Can We Trust Wikipedia?

Wikipedia is increasingly becoming the first and often only source to which many people refer for getting information on many subjects. It has several features that distinguish it from traditional sources of knowledge: It is democratic, collaborative, and constantly changing.

But can we trust it? The aim of this post is to review some of the arguments for and against trusting Wikipedia, and arrive at some tentative conclusions.

The Case against Trusting Wikipedia

Wray (2009) makes a comparison between the epistemic culture of science and Wikipedia, from which he draws the general conclusion that Wikipedia cannot be trusted. Wray makes two central claims:

  1. Unlike scientists, Wikipedians are not experts in the fields on which they write, hence they lack the necessary epistemic competence to produce reliable knowledge.
  2. Unlike scientists, Wikipedians are anonymous and do not care about their reputation and credibility, which are the forces that drive the “hidden hand” of credibility in science.

There are, however, several difficulties with Wray’s claims. Wray does not distinguish between different forms of expertise. While most Wikipedians may not have contributory expertise, i.e. the expertise needed to successfully participate in scientific research, many Wikipedians, e.g. doctors and engineers, may have interactional expertise, i.e. they can read and understand scholarly sources in their professional fields. Some may also have experience-based expertise, i.e. non-credited knowledge of special domains (Collins & Evans 2007).

Anonymity has its epistemic advantages. It mitigates the negative effects of unjustified social stereotypes in credibility assessments (Fricker 2007), and enhances critical outlook (Rains 2007). A significant number of Wikipedians in fact do reveal their material identity online. They tend to do that more than users of other online forums (Amichai–Hamburger et al 2008).

Wray does not distinguish between identity and identifiably (McLeod & Leshed 2011). While many Wikipedians do not disclose information that may identify their material identity, they reveal other components of their identity, which are relevant to their epistemic assessment (Amichai–Hamburger et al 2008).  In particular, many Wikipedians include links to articles they have previously worked on, allowing other users to learn about each other’s interests and levels of expertise (Kuznetsov 2006).

Moreover, online users tend to construct virtual profiles that resemble and moderately improve on their material identity (Messinger et al 2008). They tend to be engaged in maintaining their virtual identity’s reputation and care about it as well as that of their material identity (McLeod & Leshed 2011). And Wikipedians report recognition as one of their motivations, and Wikipedia has an intricate system of bestowing recognition, honours and rewards, such as “Featured Articles” (Kuznetsov 2006).

It is also worth mentioning that the existence of a “hidden hand” of credibility in science is disputed (Solomon 2001).

The Case for Trusting Wikipedia

Drawing on considerations from the epistemology of testimony, Tollefsen (2009) argues that there are good reasons to trust Wikipedia, and they are improving as Wikipedia is slowly maturing.

She argues that Wikipedia can be characterized as a mix of Wikipedians’ individual testimonies, and the group testimonies of Wikipedia as a collective body. She further argues that Wikipedia has institutionalized mechanisms of critical discussion and quality control, by which it evaluates Wikipedians’ individual contributions (testimonies).

As a consumer of group testimony, a Wikipedia reader may trust, to some extent, these mechanisms. He can also critically evaluate the content of the articles in light of his previous background beliefs.  By doing so, so Tollefsen argues, the reader may satisfy the conditions for forming warranted testimonial beliefs on a reductionist account of testimonial justification. (If you are not familiar with these terms, the claim is roughly that we have a right to trust Wikipedia, which is similar in strength and scope to the right that we have in ordinary circumstances to trust the say-so of other people, such as teachers, doctors or random strangers we ask for directions.)

As I argue in the next section, however, Tollefsen’s account overlooks systematic features of Wikipedia‘s epistemic design, which are endemic to it.

Reasons for Concern

Both accounts have misidentified the salient reasons for concern. I identity four such reasons:

  1. Inaccessible Academic Content – Most quality academic content is found either on closed online databases, such as JSTOR, which require subscription, or in academic libraries, which are largely inaccessible to the general public.
  2. Long Testimonial Chains – In the absence of direct accessibility to primary research, Wikipedians rely on second- and third-hand sources. Much of these sources are simplified and popularized accounts of scientific research, which tend to distort it (Miller 2009).
  3. Lack of Tempered Equality of Intellectual Authority – Longino (2002) requires that in a knowledge-producing epistemic community, intellectual capacity and relevant expertise be the only criteria by which people are given the right to participate in the collective discussion. Due to Wikipedia‘s democratic nature, this is not the case there. Lay users have an equal say as experts and they often dispute experts’ claims, wear them out in tedious discussions, and eventually drive them away.
  4. Insufficient Social Diversity – Social diversity has a significant epistemic role in facilitating critical discussion in which different perspectives are represented. The Wikipedians’ population sometimes seems too homogeneous, consisting mostly of exactly those like-minded of people who find intellectual online activity to be satisfying.

