Continuation of: The Power of Positivist Thinking

Consider this statement:

The ultra-rich, who control the majority of our planet's wealth, spend their time at cocktail parties and salons while millions of decent hard-working people starve.

A soft positivist would be quite happy with this proposition. If we define "the ultra-rich" as, say, the richest two percent of people, then a quick look at the economic data shows they do control the majority of our planet's wealth. Checking up on the guest lists for cocktail parties and customer data for salons, we find that these two activities are indeed disproportionately enjoyed by the rich, so that part of the statement also seems true enough. And as anyone who's been to India or Africa knows, millions of decent hard-working people do starve, and there's no particular reason to think this isn't happening at the same time as some of these rich people attend their cocktail parties. The positivist scribbles some quick calculations on the back of a napkin and certifies the statement as TRUE. She hands it the Official Positivist Seal of Approval and moves on to her next task.

But the truth isn't always enough. Whoever's making this statement has a much deeper agenda than a simple observation on the distribution of wealth and preferred recreational activities of the upper class, one that the reduction doesn't capture.

Philosophers like to speak of the denotation and the connotation of a word. Denotations (not to be confused with dennettations, which are much more fun) are simple and reducible. To capture the denotation of "old", we might reduce it to something testable like "over 65". Is Methusaleh old? He's over 65, so yes, he is. End of story.

Connotations0 are whatever's left of a word when you subtract the denotation. Is Methusaleh old? How dare you use that word! He's a "senior citizen!" He's "elderly!" He's "in his golden years." Each of these may share the same denotation as "old", but the connotation is quite different.

There is, oddly enough, a children's game about connotations and denotations1. It goes something like this:

I am intelligent. You are clever. He's an egghead.
I am proud. You are arrogant. He's full of himself.
I have perseverance. You are stubborn. He is pig-headed.
I am patriotic. You're a nationalist. He is jingoistic.

Politicians like this game too. Their version goes:

I care about the poor. You are pro-welfare. He's a bleeding-heart.
I'll protect national security. You'll expand the military. He's a warmonger.
I'll slash red tape. You'll decrease bureaucracy. He'll destroy safeguards.
I am eloquent. You're a good speaker. He's a demagogue.
I support free health care. You support national health care. He supports socialized health care.

All three statements in a sentence have the same denotation, but very different connotations. The Connotation Game would probably be good for after-hours parties at the Rationality Dojo2, playing on and on until all three statements in a trio have mentally collapsed together.

Let's return to our original statement: "The ultra-rich, who control the majority of our planet's wealth, spend their time at cocktail parties and salons while millions of decent hard-working people starve." The denotation is a certain (true) statement about distribution of wealth and social activities of the rich. The connotation is hard to say exactly, but it's something about how the rich are evil and capitalism is unjust.

There is a serious risk here, and that is to start using this statement to build your belief system. Yesterday, I suggested that saying "Islam is a religion of peace" is meaningless but affects you anyway. Place an overly large amount of importance on the "ultra-rich" statement, and it can play backup to any other communist beliefs you hear, even though it's trivially true and everyone from Milton Friedman on down agrees with it. The associated Defense Against The Dark Arts technique is to think like a positivist, so that this statement and its reduced version sound equivalent3.

...which works fine, until you get in an argument. Most capitalists I hear encounter this statement will flounder around a bit. Maybe they'll try to disprove it by saying something very questionable, like "If people in India are starving, then they're just not working hard enough!" or "All rich people deserve their wealth!4 "

Let us take a moment to feel some sympathy for them. The statement sounds like a devastating blow against capitalism, but the capitalists cannot shoot it down because it's technically correct. They are forced to either resort to peddling falsehoods of the type described above, or to sink to the same level with replies like "That sounds like the sort of thing Stalin would say!" - which is, of course, denotatively true.

What would I do in their position? I would stand tall and say "Your statement is technically true, but I disagree with the connotations. If you state them explicitly, I will explain why I think they are wrong."

