Consider a statement of the form "based on knowledge and common sense and estimating the probabilities of alternative hypotheses, I believe that X". Or, perhaps, "based on Bayesian reasoning I have come to the conclusion X". How much, when an interlocutor of yours uses such a statement, should it affect your credence in X?

I claim that the answer is "negligibly", in most cases. (I discuss exceptions below.)

The most appropriate way of conducting an argument is in most cases to simply state your belief X. If you are going to adduce evidence and argument in support of your belief X, simply state these. If you've derived probability estimates, simply provide them, perhaps with the method of derivation. The "meta" observation by itself has no place in an argument.

(Isn't that obvious, you might ask? Not to everyone, as this statement form turns out to be actually used in discussions here. So, in the interest of contributing even a little to a possible theory of argumentation, I develop some further observations on this pattern.)

The assertion "my opinions on X are soundly arrived at" has an equivalent structure to saying "I am a truthful person": it cannot be verified by an interlocutor, other than (in the latter case) by looking at what statements you utter and independently verifying they are truthful, or (in the first) by looking at what arguments and evidence you adduce in support of your opinions on topic X, and independently assessing whether the evidence is credible and the arguments are individually sound.

Claiming to be a truthful person doesn't mean much. I know parents who are constantly telling their kids "oh, you shouldn't lie, it's bad", but who actually lie to their kids quite often. My own approach has been to tell my kids the truth (and to answer pretty much all their questions), while almost never mentioning "truth" as a moral topic. Empirically, I observe that my kids are growing to be truthful and trustworthy persons, more so than many of e.g. my neighbours' kids.

Claiming that you use sound methods in arriving at a conclusion of which you want to convince your interlocutor similarly has little content. It is the sort of thing that can only be assessed by looking at your justifications for belief. The claim may (if stated in an authoritative tone) actually manage to nudge your interlocutor a little in the direction of accepting your conclusions. On reflection, it shouldn't, and your argumentative skills would benefit from refusing to accept such statements as justification.

Rather than say "there is evidence", point to evidence. Rather than say, "it can be argued that", simply argue it. If there are allowable exceptions, they concern cases where you say something in addition to the bare assertion of soundness: for instance if you say "there is lots of evidence, readily obtainable from everyday sources". Your claim is above and beyond a claim of soundness.

You should be prepared, if challenged, to back up your observation about the abundance and availability of the evidence. If you say of something that it's "obvious", it had better be obvious.

Sometimes too, the use of a given methodology to arrive at a belief X is surprising information in and of itself. It makes a difference to an interlocutor to know that your belief has that origin. It is hard to think of examples where it could possibly make a difference to say that in very vague and general terms, but I can think of cases where the use of a specific method is new and surprising. For instance, if you say "I believe in X and Bayesian reasoning (as opposed to other methodologies) demonstrates X in a particularly convincing manner". In this case the claim is not only about X, it is also a reasonably interesting claim about the method itself. (Which claim might or might not stand up to examination -  cf. the ongoing argument about frequentism vs. bayesianism.)

These exceptions aside, sound argumentation shares one precept with fiction: show, don't tell.

New Comment
25 comments, sorted by Click to highlight new comments since:

- tl;dr: It takes too long

All communication takes time. We should minimize the time necessary to communicate our arguments. As a speaker (or writer for that matter), I cannot know which parts of my argument will be obvious to the listener and which ones won't.

"It can be argued"/"There is evidence for" should be used whenever the speaker assumes further detailed arguments are unnecessary but would be able to supply further details if requested without undue delays. By stochastically testing whether the speaker actually can supply correct arguments/evidence if requested, we can quickly build trust - saving us a lot of time during later communication with said speaker.

In this way, these phrases could not be used for bluffing disagreeing listeners, because they would simple request more details.

At least in philosophy (my discipline), people often say 'it can be argued that p' when arguments for p have been produced by others, and these arguments are thought to have considerable force, but it would take the discussion too far astray to rehearse those arguments. It is sometimes worthwhile to say this, rather than producing an argument for p.

This isn't just for people who are already familiar with those arguments for p. Knowing these arguments exist is often importance evidence related to p, and can guide one's future thinking.

Interesting piece.

I agree with Drahflow and utilitymonster here though. An argument needs to be made in the context of the audience. Unnecessary details about an argument may dilute the effect of your speech. And stating the obvious (which it may be to the audience) makes one look like an arrogant guy, who is assuming that the audience wouldn't know.

Yet, I agree with you in the example you give about being truthful to kids. Making an argument based on truth and stating that truth may be a good way of dealing with kids. And as you claim it is showing in their development.

Consider a statement of the form "based on knowledge and common sense and estimating the probabilities of alternative hypotheses, I believe that X". ... The "meta" observation by itself has no place in an argument.

That's only true if you're observing Crocker's Rules.

If nothing else, saying "I think that" at the start of a sentence reminds the writer not to assign probability 1 to it.

Brief comment, elaborating on the point of the post: one factor in how reliable a signal is comes from how hard it is to fake the signal. The meta-declaration is a very weak signal - anyone can say the words "I examined large amounts of evidence and deduced that the many-worlds interpretation of quantum mechanics is far superior to the collapse theories", and equally anyone can say "I examined large amounts of evidence and deduced that the many-worlds interpretation of quantum mechanics is far inferior to the collapse theories". What is expensive to fake, and therefore good evidence of sincerity, is to write a series of essays discussing the relevant evidence and key elements of the analysis leading to the particular conclusion. Only in the latter case can third parties can check your evidence and your reasoning, and anything they discover about these screens off everything anyone can say about your trustworthiness.