Tentative Conclusions

There are good reasons to trust Wikipedia articles on issues such as popular culture or current events, on which there are amply available sources and many people with relevant expertise.

There are also moderately good reasons to trust Wikipedia articles in areas such as medicine, engineering, and undergraduate-level science, in which there is reason to suppose there is a sufficient number of Wikipedians with the required interactional expertise.

There are few good reasons to trust Wikipedia articles on issues that require specialized expertise and on which there are few available reliable sources, such as advanced science, history of science, current philosophy and the like. There are reasons to suspect this will not improve with time.


Amichai–Hamburger, Yair, Naama Lamdan, Rinat Madiel & Tsahi Hayat. 2008. Personality Characteristics of Wikipedia Members. CyberPsychology & Behavior 11(6): 679-681.

Collins, Harry M. & Robert Evans. 2007. Rethinking Expertise. Chicago: University of Chicago Press.

Fricker, Miranda. 2007. Epistemic Injustice: Power and the Ethics of Knowing. Oxford: Oxford University Press.

Kuznetsov, Stacey. 2006. Motivations of Contributors to Wikipedia. ACM SIGCAS Computers and Society Archive 36(2).

Longino, Helen. 2002. The Fate of Knowledge. Princeton: Princeton University Press.

McLeod, Poppy L. & Gilly Leshed. Forthcoming, 2011. As Long as They Don’t Know Where I Live: Information Disclosure Strategies for Managing Identity in Second Life. In Reinventing Ourselves: Contemporary Concepts of Identity in Virtual Worlds, eds. Peachy & M. Childs. New York: Springer.

Messinger, Paul R., Xin Ge, Eleni Stroulia, Kelly Lyons, Kristen Smirnov, & Michael Bone. 2008. On the Relationship between My Avatar and Myself. Journal of Virtual Worlds Research 1(1): 1-17.

Miller, Boaz. 2009. What does It Mean that PRIMES is in P? Popularization and Distortion RevisitedSocial Studies of Science 39(2): 257-288

Rains, Stephen A. 2007. The Impact of Anonymity on Perceptions of Source Credibility and Influence in Computer-Mediated Group Communication: A Test of Two Competing Hypotheses.  Communication Research 34: 100-125.

Solomon, Miriam. 2001. Social Empiricism. Cambridge, MA: MIT Press.

Tollefsen, Deborah P. 2009. Wikipedia and the Epistemology of Testimony. Episteme: A Journal of Social Epistemology 6: 8‑24.

Wray, K. Brad. 2009. The Epistemic Cultures of Science and Wikipedia: A Comparison. Episteme: A Journal of Social Epistemology 6: 38-51.


Boaz Miller
Boaz Miller is a postdoctoral fellow at the Edmond J. Safra Center for Ethics, Tel Aviv University. He has a PhD and MA from the Institute for the History and Philosophy of Science and Technology (IHPST) at the University of Toronto. His areas of specialization are philosophy of science and social epistemology. He works in the intersection of philosophy of science, analytic epistemology and science and technology studies. He studies scientific expertise, the relations between knowledge and consensus, and the relations between social values and evidence. He has a BSc in computer science and "Amirim" Interdisciplinary Honors Program from the Hebrew University of Jerusalem.


  • P.D. Magnus Reply

    I think that Can we trust Wikipedia? is the wrong question. For one thing, we all do use it. Pretending that the question is whether or not to use it at all is disingenuous.

    For another thing, Wikipedia articles are highly variable in quality. Different communities of users contribute in different topic areas. Topic that have fan communities (like comic books) often have thorough entries. I also understand that the articles in physics are pretty good; I have students who were pointed to Wikipedia by their physics prof. There are other topics, however, where there is not a core community of careful editors and so the articles are shoddy.

    The right question to ask is how to use Wikipedia. Even if it is in some sense like group testimony, using it isn’t like evaluating more familiar kinds of testimony. (Rather than fill up the comment box, I’ll refer you to my paper in the same issue of Episteme.)

    • Boaz Miller
      Boaz Miller Reply

      I agree. Articles on Wikipedia deserve differential trust. My aim was to try and come up with principled ways to distinguish between different articles based on how much trust they deserve.

    • Mike Thicke
      Mike Thicke Reply

      Actually, I think “the” right (more interesting) question is: What is different about knowledge produced by Wikipedia than knowledge produced through other means? Brad Wray is right that the community structure and norms of Wikipedia are different than those of science, but it doesn’t follow that this means Wikipedia doesn’t produce reliable knowledge unless science (in its current form) is the only way of producing reliable knowledge. I think the community structure of Wikipedia is very interesting, and is worth studying apart from the question of whether it is trustworthy.

