What philosophical problems are there with obtaining knowledge by picking the correct or closest answer from a small set of solutions? And further, what are the problems/justifications for judging knowledge based on the democratization of questions and answers instituted by such methods as "up-ticking" a question or answer? Can these two systems, when fused into a Q&A site, serve to allow the user to find knowledge in a justifiable way? Or must each piece of knowledge -- no matter it's answer status or number of up-ticks -- be rationally considered separately in its own right?

5 Answers 5


Obviously, the way one approaches this question will vary depending on the epistemological tradition one is working within.

From the standpoint of the Nyāya school of classical Indian epistemology, there are exactly four epistemological warrants (Pramāņa -s):

  1. Direct Perception
  2. Inference
  3. Analogy
  4. Authority

Now, clearly Perception is not in play in this case, but the other three form the means of justification available to us when reading an answer.

Inference and Analogy are relatively straightforward: if the argument in the answer is logically cogent and sound, we can accept the answer, and we can similarly evaluate the answer by means of analogy to other knowledge we already have. Neither of these are specific in any way to StackExchange, of course-- they are the means we would use to evaluate any answer given to us by any third party.

What is somewhat different here is the means of evaluating authority. Although external credentials (such as graduate degrees or publications) are not made prominent here, the upvoting and reputation system give a means of evaluating the relative authoritativeness of the answer (and answerer) in the eyes of the community. And that community evaluation, of course, is in turn based upon the same list of Pramāņa -s as above.


Answers to questions on StackExchange sites seem to fall under the general rubric of "testimony" and thus inherit the problems and strengths of the literature on the epistemology of testimony.

When can you know something that is reported to you as opposed to something you perceived yourself? Hume's answer to this was something like "Well, we kind of know from past experience that people are typically trustworthy" (This from the guy famous for undermining induction! I know!) Here's the passage in the Enquiry:

The reason why we place any credit in witnesses and historians, is not derived from an connexion, which we perceive a priori, between testimony and reality, but because we are accustomed to find a conformity between them

There's something really classy about spelling "connexion" with an "x" like that. Damn.

The problem for this justification is, of course, that the conformity between "stuff I read on the internet" and reality is much less frequent and uniform than the conformity between reality and "stuff my friends report to me sincerely". So there are additional problems due to the nature of the medium.

Bayesianism is an approach to epistemology that has methods for dealing with problems like this: problems of gaining knowledge from faulty signals.

  • 1
    Maybe it's possible to argue that the Stackexchange sites are some sort of Bayesian calculator?
    – qgp07
    Commented Dec 20, 2011 at 23:22
  • No I think the Bayesian calculations are done by the agent having the beliefs. What the upvoting and reputation systems do for the site is attest to the trustworthiness of the signals (the answers). It is then up to the agent to work out how her beliefs ought to change in light of the signals.
    – Seamus
    Commented Dec 21, 2011 at 9:22
  • 1
    Since when did 'popularity' or the 'fifty-one-percent vote' determine truth? And, don't tell me: "Truth like justice is in the eye of the beholder." Or, "Whose 'truth' is it?" Or, "What truth?" Or, "What is what?" I'm not finding your site to be helpful at all. Stack Exchange answers are symptomatic. The 'problem' is the typical 'ideology' evidenced by both the 'self-educated' (which seems to be the overwhelming case with Philosophers Stack: i.e, too many "Johnny boys"), and public education alongside Marxian academia in today's post-modernist, politically correct Left-Wing America. Commented Dec 4, 2014 at 18:09
  • @DarcyDavis It's not about determining truth: it's about working out what are reliable indicators of truth. The basic idea is "if lots of people believe $x$ then $x$ is likely to be true", not "If lots of people believe $x$ then $x$ is true". And as I said, the indicators of truth in this sort of case are defeasible, equivocal and not on such solid ground as one might like. As to the second half of your comment, I have no idea what you're talking about.
    – Seamus
    Commented Dec 6, 2014 at 16:48
  • 1
    I think you do have a general idea of what "I'm talking about." Epistemic problems, justifications, knowledge. Stack Exchange is an interesting site; but the best place for information and "knowledge" is not the Internet (as an example, you should rarely use Wikipedia); a "better place" is your local Junior College liberal arts library. Commented Dec 6, 2014 at 20:33

I can think of one merit. For instance, when there some people who are living at geographically and historically speaking different places from the majority of people ( which is, in my assumption, the "Western People" ), like people from India, People from Russia, people from Africa, people from ME, people from Asia, for these who are in the majority group, I think these different people can serve well enough here since they can give the majority of people probably very near original source, which, I assume personally would be a language barrier. If you would like to know something peculiar about Buddhism, oh yes, the best choice would be to ask people from India, since after all, Buddhism was born and has been "advanced by" for a long time there ( but actually not at today's India but in terms of cultural influence ). If you would like to know Chinese thoughts, the best way is ask Chinese without doubt ( is there anybody here...? ).

One "problem" or potential demerit is that if you are not interested in the questions and answers, then it would not be helpful to you and you should go elsewhere.

If my answer does help in any kind of ways, I will feel happy.

  • Kindly allow me to give thanks to the person who edited.
    – user13955
    Commented Jul 6, 2015 at 11:30

To consider this objectively one must keep in mind when seeing the result of votes, on how they are created. Rationally one must consider that the votes on this site came from its rules, that is, because of them to get votes you have to get reputation and objectively it is just reputation. The system provides an objective valuation as far as the members are being objective. If the majority of the site are objective, the numbers are objective within their own definitions based on their causes: the rules.


There are debates in social epistemology as to the epistemic value of going with the numbers. Some hold that weight of opinion is only good justification if each individual comes to a conclusion independently and if they are sufficiently expert in the topic rather than being "nondiscriminating reflectors". But on the other hand, it can be proposed that some epistemic weight might be given to a nondiscriminating reflector Y's agreement with X (the expert) if they have some ability to identify experts (meta-expertise):

The novice's confidence in X's expertise concerning hypothesis H is rationally increased by his or her confidence in Y's meta-expertise

(Coady, 2012). On SE, there are clearly voters with varying levels of expertise - though it might be argued that all members of the community have at least the ability to recognise attributes of quality answers.

My own work (Matthews, 2014) suggests that those with higher levels of knowledge, if voting blindly, will tend to agree with the high ranked answers. Those with lower levels of knowledge will tend to pick more randomly when the voting cues are not present. This implies that the system on the whole works well as long as there is sufficient expertise among the voters.

We should also be aware that there are some sources of bias in SE sites that can mean that the highest voted answer is not the best. There is the well-known first mover advantage where the earlier answers gains votes and hence reputation for the answerer (the "fastest gun" effect, see Mamykina et al, 2011). This implies that it's always worth checking newer answers as well as highly-voted ones.

Coady, D. 2012. What to Believe Now: Applying Epistemology to Contemporary Issues http://eu.wiley.com/WileyCDA/WileyTitle/productCd-EHEP002800.html

Mamykina et al 2011. Design Lessons from the Fastest Q&A Site in the West http://www.cs.berkeley.edu/~bjoern/papers/mamykina-stackoverflow-chi2011.pdf

Matthews, P. 2014. Use of credibility heuristics in a social question-answering service http://www.informationr.net/ir/20-1/isic2/isic27.html#.VZpVIUaHhVU

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .