Thursday, April 19, 2007

Wikipedia as primary source: just what is original research, anyway?

Erik Moeller, back on March 13th, wrote this email to the WikiEN-L mailing list, entitled "Wikipedia as a primary source". It seems mainly to address the topic of when the subjects of articles wish to interact with the Wikipedia editorial process and change the content of their articles.

It is quite common for Wikipedia to have articles with factual errors. Disturbingly common, in fact. It is also not surprising that the people who have articles about them that are in error would like to have them corrected. What is surprising, at least to me, is that many Wikipedia volunteers have been refusing to correct errors in such articles upon complaint from the subject until and unless the subject goes to the hassle of publishing a statement (online, of course, since these people can't be bothered to read any sort of offline source) correcting the factually false information on Wikipedia. Erik's comments seem intended to address this quite silly behavior, and to the extent that they do so they're on point. (Incidentially, I also agree that the Foundation should not be involved in this process any more than necessary.)

However, the whole role of primary sources in encyclopedic authorship is still underdeveloped at Wikipedia. There seems to be an categorical attitude that primary sources should never be used, not even to verify (or discredit) claims from other sources. In general, this reflects back to the broad misunderstandings of Wikipedia's No original research policy. Wikipedia should not have original research, yes, this is true. However, verifying facts is not original research; it's verification. Similarly, removing alleged facts which cannot be verified or which are actually found to be false is also not original research; it's, once again, verification. Verification is such a critical part of the encyclopedic authoring process that one must wonder at the sanity of those who propose with a straight face, that Wikipedians should not do it.

And this is why I oppose the deletion of this template. A private email might not be a valid encyclopedic source to assert a point, but it is definitely a valid source to confirm one. It makes sense to provide the means to record that a fact has been confirmed, by whatever means were used. I imagine it'll be used infrequently, but it might still end up being used from time to time. Deleting a citation format won't prevent people from doing original research, and it'll just make it harder for people doing real, proper, work to document what they've done.


  1. Ah, but how do we know that the email is from who it says? An email to disprove something with no citation is fine but, per my comment in the deletion discussion, is no better in support of an assertion than Cite:phone call (or perhaps Cite:what some bloke down the pub said)

    Using information from emails to remove crud is needed but, unless the definition of a wikipedia reliable source is rewritten, can't be used in citations.

  2. Oh my, I wish I had seen that debate earlier. I've used email and phone calls several times to discredit and verify facts. There is nothing "original" about calling the first employee of a major corporation and asking where the original store was, as I did here. -