Against accusing people of motte and bailey

post by Kaj_Sotala · 2018-06-03T21:31:24.591Z · LW · GW · 14 comments

Motte-and-bailey is a pretty famous term in rationalist circles. To quote the canonical SSC post on the topic:

So the motte-and-bailey doctrine is when you make a bold, controversial statement. Then when somebody challenges you, you claim you were just making an obvious, uncontroversial statement, so you are clearly right and they are silly for challenging you. Then when the argument is over you go back to making the bold, controversial statement.

I think that this is a real thing that very definitely happens.

I also can't remember having been in a single discussion that would have been improved by someone using this concept, and several that were worsened by it.

What's wrong with pointing out a motte and bailey if you see one?

Well, consider this conversation, whose content is fictional, but whose structure is similar to several very frustrating conversations that I've had:

Person: LW rationality isn't worth looking at. It's involves basing all your thinking on explicit calculations, which never works.
Me: Actually, that's not the case! People in the rationalsphere have explicitly said that that you shouldn't do that, and that emotion and intuition are actually important inputs to your decision-making process as well. For instance, here's a link to Julia Galef on straw vulcans [LW · GW], and here's an article saying that you shouldn't always use probabilities [LW · GW], and...
Person: Right, but that's just the motte. What they actually recommend in practice is just doing an explicit calculation about everything.

I am now faced with an impossible situation. Given that this person has decided that all the explicit declarations of "use emotion and intuition too" are just the motte, there's practically nothing that I could do to change their mind. Any links to articles saying that will be dismissed as just being parts of the motte. I could try to challenge their claim by asking for links that would support their case - but no doubt that the person can dig up some links which on an uncharitable reading at least sound like the thing that the person is saying. And possibly the person's actual crux [LW · GW] is that they've actually run into some people who did misinterpret LW rationality in that way, and now this person has decided that those people were actually representative of all LW rationality.

Now, in this particular example, I might still be able to change the other person's mind: it's not like there are lots of self-described LW rationalists who are saying "down with intuition, yay formally calculating everything".

But suppose that we were discussing something of which there were both sensible and crazy interpretations - held by different people. So:

and maybe even:

Now we are at the worst possible situation. Suppose that I belong to group A, and want to defend my group against an accusation. I say that no, we don't believe in crazy claim B1, we actually consistently maintain claim A1, and always have. I have links to back this up.

The other person says that this is just a motte and bailey, digging up links of group C using A1 as a motte - and is entirely correct.

Maybe I could say "you are referring to people belonging to group C, I belong to group A, see, members of my group are better than that." But it's unlikely that everyone involved explicitly declares their allegiance to different named groups, and that all members of the explicitly named group behave the same.

And even if they did, maybe that was part of the motte-and-bailey as well. There could be one group whose mission it was to say outrageous things, and another group whose mission was to say the sensible thing, in a way where the sensible thing still ended up supporting the outrageous version but established plausible deniability. This kind of thing actually happens: e.g. the youth organizations of Finnish political parties often make more radical claims than the parent parties would. To some extent this is likely deliberate and to some extent it's likely just younger people being more radical - and it being impossible to say the extent to which it's deliberate, is of course exactly what you want for the sake of having plausible deniability.

Scott's original post actually noted that motte-and-bailey accusations might derail a conversation, and had a recommendation for how to deal with it:

This means people who know both terms are at constant risk of arguments of the form “You’re weak-manning me!” “No, you’re motte-and-baileying me!“.
Suppose we’re debating feminism, and I defend it by saying it really is important that women are people, and you attack it by saying that it’s not true that all men are terrible. Then I can accuse you of making life easy for yourself by attacking the weakest statement anyone vaguely associated with feminism has ever pushed. And you can accuse me if making life too easy for myself by defending the most uncontroversially obvious statement I can get away with.
So what is the real feminism we should be debating? Why would you even ask that question? What is this, some kind of dumb high school debate club? Who the heck thinks it would be a good idea to say “Here’s a vague poorly-defined concept that mind-kills everyone who touches it – quick, should you associate it with positive affect or negative affect?!”
Taboo your words, then replace the symbol with the substance. If you have an actual thing you’re trying to debate, then it should be obvious when somebody’s changing the topic. If working out who’s using motte-and-bailey (or weak man) is remotely difficult, it means your discussion went wrong several steps earlier and you probably have no idea what you’re even arguing about.