YSITTBIDWTCIYSTEIWEWITTAW is a little long for an acronym, but ADBOC for "Agree Denotationally But Object Connotationally could work. [EDIT: Changed acronym to better suggestion by badger]


0: Anatoly Vorobey says in the comments that I'm using the word connotation too broadly. He suggests "subtext".

1: I feel like I might have seen this game on Overcoming Bias before, but I can't find it there. If I did, apologies to the original poster.

2: Comment with any other good ones you know.

3: Playing the Connotation Game a lot might also give you partial immunity to this.

4: This is a great example of a hotly-debated statement that is desperately in need of reduction.

New to LessWrong?

New Comment
59 comments, sorted by Click to highlight new comments since: Today at 7:51 AM

This is one of my favorite posts yet, but I'm not sure I understand your full chain of reasoning. I understand you to be arguing that we should only be affected by the denotational content of a statement, and ignore connotations as best we can. I entirely agree we shouldn't confuse the two, but I don't see how to go from that to your full conclusion. Is the danger of confusion so great it is worth giving up the extra expressiveness of connotation? I'd appreciate some clarification.

I really like the idea of an acronym, but I'd like one that can be used naturally as a verb. My best shot is "agree denotationally but oject connotatively", e.g I adboc that the rich party while the poor starve.

I don't think we should eliminate connotations from the language, if that's what you're asking.

But in cases where we're trying to use X-Treme Rationality on things that have otherwise resisted debate, we have to use special techniques make the problem more tractable. And one such technique is positivist thinking and putting a big wall up between connotations and denotations.

I like your acronym.

Thanks for the correction. This makes much more sense as a technique than a general principle.

Added your acronym to main post. Please do not doci. Adboc instead.

I don't suppose it's possible to view the version history of the post, so can you state for posterity what "DOCI" used to stand for?


It appears to have been something like "denotation OK, connotations iffy". (Someone objects to "iffy" in one of the comments.)


approve strongly of "adboc" as a new phrase...among less technical crowds, i think speaking the entire phrase "agree denotationally but object connotatively" sounds precise and communicates effectively what is intended...i will be incorporating it.

Does anyone have a good acronym for "I agree with your reasoning, but the fact that you're applying it only here and nowhere else probably proves something bad about you"?

Do You Always Reason Thusly?

That seems similar to what the same author later called an isolated demand for rigor.

This may be excessive nit-picking, I'm not sure; but I just don't think that your examples of the Connotation Game provide connotations as you define them. "demagogue" and "eloquent" have different denotations - it's not just the emotional aura of the words, their actual meanings are different. Ditto for "patriotic" and "jingoistic" and so forth. The point of the game is that the same person/act can be described by flattering, neutral and disparaging words, but even though the same behavior is described by different words, the meanings of the words are different. The game points out inherent subjectivity present in those meanings.

If you used e.g. "black", "colored" and "African-American", then it'd be about connotations. But that's not how the game is played.

More generally, I think you're trying to give the concept of a connotation a larger role than it's used to. Connotations are about differences in phrases like "ultra-rich", "stinking rich", and "independently wealthy". In the phrase about the ultra-rich you quote, it seems difficult to assign the subtext being carried to connotations, unless you want to use "connotation" in the sense of "subtext, any implied meaning". But I don't think those shoes will fit the word.

Subtexts are very important, and are used insidiously all the time. They're not always emotional; sometimes they exploit logical structure, as in the canonical example "Have you stopped beating your wife?" It's difficult to stand your ground against them, and forcing them into the open, demanding that they be made explicit, as you suggest, is a useful technique. Unfortunately, it usually won't work on someone who feels entitles to use a subtext against you - due to either intentional demagoguery or misguided self-righteousness, for instance.

One framework that seems to make a careful study of implied meanings is that of framing. I don't really know what they're doing with this and whether it's more than a useful metaphor. But, to repeat myself, I think that "connotations" might be too straight a jacket to use for the notion.

