The response from various right-wing circles about the TCS brouhaha is either charmingly antique or extraordinarily naive. The position seems to be that we should ignore who’s paying the piper and just listen to the tune to see whether we like it. Arguments, they say, can be evaluated independently of the context they appear in. But this relies on views about the nature of testimony that don’t stand up to empirical or philosophical scrutiny. As Grice put it, communication requires cooperation, and since advertising masquerading as honest opinion is not particularly cooperative, it is unlikely to be communicative, but without successful communication there simply isn’t a presented argument to evaluate. (This is long, so I’ve moved most of it to the extended entry.)

Let’s make that all a bit less abstract.

The role of language is not, much as you might believe from the antics around here, to facilitate debate. Just exactly what it is is a matter of some controversy, but it’s a reasonable guess that it’s something much more cooperative than debating. Maybe the primary role of language is the expansion of common beliefs. Maybe it’s the bringing about of shared plans. Maybe it’s the coordnation of those plans. It certainly isn’t adversarial. This is not to say it can’t be used in debating, because of course it can. Plenty of things can be used for other than their basic role. The role of tree branches is not to be instruments for hitting baseballs, but they can be well adopted to that role, as long as one is sufficiently careful. If one is not careful you’ll end up with weak groundouts and broken bats. And language can be well adopted to debating, as long again as one is careful. If care is not taken, all parties to the conversation can be harmed.

When people are engaged in cooperative behaviour, various assumptions can be safely made about their behaviour, assumptions that are not obviously entailed by their actions. If A and B are cooperating on repairing a car, and B walks away from the car for no apparent reason, A should assume that B has a good reason, relative to the shared project, for walking away. Maybe she needs to get a tool, maybe she needs to consult a manual, maybe she is thirsty and needs a drink before she can effectively continue, or whatever. As long as A and B are cooperating, A won’t, and shouldn’t, question B’s motives at every stage. The comparison with a case where A and B are not cooperating, where B is only there under some kind of duress, or for some ulteriour reason, is quite different. If B is a new employee at A’s shop, A might be justified, indeed required, to inquire why B is walking away from the car. The general point is that when we know that the situation is a cooperative one, we can safely make assumptions about the behaviour of other participants, and these assumptions can make our interactions more pleasant and efficient.

The same kind of principle is true of language, with the added twist that it seems to be hard-wired into us to treat conversations as if they were cooperative enterprises. I can’t find the citations immediately (and I’m not sure the evidence is that compelling, but it’s interesting) but there have been experiments suggesting that when a subject hears a sentence, her default behaviour is to treat it as true.

From memory, the important experiment invovled experimenters read a series of not very plausible sentences to subjects while having the subjects perform various distracting tasks. Subjects were more likely to believe the sentences read out than subjects not so distracted. One might have had a model for the mind where we hear sentences and then evaluate/decide whether to incorporate them as beliefs. One interpretation of this data is that’s just mistaken – we hear things as true, and, if and when possible, decide to remove them from our beliefs. Note that this is perfectly rational behaviour if language is basically a mechanism for cooperative action, which fundamentally it is.

This kind of hard-wired practice gets incorporated over time into social norms. There’s a norm of conversation that one should only say what one knows. And this has many consequences.

One, if there are good reasons against a belief you have, even if you think they are outweighed, these might defeat that belief’s claim to be knowledge, so you should not express the belief without mentioning the reasons pointing in the other direction.

Two, you should only speak on areas in which you have some (relative) expertise. In a cooperative enterprise, there is a mutual deference to the experts on the given task at hand. We assume the same is true of language, so anyone who speaks presents themselves as an expert on the topic, one who knows what the hearers do not know.

But now it’s clear how debating scenarios, and especially advertising scenarios, can distort these well-grounded norms. Advertisers do not (unless mandated) tell you about the countervailing considerations. They may present themselves as experts, but this is misleading.Debaters may or may not be as bad, depending on how much they value winning as compared to the growth of knowledge. (And obviously we all favour winning a bit.)

The fundamental purpose of labelling advertising as advertising then is to point out to readers in advance that this is not a cooperative framework. In a non-cooperative setting you don’t assume that the other party is acting to further shared interest. You don’t assume that what the other party is doing is well motivated. You don’t assume that what the other party is doing would not be undermined by more facts that are in that party’s possession. You don’t even assume that the other party has a particular expertise – just a desire to see the debate move in a certain direction. It’s no surprise that we socially evolved rules like labelling advertising as such to deal with the problem of people exploiting conversational norms for partisan, rather than shared, interests. The problem with TCS and its ilk is that it aims to undermine the rules we developed for this problem. But those are good rules, for it is a legitimate problem and the rules are an efficient solution, so it is wrong to undermine them.

Let’s get even less abstract, though the air will still be fairly thin.

Any argument, and it is agreed all around that TCS commentators are attempting to put forward arguments, has premises and inferences. Unless we are antique foundationalists, the premises will not present themselves with the divine light of reason, and unless we are particularly strident deductivists, the cogency of the arguments may not be obvious. If we want conversation to go anywhere, we have to take at least some of the premises someone gives us on trust, and to some extent we have to take the cogency of the argument on trust either.

Now we are normally smart enough to know not to do this with advertisers. We don’t trust either the premises or the reasoning. That’s why smart advertisers no longer try and persuade us, they mostly try to entertain and leave a positive impression. And it’s why even smarter advertisers try to appear not to be adverstiers so as to activate conventions of trust.

There’s several things that can go wrong when these conventions are broken.

1. Some writers just make things up. Their premises are false, and they either know this or don’t care. They just hope they won’t get called on them.
2. Other writers use arguments whose lack of cogency can be detected with some effort, but they hope we won’t make the effort, and that we’ll trust them that their premises support their conclusions.
3. Other writers again use arguments that are clearly enough not cogent given all the data, but they hide the data that tells against their position, so their argument is apparently good. As long as we are using ampliative inference this is a live possibility.

All three of these things are very annoying to the reader, because they involve a breach of trust. In every case the reader may well prefer to have never had to bother removing the errant beliefs from her belief box. Another reason advertising is labelled as such is to warn the reader to be on the lookout for all of these types of writers, and to read no further if she does not care to have to check for these things.

For there are other kinds of writers out there. Writers who put forward arguments that do try and account for all the evidence, and who use arguments that really are cogent. In those cases it may well be worthwhile investing time and energy in reading their pieces, because the reader may learn something she could not learn without some effort. Even if the writer has a point of view, as long as she’s operating as part of a genuinely cooperative enterprise, the growth of knowledge enterprise at its broadest, then some degree of trust that the writer is not one of the three types alluded to above is in order.

Advertisers, or even people writing in publications that are little more than advertising, do not deserve that kind of trust. But without it there is little to argue about. I cannot evaluate their arguments, because for me it is an open possibility that they are suppressing evidence that tells strongly against their claims. So I’ll keep my reading to those places (unlike TCS) where I can engage with people with different views in good faith.