The citations-versus-downloads conversation continues. Nature Neuroscience‘s editors have analysed the number of downloads a paper receives immediately after its appearance online, and find a high correlation with its citation frequency years after publication. Associate editor Noah Gray provides the details at the journal’s blog, Action Potential.
Despite well-known concerns about impact factors, he notes, “these numbers are typically used to rate the importance or prominence of a particular journal, and thus by proxy, the importance of the individual papers published within”. This flawed association often leads individuals and organizations to equate the total number of citations with scientific impact.
Instead, the editors wanted to measure readership of an article that would reflect its levels of outside interest and perceived value. Although the “number of downloads” measure is subject to misuse and has its own flaws, it provides a piece of an alternative solution for a more informative picture of manuscript influence.
For further discussion, join the Nature Network group Citation in science.
Why do I get the feeling the computer scientists will win the download game?
More specifically, Bob:
This begs for a virus that propagates onto other machines, waits a random time and downloads my paper once from each new machine.
Yes, Jon, one would need an addtional system such as COUNTER, to “credentialize” the downloads. There is some interesting debate on this at the Citation in Science forum, where Lee-Ann Coleman makes the additional point that site licenses complicate the “number of downloads as metric” proposal (there are other points, plus some clarifications from Noah Gray of Nature Neuroscience).
A system such as you discuss would require a more sophisticated program, one that selectively invades the machines of colleagues before downloading the paper-to-be-inflated. Perhaps such a program could consult a list somewhere on the web, setting an “invasion finished” flag by the name of the investigator whose machine was just invaded, then sending progeny copies to the next few targets on the list. This would establish a set of credentialized machines from which one-time downloads of the targeted paper could be executed.
A still more sophisticated program might check a distant website occasionally to see whether the nefarious author has published again and then download that new paper (perhaps with a determined probability of download so that there is variation, publication-to-publication, in the set of downloading machines).
Of course, if all that energy were spent instead on research we’d likely be better off. However, if P&T depend on downloads of my paper then perhaps it is in my best interest to learn some programming…
Agreed with your sentiments, Jon.
When I wrote “COUNTER”, I meant the compliance organisation, whose website is here
It is an industry standard, so that when a journal or other publication states visitor, page views or other web statistics, you know whether to believe them or not depending on whether the organisation is COUNTER-compliant. (For example, most people’s individual blogs or small websites are not.)
Agreed, though, that as a metric downloads have a lot to be desired, for all the reasons you say and more.