Publishers Withdraw More Than 120 Fake Papers

bmahersciwriter writes "Over the past two years, computer scientist Cyril Labbé of Joseph Fourier University in Grenoble, France, has cataloged computer-generated papers that made it into more than 30 published conference proceedings between 2008 and 2013. Sixteen appeared in publications by Springer, which is headquartered in Heidelberg, Germany, and more than 100 were published by the Institute of Electrical and Electronic Engineers (IEEE), based in New York. Both publishers, which were privately informed by Labbé, say that they are now removing the papers." Looks like journal trolling is really easy.

Comments

Computer Generated Patent applications?

[Rinikusu]

Maybe I should stop right there in case someone gets a bright fucking idea.

Re:Computer Generated Patent applications?

[sexconker]

Maybe I should stop right there in case someone gets a bright fucking idea.

Too late. I already patented the computer-generated patent.

science has no defense against hooliganism

[fermion]

I recall when someone went down my block and the window of every car parked on the street. It was a crime, but really there was no easy way to catch the perp, and we just replaced the glass. We continued to park on the street, did not pay for huge security expense, and it never really happened again. Some kids probably just goofing off. No real profit in the crime. Just hooliganism.

Which is what this seems like. The process of science is not going to jeopardize itself just because some board kids want to vandalize the walls and get attention. If we change the process not to improve it, but just to defend against the Justin Beibers of the world, what good would that do?

As it is there are safeguards in place. As much as people deride the cost of publishing, this reduces the incentive of hooligans to publish purely fake papers. Peer review, which does not protect against purposeful fraudulent papers, does keep a reign on the problem. Then there is simple principle that a single paper is just that, a single paper. It is one data point, and even if referenced widely, is in no way fact.

This also makes me recall the 'confusing' health debate. Like what to eat, what not to eat, etc. The problem is that many people read a popular media report based on a single piece of research and think it is true. This misconception indicates the problem with science education in America. That one result is meaningful. That our basic principles of science were developed fully in one paper, with no background, and no adjustment as more data was taken. For instance, relativity was based on at least hundred years of research. Einstein pretty much observed single discrepancy in the magnetic/electrical field and formulated a correction.

Re:science has no defense against hooliganism

[Obfuscant]

Just hooliganism. Which is what this seems like.

This. Someone found a system that was based on trust and decided to try to beat it. Yawn.

Conferences are not journals. The peer-review comes during the presentation, not when the abstract is submitted. If the session moderator doesn't know the submitter, maybe he'll look at the abstract a bit more closely, but he's not going to send the abstract out to three other people in the field to vet it. So it gets published.

A very long time ago someone did this as a joke at a conference I went to. The talk was about "a hole in the bottom of the ocean. There's a log in the hole in the bottom of the ocean. There's a frog on the log in the hole in the bottom of the ocean..." OMG! How awful. A bit of fun on the session moderator's part. Nobody got shot or fired. We all survived.

Re:science has no defense against hooliganism

[enzo1]

Peer review, which does not protect against purposeful fraudulent papers

We aren't talking about fraudulent papers; we're talking about joke papers that aren't even reviewed, so much as proofread and rubber-stamped.

Re:science has no defense against hooliganism

[pz]

Conferences are not journals. The peer-review comes during the presentation, not when the abstract is submitted. If the session moderator doesn't know the submitter, maybe he'll look at the abstract a bit more closely, but he's not going to send the abstract out to three other people in the field to vet it. So it gets published.

Depends on the field. Some conferences are very hard to get into, and there is a rigorous peer-review process (I can think of two of the conferences in my field with acceptance rates at or below 30%). Others accept essentially anything and everything (like those semi-scam conferences in China I keep getting invited to). I'd wager, however, that the majority of meetings are somewhere in the middle, accepting most submissions that sound reasonable, as the organizers trust that the cost of attending and presenting is sufficient disincentive to trolls.