Even more confusing: often statements carry useful implications in their subtexts, and often these useful implications are obscured or removed if one tries to keep all meanings explicit and all claims {verifiable/falsifiable}. For example, consider the use of the terms "bleggs" and "rubes" in Eliezer's post disguised queries. The foreman in the story tells Susan about objects that are "bleggs" and "rubes", rather than just substituting in the visible criteria about blue-ness and egg-shaped-ness, so that the implication that said objects represent "clusters in thingspace" can guide Susan's further learning about the objects.

Much useful inference is done by updating the meaning of a term after its creation (as with the bleggs and rubes, or with Einstein's vs. Newton's notion of "space"). These update-possibilities would be harder to notice if we regarded words as mere shorthands for particular verifiable criteria. I agree with Yvain's suggestion to more often make our disputes verifiable, but I have also seen people insist that everything be operationalized into concretely verifiable terms in a manner that may make it harder for them to update their deeper concepts.

If you used e.g. "black", "colored" and "African-American", then it'd be about connotations.

Well... I'm pretty sure a dark-skinned Nigerian who has never been to America and has no recent American ancestors would count as black but not as African-American.

This post just appeared at Language log and has this to say:

Language Log could devote a thousand posts to the project of underlining and elaborating the ways in which grammar does not protect us against misunderstanding the sound of an uttered name, and logic does not protect us against what we say having double meaning. Come to think of it, the thousand posts may already have been written: there are over 5,500 old posts searchable here and already over 1,200 new ones on the present server searchable here.

Asking others to clarify their assumptions is good, but translating them into positivist language might not be as precise as it sounds

One theme of OB and LW has been to take the fuzzy complexity of the real world and show that in principle at some level it's related to something precise. We can't actually do Bayesian math in our heads for real-world calculations, but just knowing how to work the ideal case protects us against certain real-world errors. Likewise, Eliezer's meta-ethics reduces morality down to some horrendously complex thing that we can never calculate, but it's nice to know that morality does reduce to something when we're wondering whether it exists at all, whether it's all relative, or so on.

The real positivists thought they could reduce all language to their positivism and spent thirty years trying. I don't think I'm going to do that in a few days of posting about stuff on a blog. But if I can sketch a few really-large-scale things and then let people's common sense fill in the blanks, that'll still be better than nothing.

Speaking of themes; I guess one thing that bothered me about this post (many of your other posts are very good) is that this post doesn't seem to serve a point; questioning assumptions is often brought up on OB and LW and asking others to be more precise in describing their assumptions is also very common. Any connection to positivism here seems very tenuous; criticizing positivism has little or no impact on soft-positivism.

I feel that there are many other OB and LW posts that would address this issue more effectively and it might be better to just make this page an index of them as opposed to a full post.

That said, there is no way to easily recognize a lot of the themes here and LW in particular runs the risk of just becoming a repository of the same things over and over again.

Are we really refining the art of human rationality here?

You could be right; I have no formal training in this field. It sounds like you don't like "subtext". Suggest another word and I'll switch to it.

I think "subtext" is fine in this role.

I feel like I might have seen this game on Overcoming Bias before, but I can't find it there

The game is familiar to me from Yes, Minister, a TV programme which was an expert satire on British politics. Bernard Woolley, a senior civil servant, would refer to these as "irregular verbs". From the quotes page at IMDB: "That's one of those irregular verbs, isn't it? I give confidential security briefings. You leak. He has been charged under section 2a of the Official Secrets Act. "

More irregular verbs here.

The first one of these I can remember reading was "I'm erotically open-minded; you're kind of kinky; he's a disgusting pervert."


Objections to this statement seem to be 1) the highly loaded descriptions of the rich and the poor and 2) the juxtaposition of the descriptions without an explicit relationship.

While an examination of word choice might allow you get to a less loaded formulation of its content. eg. The rich, who have much, enjoy luxury while the poor suffer. It doesn't get to the fact that the statement is an attempt to draw us into an implicit connection between the two descriptions. The statement is only connected by a "while", which might connect any two facts, but in practice we only bring the facts together in order to contrast or compare them. This is a nice set up for the rhetorician because the connection we make is ultimately our own, and often not explicitly known to ourselves. The capitalist objects, but is not sure to what, precisely, because he can't find it in the statement.