    • W. Dean Reply

      “How can we use Wikipedia?” isn’t the right question either, unless it’s a rhetorical question and the answer is that we can’t. The only reason to use Wikipedia (or any other encyclopedia) is as an authoritative source. But we can’t cite Wikipedia as an authority because its authorities are anonymous. That’s the bottom line. So it hardly matters, from a practical standpoint, whether this or that article is reliable or plausible or un-shoddy (even if these questions provide an interesting case for studying expertise and authority).

      I don’t mean to suggest that I don’t read Wikipedia articles, because I do. But it seems perfectly obvious to me that these articles carry the same or less weight than the opinions of an acquaintance who happens to have a degree in x, work in field x or read a lot about x (etc.). And I wouldn’t cite such knowledge as authoritative except in extremely unusual circumstances.

      • Mike Thicke
        Mike Thicke Reply

        But not even “we” when “we” means “academics” only value authoritative or citable sources. Wikipedia is, as you imply, useful for getting an introduction to a subject, but I think we can also gain knowledge from Wikipedia—that is, beliefs gained through reading Wikipedia can in many circumstances be considered justified. Not justified to the level that you would want to put such a belief in print without some other verification, but justified enough for most cases.

        As Boaz says in the article, Wikipedia isn’t truly anonymous, and it does cite other, non-anonymous, sources. Even if it didn’t, I don’t think anonymity is a sufficient argument against Wikipedia giving us knowledge.

  • Mike Thicke
    Mike Thicke Reply

    Great article Boaz.

    I have some concerns with your reasons for concern, however. Regarding 1 & 4, my understanding is that a large portion of Wikipedia editors, particularly of those articles requiring “specialized expertise” that you trust the least, are graduate students. This means that many of Wikipedia’s editors do have access to academic content not accessible to the general public. And AFAIK, there is nothing in Wikipedia’s rules forbidding the citation of such content. It doesn’t seem necessary for everyone to have access to this material, since it only takes one informed editor to convey the relevant information.

    If Wikipedia editors’ demographics roughly match those of graduate students, then any problems of diversity for Wikipedia are probably going to apply to academia as well. Though this doesn’t mean it’s not a problem, it suggests it’s not a problem unique to Wikipedia.

    Regarding 3, I wonder at your inference from lay participation to there being criteria other than Longino’s in determining who contributes to Wikipedia articles. The democratic nature of Wikipedia means that the enforcement of those criteria will operate differently than in science, but not necessarily that those won’t still be the enforced criteria. As you pointed out in the beginning of your article, there are different kinds of expertise, and not everyone needs contributory expertise to have the proper expertise for contributing to Wikipedia. It seems quite possible that Wikipedians will, democratically, enforce just the standards that Longino recommends.

    • Boaz Miller
      Boaz Miller Reply

      The demographics of Wikipedians is an empirical question, and I hope that somebody has studied it. I’ll try to look for some studies when I write my paper to avoid mere speculations. There are no doubt graduate students working in Wikipedia, but my impression is that most contributors are amateurs. This may also vary with respect to different subject matters and Wikipedias in different languages. At least in philosophy, many articles are in bad shape, and I would be surprised (and disappointed) to learn that graduate students are responsible for it.
      With respect to accessibility to academic sources, just count how many references there are to the 1911 Encyclopedia Britannica, which is public domain, but hardly the best and most up to date source to cite from. If Wikipedians had access to better sources, they would cite them (at least I hope so).
      Longino’s tempered equality condition applies to any epistemic community, not just to science. There are difficulties with how to apply it to concrete real-life cases, but as an abstract principle it has a sound epistemic rationale – only relevant expertise should matter with respect to having the right to say-so in critical deliberation within the community. Wikipedia’s democratic nature, where each member has a vote regardless of relevant expertise clearly violates this principle.

  • Gilly Leshed Reply

    In a highly publicized Nature article, Wikipedia and Encyclopedia Britannica were found to be similar in their accuracy on 42 articles that were sent to experts for review (Giles, 2005). This was considered surprising to some, and natural to others, given the democratic nature of Wikipedia article writing. Note that this study was published only about 4 years after the launch of Wikipedia in 2001. What does it mean? That we can trust Wikipedia as much as we can trust Britannica? If trust is a direct product of quality, the answer may be yes. But people may have other reasons to trust one source of knowledge rather than the other, including access (discussed above), the type of knowledge they are looking for (e.g., see Diana Forsythe’s discussion of the kinds of knowledge represented and not represented in a medical information system for Migraine sufferers, 1996).