Re:science has no defense against hooliganism

[drinkypoo]

These papers should all have been rejected. They weren't. This opens a big can of worms

It doesn't. This is a known problem. It's accounted for in two ways. One, most of those papers will never be cited by anyone. Two, if they are cited by anyone, it increases the chances that people with brains will read them and decide they are bullshit; they may even try to reproduce the results.

Maybe if academic departments valued quality.

[hey!]

I recently did some literature research into ontology technology, and was shocked by how many papers were pot-boilers that disguised trivial ideas with inflated language. These were papers that had absolutely no discernible academic value other than to pad a resume, and collect but a smattering of citations, mostly from similar papers. In comparison the seminal papers, the ones that get tons of citations for years to come are robust, thought-provoking and well-written.

Granted the well-written part probably has something to do with attracting future citations, but I think the trivial nature of the useless papers probably has something to do with their obscure style.

The key here is "Conference Proceedings"

[JoshuaZ]

In many fields conference proceedings have little to no oversight. These papers don't get noticed at all or cited and for most purposes don't exist. The only real issue I can see here is that a large fraction of these are apparently coming from China and this is consistent with prior reports of serious problems with academic quality coming from China. It is possible that people are using these essentially fake papers to boost their publication counts which may give them some advantages as long as no one looks closely, but any institution that is a serious institution will look at everything one has published. I actually found this point more interesting:

Labbé emphasizes that the nonsense computer science papers all appeared in subscription offerings. In his view, there is little evidence that open-access publishers — which charge fees to publish manuscripts — necessarily have less stringent peer review than subscription publishers.

Considering how many complaints there are about low-quality open-access journals, this suggests that that isn't nearly as much of an issue as some people are claiming.

Re:The key here is "Conference Proceedings"

[Anonymous Coward]

In CS, well-refereed conference proceedings often exceed journal publications in their contemporaneous impact and prestige value to authors. CS journals typically take 2 years to process a manuscript to publication. Way too slow for a fast-moving field. Good conferences will have a 20% manuscript acceptance rate (been there, done that, as author and reviewer, many times) with only useful papers presented and put into proceedings, only a few months after research results have been written up.

I had the privilege of educating a dean about the value of selectively-refereed conference proceedings for academic computer scientists. It worked. Also educated about the adjunct value of releasing research software to the specialized communities, if the usage and impact can be documented.

Re:The key here is "Conference Proceedings"

[don.g]

I have no mod points, so take this as a "me too". CS conferences have prestige and a high impact value -- and the papers in the proceedings are full papers, not just abstracts.

In some other fields (a friend of mine tells me this is what biology is like), conference talks may be submitted with just an abstract, and the proceedings may not contain much more than Powerpoint slides. I've never seen a CS conference where that would be considered remotely acceptable practice.

Re:Maybe he could submit some news to slashdot!

[plover]

Assuming he hasn't already...

He obviously hasn't. His algorithms are too good to produce all the dupes we see here.

Reject this Crap

[DrNico]

As someone who reviews papers (by humans) for conferences and regularly says "reject this crap" (politely, and with reasons) only to see the paper accepted, I'm not too surprised.

EXCEPT --- It's worse than THAT!

[TrollstonButterbeans]

Most of these computer generated papers have valuable ideas we need to consider.

Statistics indicate that 1 in 24.3 of these computer generated papers have uniquely valuable scientific advancements. But the real-world ratio is about 1:99.7 --- the 3 sigma rule.

If these computer-generated papers are exceeding the productivity of the actual papers by a 4 to 1 margin, a big opportunity is being missed and it doesn't matter why.

A true case of an unintended result exceed the effectiveness of your average deliberate result. Short version: a 4% rate actually exceeds the real-world discovery rate. This should not be ignored, coincidence or not.

Next step is getting citations :)

[jopsen]

Next step must be to make computer generated papers that gets citations... :)

Re:Next step is getting citations :)

[StripedCow]

Well, just produce a paper-generator that adds citations to previously generated papers.