Now if I object to the statement of the grounds that it is emotionally loaded, I am not directly addressing this implicit point of the statement. I could fall into the trap of being seen as an apologist for the rich if I only object to their description but not the implication.

But I wonder if I should object that statement is not untrue as such, but that the point of it isn't clear, would that be taken to mean that I just don't get the implicit point of it? That I'm simply unsympathetic to the emotions of the statement?

Why not make the implicit explicit (as it can be)? Why not treat the statement as if was an explicit statement of what it tried to be implicitly?

I think that point of the statement is to assert either that the rich are directly responsible for the condition of the poor, or at least that the rich fail in their moral obligations toward the poor; response:

Sorry, are you saying that the rich are directly responsible for the condition of the poor, or just that the rich fail in their moral obligations toward the poor?


This is a nice way of summing up something I do with my clients, teaching them to remove the negative connotations of belief statements. For example, a statement like "life has no meaning and everything I do is pointless" can be denotationally true, but that doesn't automatically make it a bad thing.

One easy way to expose connotations in your own thinking is to ask "So what?" or "What's bad about that?" or "What's good about that?".

(The latter's a lot less useful though, since the majority of motivated reasoning is negatively-motivated. "What's bad about that" is at least 10-20 times more likely to produce a useful answer, within the context of my work anyway.)

I actually think people who say "Life has no meaning and everything I do is pointless" are actually making a deeper mistake than confusing connotations with denotations... I think they're actually making a denotational error in missing that e.g. "purpose" or "pointfulness" typically denotes ternary relationships of the form "The purpose of X to Y is Z." In other words, one must ask or tacitly understand "purpose to whom?" and "meaning to whom" before the statement makes any sense.

My favorite connotationally-heavy follow-up to this is that "My life has as many purposes as there are agents it has a purpose to."

I've been wanting to post on this under the name "Relation Projection Fallacy" for a while now, so I just did :)

This reminds me of something I heard someone say recently - that much of the content of speech is actually sleight-of-hand. It isn't there to inform us, but to distract us into evaluating and agreeing with its denotation while it slips in its connotation unnoticed.

I though this had a name: spin.

Sure, but it is not widely recognized that a statement may be both spin, and factually true.

Isn't it? "Spin" doens't mean lying, it surely means twisting interprertations, connotations...

For instance, WP' list of spin tactics does not include outright lying.

  • Selectively presenting facts and quotes that support one's position (cherry picking)
  • Non-denial denial
  • Non-apology apology
  • Mistakes were made
  • Phrasing in a way that assumes unproven truths
    • Euphemisms to disguise or promote one's agenda
    • "Burying bad news": announcing one popular thing at the same time as several unpopular things, hoping that the media will focus on the popular one.

I find just saying "and?" has a similar effect without requiring explanations.

Quoting someone else's comment from another thread:

there is also the matter of the habit that some intellectuals have of shocking the bourgeoisie. If you say that the bad guys think that they're the good guys and we're the bad guys, you probably won't raise any eyebrows. But if you make the statement in a way that implies that you agree with the bad guys' assessment, or that you are positioning yourself as a neutral party who favors neither side, then you will probably raise some eyebrows. And based on my own experience, an awful lot of people like to present the rather familiar and tired and unremarkable view that the bad guys think that they're the good guys, in just such a way, so as to maximize their effect on their listener. This seeming undercurrent of support for the enemy is something that can be easily avoided without changing the factual content of what you're saying, but it is in my experience often not avoided, indeed, it seems to be sought out and nurtured. And then, when the predictable reaction occurs, like clockwork Mr. Epater-les-bourgeois loudly complains about the impossibility of making obviously true statements in front of the the foolish masses.


I usually use the phrase "only literally."

I feel I must point out that 'connotation' and 'denotation' tend to be used differently by professional philosophers. 'connotation' refers to intension, while 'denotation' refers to 'extension'. That is, 'denotation' refers to a word's referent (the set of things in the world to which it can refer), while 'connotation' refers to its meaning.