(Isn't that obvious, you might ask? Not to everyone, as this statement form turns out to be actually used in discussions here.

In LW's collective defence, that comment was heavily downvoted.


Not as heavily as it's going to be now that it is linked to in a top level post! :-)

Ouch, yes!

New game: Karma sniping.

there is lots of evidence, readily obtainable from everyday sources

I think someone called this the appeal to Google

I agree that a claim of sound reasoning methodology is easy to fake, and the writer could easily be mistaken. So it's very weak evidence. However, it's not no evidence, because if the writer would have said "my belief in X is based on faith" that would probably decrease your trust in his conclusions compared to those of someone who didn't make any claims about their methods.

A promise has value when its promiser has a record that warrants trust. Otherwise, I agree. Show your work.

The most appropriate way of conducting an argument is in most cases to simply state your belief X...

By what criteria do you judge this to be the most "appropriate way"? I think that any such code of conduct should be subject to an expected utility maximisation framework, and it's not clear to me that this particular code stands up.


The "meta" observation by itself has no place in an argument.

(Isn't that obvious, you might ask? Not to everyone, as this statement form turns out to be actually used in discussions here

I think the key part of this statement is "by itself." Morendil, please acknowledge that elsewhere in that thread I supplied the observations underlying my conclusion. Then please apologize for implying that I did not do so.

So, wait, does that mean no one was persuaded by my comment here? So why did it get up to 8?

And how am I supposed to make people believe that stuff?

I just read through that thread. Did I ever mention that I love this Clippy character?


So are you going to give me extra metal and treat your existing paperclips respectfully?

And how can I make other humans love me?

I love you, Clippy, because your comments are brilliant and witty and make me laugh a lot. And I still want to know how you feel about melting down one clip and forming two from it.

I still want to know how you feel about melting down one clip and forming two from it.

I think that's GREAT, as long as the two resultant clips are sufficiently functional as paperclips (which means being able to hold together a certain number of sheets of paper that I'm not quite precise about yet). Otherwise, that's just wasteful, don't you think?

You're welcome to have all the metal in my cans of food and drink once I've emptied their contents. I don't think I have any paperclips, but if I end up with any, I'll do my best to treat them respectfully.

My reaction was caused by your all-around adorable behavior. I'm not sure of how well this reaction generalizes to other people, but it seems like a plausible assumption that others might also react the same way.

I parsed that comment as meaning that you see some advantage to having us believe that you have the traits in question - a fact that can be relatively safely assumed about humans, but that I hadn't been assuming about you. I therefore took the comment as evidence that you will probably want to signal those traits in the future, which would involve taking useful-to-me-or-others action in relevant situations, but only if the cost of doing so was low enough. I have very little useful evidence regarding how you calculate such costs and how you define 'low enough' - less than I have regarding a human stranger, even - so I'm still assuming that you can't be relied on to take such actions in non-trivial situations. (Yes, I'm privileging a hypothesis in making that assumption. Doing so in this case seems to be instrumentally rational: Being betrayed is dangerous.)

As to how to get people to believe you, the wikipedia article on signaling games seems to be a decent place to start looking to figure that out, though the section on 'costly signaling' doesn't cover that concept as well as might be ideal.

I like how you do a lot to help me understand humans. That's quite paperclippy of you!

Is there some reason I have caused you to want to do this?

(Tried to do this as a private message, but that keeps failing like extrusion of mercury.)

*chuckles* You shouldn't be assuming that I'm Paperclippy any more than I should be assuming that you're Friendly. The truth of the matter is that you're doing a good enough job of coming across as human - and particularly, as a rather autistic human, which is a subtype that I'm personally inclined to go out of my way to help - that I'm having a hard time not treating you as such. It's an alief/belief disagreement issue [pdf].

The practice at explaining stuff isn't a bad thing, either.

(Meta: Of course, if I wasn't certain that I was actually dealing with a human roleplaying as a paperclipper, this particular alief/belief issue would be being resolved in the other direction. I'll admit to being impulsive, but I do grok the implications of actual paperclippers. It's just not worth the effort of overruling the alief in this case.)

What's autistic about me, and why is that good?

The primary similarity is that autistic humans, while exhibiting the same range of intelligence as humans in general, tend to be much worse at understanding the subtleties of social interaction. Most humans would consider the type of information that I'm giving you to be very basic, or even instinctive, but it's not terribly uncommon for autistic humans to reach adulthood with little to no functional understanding of one or more subsets of that information. (From what I've read, and from my own experience, most autistics do seem to have instincts relating to socialization, but those instincts tend to be nonstandard enough to be useless or even counterproductive.) There are a few other similarities, too - autistics tend to take most communication very literally, for example, and we tend to have 'special interests' that are similar in some ways to your interest in paperclips. (My fondness for multiples of the number three is similarly arbitrary, and appears to be a similarly intrinsic part of my selfhood, though I have other values that are more important than that one, so I'm not much inclined to spend my time producing instances of the numbers 9 and 81, even though I find them particularly pleasing.)

Appearing autistic is useful in this context because it will - in this context - result in you getting useful information. It's also useful to me and others because the questions that you're asking involve information that we tend to take for granted, and explicitly discussing that information or seeing it discussed explicitly allows us to better understand it and possibly notice irrationalities in how we've been using it. However, in other contexts, failing to adhere to the social expectations of others tends to have rather less useful results, for a variety of reasons, most of which are complex and some of which are fairly arbitrary.

Mercury-based clips are liquid at room temperature, not to mention highly toxic, so I'd recommend against their use in any human-habitable environment. That said, there might be places outside Earth's gravity well where mercury would be useful to you.