NNNS Chemistry blog
Prevous: You can call me Al=Al
Next: Spectacular magnet failure

All blogs


07 October 2017 - Wikipedia

thompson20172.PNGFrom September 2014 to June 2017 Wikipedia user Carolineneil was creating one new chemistry page after the other. Typical titles were 'Modified Wittig-Claisen tandem reaction' and 'Multivalued treatment'. These articles had serious issues and fellow editors and administrators spent many hours sorting everything out. What was wrong with these articles? The scope was too narrow, the content topic often existed already in another article. The articles invariably were not up to Wikipedia styling standards. And Carolineneil was never available for discussion.

The problem is that on a regular basis new editors contribute new articles, very often in the context of a school or university project. And why not help these people out, we are doing volunteer work after all? It is a regular discussion item in the talk pages, does the Wikipedia chemistry project benefit (new content, potential new editors) from school projects ? or is it a burden (rubbish content)? But now it has emerged Carolineneil was not selflessly managing a school project but was rather advancing his own interests, abusing Wikipedia for the purpose of a science experiment.

Carolineneil is Neil Thompson, an MIT professor and the resulting paper is out here. In it Thompson attempts to prove that Wikipedia has a distinct influence on scientific papers both in ideas and in the language used. He paid a bunch of PhD students to create all these chemistry articles, then planted them in the Wikipedia universe over the course of 3 years and then recorded how the body of scientific literature changed accordingly. But does an article on the Wittig-Claisen tandem reaction created in 2014 inspire new scientific research or does the specific wording re-appear in subsequent Wittig-Claisen articles? Is this something you can measure? Is it more than a plagiarism-detection tool?

The Wikipedia community has responded in force (discussion here) and not in a positive way. Biggest offences: the articles are rubbish, volunteer time was wasted and editors should never be paid for writing content. To make matters worse, the Wikimedia Foundation was in on it, providing advice.

And the Thompson Big Data publication itself? Why would scientists take note of Wikipedia? The encyclopedia is never ever cited in scientific articles and if it happens it is only ridiculed. Those active Wikipedia editors with an scientific research background more often just complain (plastering pages with "citation needed" tags) or abuse the system for self-promotion. The Wikipedia community was negative about the article, Mark Zastrow writing in Nature is enthusiastic (DOI). It is certainly a long read with over 30 pages. The gist of it: in the set of published Wikipedia articles the document similarity with scientific papers tended to increase more than in the control set of unpublished articles.

The peer-review process will provide a final judgement on the article but for now I do have a personal beef with Neil Thompson: as a Wikipedia editor (V8rik), do not waste my time! Thanks.