Science community looks for ways to stop fake papers
Some science pranks are good fun, but fake scientific reports are an ongoing problem for journals
The start of April brings, of course, April Fools' Day — and scientists aren't above getting in on pranks and practical jokes.
But as CBC Radio science columnist Torah Kachur explains, those pranks can sometimes be harmless fun — or can present a real problem for the scientific community.
What notable pranks have we seen in the science world?
Perhaps the most famous scientific April Fools' Day prank was in 1976, when BBC Radio 2 astronomer Patrick Moore convinced many in Britain to jump into the air at exactly 9:47 a.m. to take advantage of his entirely made-up claim that an auspicious alignment of Jupiter and Pluto would alter gravity and allow Earthlings to experience weightlessness.
And there are many others, like the story of a famous French scientist Jean-Baptiste Perrin, who in the 1920s reportedly put a gyroscope in a piece of luggage and left it on the train platform for someone to help carry it.
A gyroscope is a momentum machine which drives you in one direction — and since it was left running in the luggage, the momentum carried the luggage only in one direction.
So as soon as the luggage carrier tried to change direction, the luggage wouldn't budge. And according to legend, the porter dropped the suitcase and ran off screaming, "The Devil himself must be inside!"
And even here at the CBC, former Quirks and Quarks host David Suzuki once got reeled into a prank interview about a giant shark.
Have there been published scientific papers that were actually pranks?
Sort of, but not specifically April Fools' pranks.
In 2005, three MIT computer scientists decided to design a computer program called SCIgen, which can randomly generate computer science papers — complete with graphs, citations and figures. They look, and sound, very real.
Sounds very smart, but it's really gibberish. The stated point of this website is simply to "maximize amusement" — which it does.
One sentence from the fake paper I generated reads, "Suppose that there exists the simulation of the partition table such that we can easily visualize pseudorandom epistemologies."
Which you might think clearly looks like pseudo-scientific nonsense. But consider a very real paper I wrote during my PhD, which was published in a high-tier journal, and included the sentence "In situ hybridization shows that unc-45 mRNA is enriched in the gonad of adult worms as well as being quite strong in two- and four-cell embryos."
Kind of hard to tell which one is real and which one is fake, isn't it?
How often do these fake papers get published?
More than the scientific establishment would probably like to admit.
Fake and flawed papers are supposed to get weeded out in the peer review process. This is where the publisher farms out the editing work to experts in the field that the paper covers. The idea is that the legitimacy and accuracy of the science gets vetted by peers.
That peer-review process is a cornerstone in science, and works very well for high-tiered journals. But in some smaller trade journals, the existence of fake papers and fake reviewers seems to be fairly high.
There's even a blog dedicated to it. Retraction Watch follows dubious practices in science, and has reported that over the past three years, 15 per cent of all retractions happened because the papers were vetted by fake reviewers — not because of faked data or experiments that lack control, but entirely fictitious reviewers, with fake email addresses and the like.
In 2014, more than 120 papers were removed by publishers when a French researcher found they'd been created by the random generator SCIgen.
Which may sound funny, but it's definitely no joke for the scientific establishment — because there are definitely people who have started to figure maybe they can get published without doing all the work that usually entails.
What can be done about fake scientific papers?
It's really up to the publisher to be more rigorous. Not accepting reviewers with emails like firstname.lastname@example.org is a start. Publishers generally require institutional emails for any contact these days.
And they're fighting back with their own computer program. SCIgen, which started as a tongue-in-cheek program, has actually helped the publisher Springer develop something called SciDetect as a way to fight back against fake papers.
It's a computer program that quickly scans papers for typical grammar and topics that fake papers are known for. So far, its creators say it has picked up hundreds of fake papers and prevented them from being accepted for publication.
The bigger issue remains that SCIgen was made as a joke, but it's being used as a tool by desperate scientists to publish, lest they perish — the moral of the story being that sometimes, a prank can truly go awry.