This is often good advice. But as Scott himself has also discussed, sometimes you actually need to defend a group against an accusation:

I always thought that having things like political parties was stupid. Instead of identifying as a liberal and getting upset when someone insulted liberals or happy when someone praised liberals, I should say “These are my beliefs. There are other people who believe approximately the same thing, but the differences are sufficient that I just want to be judged on my own individual beliefs alone.”
The problem is, that doesn’t work. It’s not my decision whether or not I get to identify with other liberals or not. If other people think of me as a liberal, then anything other liberals do is going to reflect, positively or negatively, on me. And I’m going to have to join in the fight to keep liberals from being completely discredited, or else the fact that I didn’t share any of the opinions they were discredited for isn’t going to save me. I will be Worst Argument In The World-ed and swiftly dispatched.

In terms of the example which started this article, many of us probably think that LW rationality is important and valuable and want to encourage other people to take a look at it. So if someone shows up to say "LW rationality isn't worth looking at, they might say the occasional reasonable-sounding thing but that's just the motte, the bailey is something quite stupid", well, I'll probably want to say something in response.

I don't know what the right approach here is, given that sometimes you do need to talk about the opinions of groups, and sometimes groups do actually engage in motte-and-bailey. I don't want to go as far as to say, never accuse people of it. But I do suggest that you should be very cautious about using that accusation, because you may be placing yourself in a situation where you will never change your mind.


Comments sorted by top scores.

comment by Jiro · 2018-06-04T01:46:28.084Z · LW(p) · GW(p)

If different people in the group make sensible and crazy interpretations, and you're arguing with someone who claims to be making only the sensible interpretation, I'd expect that that person would at least be willing to

1) admit that other members of the group are saying things that are crazy. They don't have to preemptively say it ahead of time, but they could at least say it when they are challenged on it.

2) treat known crazy-talking people as crazy-talking people, rather than glossing over their craziness in the interests of group solidarity.

I'm also very suspicious when the person with the reasonable interpretation benefits too much from the existence of (and the failure to challenge) the person with the crazy interpretation. His refusal to condemn the other guy then looks suspicious. The term for this is "good cop, bad cop", and the fact that we have already have a term for it should hint that it actually happens.

And finally, sometimes as a practical matter, it's necessary to go against the bad cops. If the motte is some kind of reasonable objection to James Damore, and the bailey is "Damore said (list of things he didn't actually say)" and the bailey is all over the media and Internet and is used to attack engineers, that bailey is the one to be concerned about and the one to focus most of my effort against. It's not just argument, it's argument in service of a goal, in this case, not to be stomped on by people using baileys.

Replies from: Viliam
comment by Viliam · 2018-06-05T22:58:49.867Z · LW(p) · GW(p)
His refusal to condemn the other guy then looks suspicious.

This. I can't stop someone from believing that "LW believes [a stupid thing]" even if I am totally convinced that LW does not believe any of that, but I can make a clear statement that "I do not believe [the stupid thing]". And, who knows, if many people on LW make the same statement, maybe the person will update on the opinion that "LW believes [the stupid thing]".

(Or we get the frustrating outcome when the person says "well, Viliam, good for you to not believe [the stupid thing] but I am still convinced that LW believes that", in which case... well, at least I tried.)

But if someone refuses to take this step, and their approach is "well, I do not agree with X" in private, but at the same time they take great care to be never seen saying "X is wrong" publicly, then I believe I have a good reason to suspect them of defending one version of the belief publicly, and only falling back to another version privately when caught. ("Hey, my group totally does not believe X! That's just a lie spread by our enemies!" "So, would you be willing to go to your group's communication channel and post 'by the way, X is false, obviously'?" "No, why would I do that?" And the next day someone else from the group posts "X" and this person will share the post on social networks.)

comment by qualiyah · 2018-06-04T00:57:43.866Z · LW(p) · GW(p)

I found this illuminating, and agree that the use of the motte-and-bailey concept you describe isn't fair or helpful. But I'm not convinced the real problem here is with the appeal to the motte-and-bailey fallacy.

I find the motte-and-bailey concept quite helpful, but you can only legitimately pick out a motte-and-bailey if the self-same person is guilty of shifting between the two within the same work (or within multiple works intended to be consistent with one another). It's extremely common for individual people to personally motte-and-bailey within the space of a conversation (or 5-page paper), and having a concept to describe that is quite helpful.

