In academic scandal, publishers embrace the next step in digital publishing: Machine-generated works
February 28, 2014 | 6:48 pm
In one more of the series of scandals currently impacting the world of scientific digital publishing, a French researcher has discovered that two prominent scientific publishers released some 120 papers that have now been proven to be machine-generated fakes. As reported in Nature, Cyril Labbé of Joseph Fourier University in Grenoble exposed 16 fake papers published by by Springer and more than 100 by the Institute of Electrical and Electronic Engineers, in 30 sets of published conference proceedings.
Labbé now operates the SCIgen detection website, for detection of papers created by the popular SCIgen automatic gibberish thesis generator, originally written as a satirical joke but now used in all seriousness, apparently. In Nature, he attributes the glut of drivelish research to the pressure on researchers and institutions to publish as much as possible, with quantity rather than quality the main criterion, and on publishers to turn a profit. The publishers themselves claim to be maintaining their quality control standards, but this latest development cannot give any comfort. The whole process of peer review may need re-evaluation in this context, but Labbé’s comments also indicate that open access publications are just as stringent as their more commercial brethren – and if this string of fakes is anything to go by, possibly more so.