Forgive me if I'm repeating anyone here, but it seems to me that there's a very large problem with reducing a statement like this to a strict statement of fact that's independent of phrasing or specific connotations of the words used.

The problem is that a human brain automatically interprets the statement as having been issued by an agent, with an agenda. Therefore it becomes not an unbiased sample from truth-space but a deliberate selection, chosen with intent. Phrase it how you like, you've still made an implicit statement about the importance of all the other things you didn't say. If I were the capitalist I imagine my reaction would be to think, fine, but... [some other true fact I consider more important]. It's the prioritization of the original statement that I take issue with.

For some reason I'm a critic this evening, forgive me, but I don't think this one meets the standards of your previous articles; it feels to me that this doesn't move us far beyond the widely appreciated fact that for every portion of direct implication an utterance has, there's usually a generous side-salad of connotation. Am I missing a key novel move?

No. I almost put a footnote in saying something like "But you already knew all this, didn't you?"

The reason I wrote it anyway is because of the last four paragraphs. The particular debate failure mode I outlined there - Person A makes a denotatively true but connotatively controversial statement, Person B is offended by this statement so tries to argue against it poorly or make some similarly offensive statement - is one of the most common problems with the arguments I hear. I want to be able to say "Hey, guys, adboc" and give a link to this article.

It also fills in something I thought was a bit of a hole in last night's post. A statement like "Islam is a religion of peace" is a "dangling variable" underdetermined by the empirical facts it purports to relate to. So why does it matter which way we evaluate the proposition? It matters because of all the associated connotations. Since it has practically no denotation, saying "Islam is a religion of peace" is just saying "Please accept all these connotations here" without even the saving grace of a denotation to attach them to.

...tell me what you thought of that preceding paragraph. I can't decide whether it's obvious and a rehash, whether it's uninteresting, or whether it deserves a post of its own.

I'm glad you wrote about the debate failure mode. The clear picture you gave of it will probably help me in future conversations; I'd kind of noticed the phenomenon, and I'd kind of noticed how to dodge it, but much less clearly.

Although I might appreciate reviews of the basics more than most people do. It seems to me there's a great deal that I kind of roughly get on an abstract, "get the right answers on a multiple choice test" kind of level, but that I haven't yet integrated into my root vision of the world. Meditating on the basics helps me, there.

There's an old quote that poor people are crazy and rich people eccentric... I'm sure someone can come up with a third word to fill in the trio.

I think this technique is often useful in arguments, and I imagine most people here use it to some extent. A reasonable summary of "YSITTBIDWTCIYSTEIWEWITTAW", is "true, but so what?" (although we might need something with a less aggressive connotation...). Although this may be part of the "Dark Arts", it's often very unnerving when you're arguing with someone when they accept without question factual statements you were all set to have to defend to the hilt.

I'm well-off, you're wealthy, they're ultra-rich.

I can't think of a word that sits well between eccentric and crazy, but the following works similarly:

I'm eccentric. You're unusual. He's a freak.

There was a three-page version of this joke in all variations in a Mad Magazine. Too bad it's probably not even in the deep web not in the web at all.

I'm normal, you're eccentric, he's crazy.

Shouldn't this be "I'm unique [...]"?

Conversations seem to occur on several levels simultaneously. There's a level of literal truth. There are also multiple dimensions of politics. (What I call "micro" and "macro," in a way analogous to the application to economics.) There's even a meta-level that consists of just trying to overwhelm people with verbiage.

Something related happens with me every once in a while when someone makes a statement of the form A -> B and I say 'yes' or 'ok' in response. By saying 'ok' all I am doing is to acknowledge the truth of the statement A -> B, however, in most cases, the person assumes that I am agreeing that A is true and hence ends up concluding that B is true as well.

One example is this -

I go to a stationery shop and ask for an envelope. The storekeeper hands me one and I start inspecting it. The storekeeper observes this and remarks, "If you want a bigger envelope, I can give you one." I say, "Alright." He hands me a bigger envelope.

When someone asks you if you could pass the salt, do you pass the salt? Or just say "Yes"?

