It is good to be polite, bad to be impolite. But politeness has its costs. Among these are epistemic costs. We do not claim that these costs should prompt us to give up our commitment to politeness; only to recognize the cost, in the spirit of an honest accounting.

One is polite when one is courteous and civil towards others, as exhibited in one's speech and behaviour. We might speak of politeness norms as norms that mandate the forms of courteousness and civility expected in one's community. We can distinguish at least two different dimensions of such norms: the standards they impose (i.e. what they regard as polite and impolite), and the weight assigned to following the standards. No doubt, politeness norms differ along these dimensions from community to community. Politeness standards clearly differ from community to community: in some communities, belching after a meal is considered impolite, in others it is not; in some communities standing within a foot of another person is considered impolite, in others it is not. So too for the weight that is attached to being polite. Some communities regard politeness as of fundamental social value: in such communities, from the fact that a candidate act violates one of the community's politeness norms, it follows that one has a very weighty reason not to perform the act. We can even imagine communities in which this reason overrides any other reason a subject might have to perform the act in question. Of course, not all communities are like this. Thus, there are communities in which the weight assigned to one's reason not to violate a prevailing politeness norm, while not insignificant, is nevertheless such that it often happens that other reasons (in favour of performing the candidate impolite behaviour) are weightier. (New York City comes to mind here.)

A special case of politeness involves how one responds to another person's claims or assertions. What we will call a norm of extreme conversational politeness will mandate a standard whereby it is regarded as impolite to express any public dissent or doubt about another person's claim. If a community places an extremely high importance on matters of politeness, and if the norm of extreme conversational politeness prevails, people have an extremely weighty reason not to express dissent or doubt publicly.

There is one obvious epistemic cost of politeness in such communities: when a person makes a claim regarding which many in the audience harbour warranted doubts, those who are unaware of the grounds for doubt lose out on the opportunity to become apprised of these grounds. We suppose that everyone can acknowledge this ‘epistemic cost’ of politeness.

We are interested in arguing for a more subtle epistemic cost of politeness. In particular, we submit that in such communities the opportunity to use audience silence as a signal of the acceptability of a publicly-made assertion is lost. To say that this opportunity is lost is to suggest that the prospect of having that opportunity is one that might reasonably be hoped for. This needs defending.

Let us say that the epistemic value of a signal s of a condition c is a function of how reliably s indicates the presence of c: the more reliable s is – the more probable c is given s – the higher s’s epistemic value as a signal of c, with the result that the higher the degree of warranted confidence a subject can have in the presence of c, given s. Now we want to say that there are communities in which audience silence has a high epistemic value as a signal of audience assent. These will be communities in which (i) there is a general expectation that if one has objections to an assertion made in one's mutual presence, one will indicate this, and (ii) people regularly conform to this expectation. In such communities, when the audience is silent in the face of an assertion made in their mutual presence, this is a (high-value) signal that they assent to it. Of course, the fact that people assent to another's assertion might be regarded as a fact of little interest in its own right. However, under two further assumptions this fact becomes very interesting from an epistemic point of view. Suppose that in addition to (i) and (ii) above, it is also true that (iii) people in the community are very well-informed on many topics that are discussed, so that were there grounds for doubt (regarding something asserted on one of those topics) they would probably be aware of these grounds, and (iv) people are also very well-informed about when a topic is such that others among them are well-informed on that topic. Under such conditions, the fact that an audience is silent, when the observer has grounds for thinking that the topic is one on which they can be expected to be well-informed, is itself evidence of the acceptability of the assertion itself. In such communities, people are well-placed to discern when silence in the face of a publicly made assertion indicates (not merely assent, but) acceptability.

Now insofar as there are communities in which (i)–(iv) hold, there is the prospect that, for at least a broad (and discriminable) range of topics, members of the language community can regard audience silence as an indication of (audience assent, and so) the acceptability of publicly made assertions on those topics. This is a great epistemic boon; in effect, those who satisfy (iv) are in a position to exploit the knowledgeableness of their fellows in determining the acceptability of what is said in their mutual presence.

It is the opportunity for this sort of epistemic social work – of being in a position to exploit the knowledgeableness of one's fellows in determining the acceptability of what is said in our mutual presence – that is lost in communities in which politeness norms are both demanding in their standards and overriding in the weight that is assigned to following those standards. In such communities, silence does not signal assent, and so even if conditions (iii) and (iv) hold, silence gives no basis for reaching a determination of others’ sense of the acceptability of the assertion.

Of course, the community that places such weight on following stringent politeness norms is an extreme example. In saying that it is extreme, we do not mean that such communities do not exist; on the contrary, we suspect that there are many such communities. We mean only that these communities occupy an extreme along the two dimensions just described: the demandingness of the politeness standards (what is considered polite and what not); and the weight assigned to following them. We can imagine communities that exhibit intermediate values along one or both of these dimensions. For example, we can imagine a community in which the range of conversational behaviours that are considered impolite is more restricted: following these more ‘liberal’ politeness norms would not rule out publicly expressing disagreement with others, but instead would constrain when and how such expressions ought to be made (e.g. avoid disagreements on deeply contentious matters of politics and religion; when disagreeing do so with respect, and only after a charitable reconstruction of the other side's views, and do not persist more than one or two rounds – after that the matter should be dropped etc.). And we can imagine a community in which the weight assigned to following the prevailing politeness norms, though significant, does not ensure that one's reasons to follow them will always outweigh one's reasons to violate them.

This brings us to our main claim, which is that to whatever extent a given regime of politeness norms mandates silence, to just that extent it degrades the epistemic value of silence as a signal of assent. The extreme community described above is only a vivid illustration of this. In cases involving more liberal (permissive) norms, or where the weight assigned to following the prevailing norms is decreased (relative to potential reasons one might have for violating such norms), there will be fewer cases of politeness-induced silence. In such communities, silence may still have epistemic value as a signal of assent. But it should be clear that its epistemic value is degraded relative to the epistemic value silence has as a signal of assent in communities in which (i)–(iv) hold.

We make no claims about how this epistemic cost of politeness should be traded off against the value of living in a polite society. That is a difficult question in value theory that we will leave for others to resolve. We only ask that the accounting be an honest one, when it is done. (Perhaps New York City will not turn out so bad by comparison).