Ad Lagendijk
8 September 2009
Can we refer to Wikipedia articles in a scientific paper?
Tags: references, WikipediaPosted in Getting published, Web 2.0
I think Wikipedia articles should never be allowed as references in the primary scientific literature.
Generation gap
The young generation is on Facebook, MySpace and Twitter. The older generation, if participating in a social network, will join the more sober-headed LinkedIn. Young people even leave a virtual social network if they discover that too many members are of the old generation. I still remember situations where faculty members refused to use a computer or email.
In many cases they were cheaters, because all that computer work was done by their secretaries. Innovations are invariably accompanied by people denying their usefulness. After a period of habituation the new development is widely accepted. This pattern of denial, hostile reception to acceptance is notorious. Any critical remark on a modern development is returned with “You are old-fashioned. You better get used to the new situation”.
By the same token, any critical remark on the importance of web 2.0 developments for science is reciprocated by this “Wake up old guy!”. I am sure the rest of this post will meet the same resistance.
Wikipedia
The Internet encyclopedia Wikipedia is great. I use it frequently. Not very often for my disciplines mathematics or physics, because I discover that these entries are in the majority of cases unclear, incomplete, if not wrong. I am sure that professionals in other disciplines have the same experience with Wikipedia entries in their fields. But if you want to get a quick impression, or get useful references, Wikipedia is excellent.
Problems with Wikipedia articles
There are a number of design principles that make Wikipedia articles unsuitable to function as scientific references:
- The articles change over time
- No single person, or group of persons, is responsible for the content
- Contributors are anonymous, pseudonymous, or are using a false name
Only if all three problems would be solved my opinion would change. The best solution would be to only allow trusted people to contribute, and permit them to contribute only under their real name. If a certain number of trusted contributors (say five) agree, the article can be irreversibly frozen. Somebody is trusted if (i) his/her identity is known and confirmed and (ii) he/she knows about the subject. How one can make somebody “trusted” can be solved in many ways. I will come back to that later in an other post.
So people who complain that referees are old-fashioned when the referee of their scientific paper deletes their referrals to Wikipedia articles. or reject their paper because of the presence of references to Wikipedia articles, still have a lot to learn.
References to web sites in scientific papers
Scientific journals accept references to web sites. Don’t the above objections 
also apply to these references? I think not. If the references are to well-maintained archive sites my objections certainly are not relevant.
If the reference is to a personal website, it will be the website of the authors of the paper. That means we know who is responsible. Furthermore the web site uri is given in order for the reader to get quicker access to papers (for instance downloading pdf fies) of the authors. These papers have been published in scientific journals. That is too say the reference to the web site is never used – or at least should never be used – to point to a location where scientific arguments can be found.
Or subscribe to our RSS feed.
8 Sep 2009 12:57, Jacopo Bertolotti
Having been a Wikipedia admin for over 3 years (the italian one) I totally agree that referring to wikipedia entries in a scientific paper should be strictly forbidden. What I don’t agree about are the reasons.
What is here presented as a list of problems is in fact the very core of why and how wikipedia works. There were attempt to create a free on-line encyclopedia made by trusted (as per the definition above) people. Nupedia was an early attempt but Citizendum or Knol are modern and well designed. The problem is that they don’t work. Their articles, although written by recognized specialists, are no better than wikipedia’s ones and their number is so small than you will likely not find there what you are searching about. The best proof of their failure is that you are not using them and that most of you never even heard their name before. Wikipedia if full of omission and plain mistakes but is the best we have. Moreover in most cases is enough: right today I needed to refresh what a Toeplitz matrix is and the (very short) entry in wikipedia was more than enough. Even the entry in MathWorld contained less useful informations.
So why I’m strongly against including references to wikipedia in scientific papers? The reason resides in the very nature of what wikipedia actually is: Wikipedia is a secondary (sometimes tertiary) source of information. This means that Wikipedia do not attempt to present new and original knowledge, it just aim to present you with an easy-to-find and somehow pre-digested version of the knowledge you can find somewhere else. All the information you can find there on Toeplitz matrices can also be found on a standard textbook, and often the books and/or paper from where the information comes from are explicitly stated at the end of the page. If, for some reason, in the future I will need a reference for a property of a Toeplitz matrix I can just go and take one of the many works of Robert M. Gray (I don’t know him but Wikipedia links to his review on Toeplitz matrices so I just guess that searching two minutes I’ll find a properly peer reviewed papers where the property I’m interested in is properly discussed). Referring to Wikipedia is just lazyness, just like it is lazyness to refer to the Encylopedia Britannica
8 Sep 2009 13:55, DM
In comparison to Encyclopaedia Britannica or Encyclopaedia Universalis, on mathematics or theoretical computer science, Wikipedia is much more complete. Of course, it does not compare to a specialized monograph – but the purpose of an encyclopedia is to give you a quick refresher and point you to relevant articles, textbooks or monographs, rather than give you a complete course.
