Just a few (i.e. 120) “gibberish” research papers redacted


It’s akin to the Sokal hoax of our digital age, with a touch of Frankenstein in there too, and possibly some monkeys with typewriters. Two major publishers, Springer and IEEE, have identified over 120 papers that made it through their peer-review process but which were actually generated using a software package designed to produce gibberish. You could hardly make it up (<< deliberate pun).

As an attempt to highlight how lax the publication policies of some scientific journal were, researchers at Massachusetts Institute of Technology developed a program that randomly combined convoluted technojargon to create pastiches of scientific research paper abstracts. Stuff like the following, as reported by Gizmodo:

In recent years, much research has been devoted to the construction of public-private key pairs; on the other hand, few have synthesized the visualisation of the producer-consumer problem. Given the current status of efficient archetypes, leading analysts famously desires the emulation of congestion control, which embodies the key principles of hardware and architecture. In our research, we concentrate our efforts on disproving that spreadsheets can be made knowledge-based, empathic, and compact.

I both love and fear that abstract.

They then submitted some stuff and got it published. We can tell that the designers were intending to amuse and teach their audience, rather than to aid the production of garbage by bogus scientists. We can tell this because they published their activities in a report entitled, “Computer conference welcomes gobbledegook paper.”

But it now turns out that other less well-meaning folks have since been using this software for more nefarious ends: i.e., to get their own abstracts written, submitted, and ultimately accepted for publication, efficiently expanding their otherwise disappointing CVs with new self-references to cite.

Over 100 of these were published as conference papers by the IEEE (the Institute of Electrical and Electronic Engineers), which in my experience is held up by some computer scientists as akin to getting a paper published in a prestigious hard-copy journal (computer scientists have moved beyond paper-printing, you see). I’ve no doubt that is generally true, but it is surely embarrassing that the best science is sometimes hard to distinguish from gobbledegook.

Of course it is interesting that Nature is the journal to highlight this affair. As has been covered here before, Nature is a prominent journal that stands as an archetype of the traditional publishing model. Critics want a new model instituted, one that charges scientists themselves (rather than readers or other tax-payers) for the costs of publication.

Some have pointed out that this might introduce a particular profit motive for publishers that ends up lowering editorial standards and encouraging the wholesale publication of less-than-perfect science. The IEEE happens to offer such author-pays options for submissions, so this whole affair happens to neatly highlight the problem of moving towards this form of “open access” publication.

So it supports the view of journals like Nature, who broke the story. Coincidentally convenient, of course.

Leave a Reply