Wikipedia’s epistemological methods

A colleague of mine recently asked me about Wikipedia’s policy on sources and evidence, Wikipedia:Verifiability (WP:V). In short, the threshold for including content in Wikipedia is that it be “verifiable, not true”. Truth alone, without appropriate evidence that fits with the Wikipedia community’s standards, is not enough to justify adding something to Wikipedia.

You can interpret this in a number of ways. For some, it’s an embodiment of post-modern notions of truth and subjectivity (people disagree about truth, so we don’t let people simply add what they know to be true, instead relying on authority). For others, it’s just a practical concession to the sociological nature of Wikipedia, in which some people are more objective and more capable than others (and those are the people that know how to leverage authority effectively). The Verifiability standards could also be taken as a fundamentally rhetorical, rather than epistemological, policy: communal standards of evidence ensure a basic level of apparent reliability, since readers can be pointed straight to relevant authorities. (Citizendium, in contrast, as has looser evidentiary standards and relies in part on the personal authority of its Authors and Editors.)

From an academic standpoint, there are plenty of relevant sets of literature that bear on the problems that Wikipedia’s evidentiary standards and policies attempt to deal with. But from my own perspective as a historian of science, I think the parallel to scientific epistemology and evidentiary norms is an interesting one.

WP:V works in ways that are closely paralleled in scientific (and historical) method as it is actually practiced. Communities of scientists have various norms (mostly unwritten) for what does and does not constitute legitimate evidence for making novel scientific claims. These norms are highly context dependent, and can include (for exclude) experiments, reference to the work of others, reasoning and rhetoric, visual evidence, artificially simulated data, etc., depending on field and venue. Verifiability in the traditional scientific sense of experimental repeatibility is actually very rarely a consideration in science (and in fact, many philosophers and historians of science have argued that repeating experiments is rarely possible and almost never desirable… the questions instead are, do the results accord with the results of related experiments?, can we build on these results?, etc.)

Science, as scientists are increasingly willing to admit in recent decades, is about what is verifiable rather than true in a similar sense to WP:V, since experimental science is increasingly conducted in largely artificial physical contexts. What happens in the lab is hoped to be a faithful reflection of what happens in nature, but the whole point of the lab is to isolate certain parts of nature so that they can be studied without all the complicating factors…and sometimes those complicating factors mean that a given experimental result may actually only be “true” for the very peculiar and artificial set of circumstances tested. The analogous situation on Wikipedia is when a seemingly reliable source is wrong; all the Wikipedian can do, without other sources to compare it to, is either limit claims to “source X says Y” (instead of just claiming Y and citing X) or ignore the source altogether. On Wikipedia we also hope that what the sources says accords with reality, but (for sociological rather than technical reasons) editors can’t go out and probe reality in its full complexity and must stick within the (negotiable) norms (which, like in science, are tailored to try to maximize the chances of accord between evidence and reality).

WP:V, and Wikipedia’s approach to sourcing and evidence more broadly, is just a different set of evidentiary norms, suited to a different group of people with a different purpose.