So it seems to me that the real problem is not with the concept of motte-and-bailey arguments at all. All the actual problems are located around the tendency to hold every individual who calls themselves an X accountable for (simultaneously) all the opinions ever propounded under the label of X. The tendency to charge adherents of X with a motte-and-bailey fallacy even if they've not personally committed one yet is just one of many instances of this general tendency. The concept of a motte-and-bailey fallacy is no more at fault than the concept of an inconsistency is.

Replies from: Kaj_Sotala, strangepoop, RobbBB
comment by Kaj_Sotala · 2018-06-04T06:35:54.375Z · LW(p) · GW(p)

I think that I endorse this.

comment by a gently pricked vein (strangepoop) · 2018-06-05T17:06:46.465Z · LW(p) · GW(p)

All the actual problems are located around the tendency to hold every individual who calls themselves an X accountable for (simultaneously) all the opinions ever propounded under the label of X.


Responding to reductions is like responding to insults: if they don't spring out of genuine confusion or you've run out of your good deeds for the day, you don't have to respond.

I mean, if a surefire way to get you to give me information is to hold your reputation hostage, then you're going to be spending an awful lot of time fielding queries from strangers.

comment by Rob Bensinger (RobbBB) · 2018-06-04T18:30:47.017Z · LW(p) · GW(p)

I agree that motte-and-bailey would be a much better meme in the world where it were applied to individuals rather than groups 90+% of the time. I think it would still be a pretty bad meme in those worlds, for reasons related to Hazard's comment [LW(p) · GW(p)] and my pro-hypocrisy stance. I also think it's plausibly a lot harder to stop people from applying "motte and bailey" to groups than to discourage the "motte and bailey" framing altogether.

comment by musicmage4114 · 2018-06-07T02:04:40.329Z · LW(p) · GW(p)

I think this post beautifully (though indirectly) illustrates a few things:

  • the importance of debating only those "in front of you"
  • the importance of separating personal ideas from group ideas
  • the relative futility of debating against or on behalf of a group

In the first example you give, the error in pointing out a motte-and-bailey is that a motte-and-bailey hasn't actually occurred in the context of the conversation. If we bring up the bailey before it has actually been presented as an argument, we imply that we expect the other person to use it, and by extension, that they will argue in bad faith. The appropriate defense then (assuming you're telling the truth) is: "There may be some people who make that argument, but I haven't and I'm not going to, because I don't agree with it. Let's focus on the arguments each of us is actually making, okay?"

Similarly, if we make the argument "You belong to this group, and members of this group often use that statement as a motte, but they assert corresponding bailey X when they think they can get away with it", then we are assuming that the other person's beliefs are perfectly in line with our perception of their group's beliefs. This is also incorrect, and on two levels this time: we don't know that this person believes everything the group believes unless they say so, and we don't know if we have an accurate understanding of what their group believes if we are not part of it (or sometimes even if we are!).

We can take this even farther! Suppose the other person belongs to Group X, whose leadership or designated representative (or group of representatives) maintains a publicly-available platform, manifesto, or statement of purpose that clearly lays out what the group as a whole believes. In that case, we might be able to say we understand what the group believes, but we still can't say for sure what the group member in front of us believes, because they still don't necessarily need to agree with everything the group says (though such incongruity can sometimes open up other avenues of debate). In both of the preceding examples, the appropriate defense would be: "I may be a member of Group X [and Group X might as a whole support argument Y], but there are places where Group X and I disagree, including argument Y. Let's focus on the arguments we each make personally, okay?"

It's also important not to get baited into arguing against something we don't personally believe. Unless we're acting in an official capacity as a member of Group X, I think it's totally okay to refuse to engage when someone brings up the group's platform by saying something like "As a member of Group X, I do have an understanding of why some members use Argument Y, but I don't agree with Argument Y, so I'm not going to defend it."

In other words, I think it's fine to point out a motte-and-bailey, but only when someone has actually used it. It seems like the real issue is people pointing it out when it hasn't happened, as Davide_Zagami has already pointed out.

comment by Hazard · 2018-06-04T13:14:39.696Z · LW(p) · GW(p)

Another thought on ways it can be unhelpful to cry motte and Bailey. I don't know what's the split between, "People who exhibit motte and bailey behavior and do so as a tactic", vs "those who accidentally exhibit motte and bailey behavior", but I find myself accidentally internally doing motte and baileys far more often than is comfortable. I'm getting better at noticing when there's the switch, when my brain starts using a seperate model, but it's a very faint feeling. If I someone was actually and accidentally doing a motte and bailey, they might really not notice the switch, and shouting "Aha, motte and Bailey!" feels like an easy way to get them on the defensive and not actually think about if they switched models.