    Inclusiveness and “the wisdom of the crowds” may be a source of reliable knowledge in Wikipedia articles (in contrast to the argument above that there is insufficient social diversity): if someone makes a mistake in the article, there is a high chance that someone else will notice the error and fix it. Note that this does not come without cost: Wikipedia editors need to continually engage in coordination in order to ensure the quality of the articles they produce (Kittur & Kraut, 2008). Furthermore, there is strong emphasis by Wikipedia editors on following conventions, guidelines, and article planning, especially in the Talk pages (Viegas et al., 2007), which may be contributing to a higher quality of articles. Again, the question is whether high quality should immediately lead to trust.

    Forsythe, D. E. (1996). New Bottles, Old Wine: Hidden Cultural Assumptions in a Computerized Explanation System for Migraine Sufferers. Medical Anthropology Quarterly, Volume 10, Issue 4, pages 551–574, December 1996.

    Kittur, A., Kraut, R. E. (2008). Harnessing the Wisdom of Crowds in Wikipedia: Quality Through Coordination. CSCW 2008: Proceedings of the ACM Conference on Computer-Supported Cooperative Work. New York: ACM Press.

    Giles, J. (2005). Internet encyclopaedias go head to head. Nature 438, 900-901 (15 December 2005).

    Viegas, F. B., Wattenberg, M., Kriss, J., & van Ham, F. (2007). Talk Before You Type: Coordination in Wikipedia, 40th Annual Hawaii International Conference on System Sciences (HICSS ’07).

    • Mike Thicke
      Mike Thicke Reply

      Thanks Gilly! I think we need a “like” button on here.

    • Boaz Miller
      Boaz Miller Reply

      Hi Gilly. Really great comments! I will read the sources you list. I am sceptical about the wisdom-of-the-crowd-argument. There are general problems with it such as impassibility theorems (incoherent group sets of beliefs), herding, group polarization etc. But frankly, my best argument is the bell curve of grades that we get after we teach a course. After attending a course, most students get a mediocre understanding of the subject matter, which represent the middle of the curve. They usually tend to misunderstand the same issues, and with respect to these issues they are more in a position to enforce each other’s errors than to correct them. I’d rather that only those who at the good end of the curve wrote and edited the relevant articles in Wikipedia.
      My point was that I see no reason to expect wisdom from the crown on matters that require specialized expertise, where most members of the crowd do not have the necessary background knowledge or right intuitions, which are necessary to correct mistakes, and may even have misleading beliefs and intuitions.
      Take, for example the seemingly well written and well referenced article The Scientific Method. No historian of science would sign on the history there, which is terribly Whiggish. The theoretical part mixes together different contradictory theories of confirmation, it has disproportional representation of some views (with all due respect to Peirce, he is not that important) and lack of proper representation of key thinkers like Hempel and Lakatos. It has very little reference to the vast debates over the different models of the scientific method, and whether any one at all is true. It is obviously written by science enthusiastics who have at most superficial understanding of the subject matter. Compare that to the table of contents of a proper academic textbook. From my experience, such articles are common. For this reason I am not impressed by the Nature article.

      • Mike Thicke
        Mike Thicke Reply

        But the collective knowledge of a group != the knowledge of its median individual. Surowiecki’s The Wisdom of Crowds isn’t arguing that the average individual knows just as much as any expert, but that average individuals making collective decisions often perform better than experts.

        • Boaz Miller
          Boaz Miller Reply

          But what’s the argument for that? Is it based on some version of Condorcet’s theorem? I find good no reason to suppose that in most matters requiring specialized expertise most members of the crowd will have higher than .5 probability to reach the truth. If that were the case, we could decide almost any factual controversy on lay majority voting. In such matters, why should we expect a crown to do better than its median individual?

          • Mike Thicke
            Mike Thicke Reply

            The argument of the book is empirical: that in many situations crowds of non-experts actually do out-perform isolated or small groups of experts. The argument for why is something like mistakes cancel each other out and what is left over is true (this is obviously approaching caricature of his actual argument but…)

            • Boaz Miller
              Boaz Miller Reply

              I’m not one of those philosophers who say, “yeah, it works in practice, but does it work in theory?”. In the case of Wikipedia, there is great variability in quality, and quality is not correlated with how impressive or well referenced an article is, as in the case of The Scientific Method I gave as an example.

  • Eleanor Louson
    Ellie Louson Reply

    Great article Boaz!

    One topic you didn’t get into was vandalism. Certain topics on Wikipedia have been well-known targets of vandalism (see especially the history of Nickelback’s Wikipedia page *warning: contains mature language* but every once in a while I’ll find a random vandalized phrase on a more mundane page. I suspect that if Wikipedia’s editors are missing those obvious examples of vandalism, they may also be less obvious errors and biased perspectives.

Leave a Reply

Your email address will not be published. Required fields are marked *