FYI, I think you were looking for "No, thanks"


The word "connotations" is cute but it's a bit of a dodge. It's a non-label label. "Everything else". Fair enough, but everything else how? The statement is denotationally pretty tight, no wiggle room. There isn't a hidden channel in HTML text. Where's the connotation communicated?

I conclude: the connotation exists in the relation of the statement to the context. First order connotations are the denotation in relation to the world as seen by the speaker or listener. Second order, as seen by the speaker guessing what the listener sees, or vice versa. And so on.

The payload in the phrase discussed above is a third order connotation: the speaker's moral weighting of what he guesses to be the listener's ethical analysis of the denotation.

(more to come)

The word "iffy" in your acronym should be replaced, I think.

a quick look at the economic data shows they do control the majority of our planet's wealth

Not really, actually. The ereason is a bit subtle.

A very large portion of the planet's wealth consists of the capacity of people to earn income by working. Most people have an asset - the value of their labor - which is no counted in standard tabulations of wealth.

Wealth in terms of financial capital is indeed distributed in a very skewed fashion, for a number of reasons. Wealth in terms of ability to earn income is still skewed, but far less skewed.

You can see this when you look at the distribution of financial wealth versus the distribution of financial income.

The difference is enormous. I consider this to be an example of a factoid - true but misleading.

What do we need the acronym ADBOC for? If we agree on denotations, our rational evaluations of the argument will be the same.

Our judgments will differ only when our judgments are affected by the emotional associations of the words or phrases, and those associations are irrelevant to the logical structure we're supposed to be evaluating.

Our judgments will differ only when our judgments are affected by the emotional associations of the words or phrases

I don't agree with this; subtexts can have claims and logical points implicit in them, and those can be relevant but mistaken.

If the argument exists only implicitly, then it's not a rational point and is necessarily ignored in rational discussion.

I think I once believed that rational argument was about forcing someone to accept a series of claims by showing that they all follow from each other, but it just doesn't work that way in practice. When it works it's much more like handing material to someone who is (at least somewhat) cooperating with you in trying to figure out the logical structure of a problem -- pushing the right buttons of insight, so to speak.

Even if you're right that rational discussion doesn't take anything into account that hasn't been spelled out explicitly, things like "ADBOC" can help us cope with the inevitable irrational elements of real discussions.

When it works it's much more like handing material to someone who is (at least somewhat) cooperating with you in trying to figure out the logical structure of a problem -- pushing the right buttons of insight, so to speak.

Precisely! The only way to really change your mind is to get into a mode where you're actually asking the questions, and genuinely looking for answers. It's the first skill a rationalist (or a mind hacker) has to develop, because it's the only way to bypass motivated reasoning and rationalization.

Checking up on the guest lists for cocktail parties and customer data for salons, we find that these two activities are indeed disproportionately enjoyed by the rich, so that part of the statement also seems true enough.

P implies Q does not imply Q implies P, surely.

I was bothered by this as well. The statement wasn't "cocktail parties and salons are patronised by the ultra-rich", but "the ultra-rich ... spend their time at cocktail parties and salons". So it's as you say, what you need to look at is not the guest lists for cocktail parties and customer data for salons, but what proportion of a typical ultra-rich person's time is spent at cocktail parties and salons. I don't have the data, but I'd anticipate the mean ultra-rich person spends more time managing their business concerns than attending cocktail parties.

Though this is all Support That Sounds Like Dissent, since none of it really detracts from the central thrust of the post, with which I broadly agree. Still, no point leaving holes in something which is political enough for people to have a good deal of motivated scepticism about it.

I think the "...and that's terrible" is pretty clearly implied. What exactly is wrong with the quote? It looks like you're dissecting a straightforward appeal to people's (stated or real) anti-unfairness values, as if it's a given that it's dishonest. I don't get it.

"All three statements in a sentence have the same denotation"

Alas, that's not the case.

Also, we already have access to a language that is purely denotational, with all of the connotations excluded. We call it mathematics. It's useful and nifty.