(Example, I've been reading Consciousness Explained, and I'm onboard with the idea that there's no Cartesian theater, but I definitely haven't completely "worked it out of my system" and you could totally catch me doing a motte and bailey on that topic)

comment by Rob Bensinger (RobbBB) · 2018-06-04T00:04:22.427Z · LW(p) · GW(p)

I agree with this. (My view of 'motte and bailey': 1, 2, 3.)

comment by Dagon · 2018-06-04T18:59:46.514Z · LW(p) · GW(p)

My first thought was "what's with all these recent LW posts complaining about online discussion peeves outside of LW"? Toward the end, you tie it back to useful advice with

But I do suggest that you should be very cautious about using that accusation, because you may be placing yourself in a situation where you will never change your mind.

The accusation of "motte-and-bailey" got a lot more popular recently, but it's really no different from accusing (when done to me) or pointing out (when I do it to you) much older terms of X bias, bait-and-switch, deceptive-sales-tactics, etc.

In most cases, there are two different problems you're reacting to:

  • you're (trying to) lower my status by pointing out problems in a group I align with.
  • you're (intentionally or accidentally) being imprecise with your proposition.

The second should probably just get to double-crux as quickly as you can. The first should usually be addressed separately from the nominal debate (but it will be very ideosyncratic in exactly how).

comment by ChristianKl · 2018-06-04T20:55:06.667Z · LW(p) · GW(p)

When it comes to LW rationality, it's not easy to be clear about what it happens to be.

If you have people say "well, because X happens I updated my probability that my date is likely to show up from Y to Z" it's easy for an outsider to get the impression that there's overvaluing of numbers going on. Even if you then defend yourself by saying: "We have this Internal Double Crux things, that totally focuses on using intuition", that just gets weirder for the other person.

comment by Davide_Zagami · 2018-06-04T10:43:47.452Z · LW(p) · GW(p)
But suppose that we were discussing something of which there were both sensible and crazy interpretations - held by different people. So:
group A consistently makes and defends sensible claim A1
group B consistently makes and defends crazy claim B1
and maybe even:
group C consistently makes crazy claim B1, but when challenged on it, consistently retreats to defending A1

I may be missing something but it seems to me that:

  • if C is accused of motte-and-bailey fallacy there is no problem;
  • if B is accused of motte-and-bailey fallacy there is a problem because they never defended claim A1;
  • if A is accused of motte-and-bailey fallacy there is a problem because they never defended claim B1.

I hope I'm not being silly: would it be fair to say that you are pointing to the existence of the "accuse people who are not making a motte-and-bailey fallacy of making a motte-and-bailey fallacy" fallacy? Could we call it "straw-motte-and-bailey fallacy" or something?

Replies from: Kaj_Sotala
comment by Kaj_Sotala · 2018-06-04T11:14:14.644Z · LW(p) · GW(p)
would it be fair to say that you are pointing to the existence of the "accuse people who are not making a motte-and-bailey fallacy of making a motte-and-bailey fallacy" fallacy?

Yes, and to the fact that once such an accusation does get made, it can be basically impossible to disprove since it's very hard to show that groups A and B actually exist and this isn't just a ploy where everyone actually belongs to C.

Replies from: joachim-bartosik
comment by Joachim Bartosik (joachim-bartosik) · 2018-06-04T15:05:34.755Z · LW(p) · GW(p)

I don't think so(EDIT: I don't think that looks like a new, separate "accuse people who are not making a motte-and-bailey fallacy of making a motte-and-bailey fallacy", it looks like something else to me). I think the situation is that there is a label for group A+B+C. Someone doesn't care about content made by group A+B+C because they perceive it as having motte-bailey doctrine.

This sounds like a bucket error (where one should get more content from A and ignore B and C) but I think it's not feasible to make new bucket that would capture only interesting people.

I haven't been in conversation where there was a problem like that. But I guess I'd try pointing out that:

- There's some good content,

- There's some bad content,

- They can checkout (specific examples) of content known to be high quality

- If they're interested in getting all of the worthwhile content they'll have to filter. Same as everywhere else.

Does it sound like something that could work?