By the way, are there really people who send articles to refereed publications and cite Wikipedia as proof of their statements? Or are you making this up as an example of what should not be done?
8 Sep 2009 14:09, Ad Lagendijk
@Jacopo.
Thanks. Really interesting to know that efforts with “trusted” authors failed.
@David
A few days ago the new editor in chief of the (Dutch) Nederlands Tijdschrift voor Natuurkunde (*the* magazine for Dutch physicists) wrote that this had exactly happened to her: as a referee she refused to accept references to Wikipedia and the authors responded in the way I outlined in my post.
8 Sep 2009 18:11, dfghj
Wikipedia would rather people didn’t cite it in papers since if you cite wikipedia wikipedia can’t cite you which makes life problematical.
8 Sep 2009 21:07, Philip Chimento
Efforts with trusted authors have not necessarily failed. They just haven’t been around long enough to succeed. When Wikipedia was as old as Citizendium is now, it was far less useful.
9 Sep 2009 8:28, David Bradley
What do you mean by “young” – according to recent news reports teenagers certainly aren’t using Twitter and while some of my middle-aged and older contacts are using LinkedIn as a serious others see it as a frivolous alternative to Facebook, which they reserve for serious business use. There are always people using tools in ways that a simple throwaway remark cannot capture…
9 Sep 2009 8:49, Jacopo Bertolotti
@Philip: Nupedia is older than wikipedia and it was such a failure that Wales and Sanger (the two creator) closed it. Then Wales tried the path of not requiring any trusting method and created Wikipedia while Sanger continued on the path of an encyclopedia created by experts and created Citizendum.
I agree that Citizendum is young but it doesn’t seems to have an exponential growing in the number of articles as Wikipedia had so I just doubt it will ever really bloom and I’m afraid it will always remain on the level of “a nice attempt”.
Another possibility is to compare Wikipedia with Matworld. Mathworld is older than wikipedia, cover just one specific subject, is written by trusted authors and is a nice place where to search for info. In my experience it is quite rare to find a serious mistake in Mathworld but often the lack of useful information on even the most basic part of math astound me.
11 Sep 2009 1:43, David Stern
My former PhD advisor Cutler Cleveland is editor of Encyclopedia of the Earth: http://www.eoearth.org/ It has a traffic rank of 123,063 according to Alexa which is not so great I guess…
11 Sep 2009 10:46, Bram van Ginneken
If you call a reference to Wikipedia lazy, then giving no reference at all is even more lazy. Giving a reference to a textbook or an overview paper without having read the book or paper, that’s also lazy. Who will testify here that you have never done that?
I think one should distinguish different types of references. A wikipedia reference is not an alternative for a reference to a journal paper and I can think of many examples where a Wikipedia reference in a paper would be inappropriate. But if you mention a general concept, and that concept may not be familiar to many of your intended audience (for example I publish about computer algorithms that analyze medical images in journals aimed at radiologists and I may mention things most radiologists are not very familiar with), and Wikipedia or Mathworld or some other web site has a nice entry on it, why would it be a mortal sin to cite that entry?
Most journals I know require references to a web site to list an access date. With that data you can go with a few clicks to the exact revision of the Wikipedia article that is being referenced. So that solves Ad’s problem #1.
Educating students about what type of references are appropriate when, and when a reference is really needed, and why you should preferably refer to a journal article instead of to an obscure conference paper, all of that is very important if you want to write good publications that are likely to be accepted. Blindly saying that references to Wikipedia should never be allowed, I find that a bit too strong.
An interesting example of a resource that might meet all three requirements from Ad is http://www.uptodate.com/. It holds basically the current state of the art in medicine. It’s written by experts. If you end up in a hospital, as most of you here will at some point, and it’s not directly clear to the physician treating you what should be done, chances are that the doctor will consult this website. I see residents in our academic hospital using it all the time. If they would actually have to read the real literature, that would take much more time. And they are not lazy, most not at least, they just have a lot of patients to treat… And yes, it’s commercial, no free access for patients, but it’s not as expensive as an old-fashioned encyclopedia either.