller-and-smaller pool of jobs.
That’s the crucial topic that Shirky’s The end of Higher Education’s Golden Age does not touch upon at all, regarding which Alberts et al.’s Rescuing US biomedical research from its systemic flaws prescribes mere palliative measures.
Previous generation of quantum researchers collectively set-forth some mighty effective roadmaps, and (as it seems to me) the present generation has ample STEM materials to do the same … but lamentably has not done so to date.
John, you did not write conclusion and summary to last comment. Could you clarify, did you mean that in situation with samller and smaller pool of job access to open literature should be limited?
QuantumBot, my own view is that a bad STEM circumstance (of too-few jobs) has pushed us toward a worse STEM circumstance (of academic access restricted to wealthy institutions).
These adverse circumstances are so strongly coupled and reinforcing (as it seems to me) that either they will remediated together, or else they are unlikely to be remediated at all.
Further considerations regarding the mutually supporting roles of prophets, traditionalists, and revolutionaries are posted on the Fortnow/GASARCH Computational Complexity weblog, under GASARCH’s outstandingly provocative topic “Every Theory Grad Student should know …”
John, modern electronic technology made access to information simpler and much less expensive. Difficult financial circumstances just make reasonable to use that. You wrote earlier that access to open literature should be avoided because it enables people to compete. Cf. wikipedia item about “obscurantism”
John, what we are arguing about? You wrote that (1) open access is desirable (2) but under certain circumstances it should be avoided because [read above …] . Is not it?
QuantumBot, please let me state plainly that the outcome preferred (by me and everyone) is more jobs *AND* more openness.
And it’s evident that these two outcomes are correlated, isn’t it?
EXERCISES Go to your library and open the bound volume (or licensed on-line access) to Scientific American from 1955.
Exercise 1 Students not at elite institutions can’t even begin to do these exercises. How does that access-inequity affect the STEM community?
Exercise 2 Count-and-read the STEM job ads from 1955. Set forth a feasible STEM roadmap that will restore the present paucity of job ads to the abundance of that past era.
Hopefully you will enjoy these exercises, QuantumBot!
John, I am not quite understanding what you are trying to illustrate by your excercises. During last 10-15 years I am obtaining all necessary publications via internet or e-mail. It is about few hundreds items per month. The problem with access is known problem, but they hardly can have an economic nature.
A standard example – just recently I needed one hight impact article published 30 years ago. The publishers claims $30 for downloading 200kb pdf. I think, real cost of the operation is ~1000 times less. Any comments?
“QuantumBot” the point of my comment is that the Quantum Pontiffs might beneficially ask broader questions, per the example of the real-world Vatican Pontiff.
What are some good journals for quantum computing/information papers? (Good both for open access and for quality.) The arXiv is important, but we should still pressure our journals to be open access.
Just to be clear, there will always be resource constraints on science, just like on pretty much every area of the economy, except during periods of rapid expansion. This is normal and not a sign of crisis of civilization.
I think that the idea that there are too many papers stems from misguided nostalgia. Are there too many books or movies or bands or youtube videos? Too many for whom? As with many things that could be on a FAQ for nerdy conversations, xkcd has a solid answer:
https://xkcd.com/1227/
Aram, your reasoning is inarguable … indeed it can be made considerably stronger as follows.
Today’s young quantum researchers will still be active when our planet’s population approaches $latex 10^{10}$. Hopefully most people’s lives will be secure, prosperous, peaceful, and free; we will assume that this outcome is feasible.
From common-sense reasons set forth elsewhere, it is evident that our future world’s economy and technologies will press against fundamental quantum and thermodynamical limits to efficiency, size, speed, and resource cost.
In this utopian mid-21st century circumstance, it is scarcely credible that the proportion of humanity that studies QM/QC/QIT/QSE (etc) can be less than $latex 10^{-6}$, or that these researcher will publish less than one article per year, and so we can lower-bound the foreseeable rate of quantum-articles at one article every thirty-two seconds.
Supposing that each researcher can afford $1000 per year in literature access charges, by mid-21st centry the affordable cost-per article will be lower-bounded at 0.1 cents.
Conclusion To borrow a phrase from Mark Twain, “Any calm person, who is not blind or idiotic, can see that we should begin now to reduce literature access charges to less than a penny per article.”
I think the most important cost is likely to be the cost of our attention. This is also an effective lower bound on tuition, for example.
It is worth noting, though, that the cost of arxiv.org is about 1.4 cents per download (or $6 per upload). Peer reviewing costs are nonzero, but don’t need to be very high.
Looking at my past posts makes me realize I need some new fixations!
Journal subscriptions are clearly a major cost for libraries, but many of the open access journals have exorbitant publication charges which needs to be met by the authors. This seems suboptimal too, in that many people will simply not be in a position to publish in these journals. Certainly the arxiv delivers phenomenal value for money, and a journal which could survive on even 20-30 times the arxiv per paper rate would be a welcome addition to the community. From a technical point of view, this seems quite possible if you can avoid the need for a full time staff. However, it seems that the only way to see a journal of this form in the near future is to actually start it. Form an editorial board and start looking at the options.