Tuesday, May 5, 2009

Who needs science when we have automatic-science!

So a few years ago some computer science students made a new computer program that will automatically create a scientist's research paper complete with text, figures, tables, methods, results etc. The papers are filled up with lots of technical terms and a lot of nonsense.

As a lark they submitted one of these automatically generated 'research' papers to the 2005 World Multiconference on Systemics, Cybernetics and Informatics, which was accepted for presentation at the conference! The 'authors' confessed to their hoax on their website and soon the conference organizers withdrew their acceptance. The program that automatically generates research papers was made available online (http://pdos.csail.mit.edu/scigen/). You can type in your name as an 'author' and you're sure to get a 'high quality' research publication. I couldn't help typing in my name and seeing what research papers could be automatically generated in my name (actually creating the papers will probably only be funny to nerds like me - note that when I tried it the program was smart enough to self cite my own fictional work 4 separate times, now that's attention to detail!).

Another automatically generated research paper (Towards the Simulation of E-Commerce, by the fictional Herbert Schlangemann) was accepted for publication in the International Conference on Computer Science and Software Engineering. To make matters even more hilarious, Herbert Schlangemann was invited to be a session chair during the conference (a prestigious invitation that involves introducing a set of speakers and moderating the question period that follows the talks).

Now, these two examples are both for conferences which are often considered an easy place to get research published. Now the main way for a scientist to advance their career is through the accumulation of peer-reviewed journal papers, which as I've said in previous posts can be a remarkably challenging thing to accomplish. Which led me to the question: If it is hard to get a journal paper published, does this mean that those papers that do get accepted for publication in a journal represent good quality research?

Absolutely not!

It turns out that a few students in Iran also took one of these automatically generated research papers but submitted it to a journal: the Journal of Applied Mathematics and Computation. Hilariously, it was accepted for publication: you can see the final accepted paper here (http://ce.sharif.edu/~ghodsi/soft-group/misc/AMC-paper.pdf).

I think a serious overhaul of the research publication process is in order!