From what I understand, technological progress happens exponentially.


Technological progress happens exponentially.

The difference is the "from what I understand" part. Other examples of such qualifiers include:

  • It seems to me
  • My impression is that
  • I could be wrong, but
  • Perhaps
  • Probably

Using such qualifiers may be annoying. It is probably easier and quicker to just skip them. It also might add "fluff" and distract from the main point.

On the other hand, I worry that omitting them would lead to overconfidence. If you say  "X is true" enough times instead of "It seems to me that X is true; I'm pretty confident but not super confident", it seems likely that you'd develop a nontrivial overconfidence in X.

Of course, the answer to "Should we use qualifiers in speech?" is almost certainly, "It depends". But despite that, it still seems like it'd be pretty useful to figure out what the "default" or "your go-to" should be.

Personally I lean pretty strongly towards using them when doing intellectual things, like here on LessWrong. But in everyday life I lean towards avoiding them, because it goes against norms, is a little awkward, and doesn't have nearly as big a benefit as when you're doing intellectual things. 

New to LessWrong?

New Answer
New Comment

14 Answers sorted by

When I was studying Lojban, I learned about the concept of "evidentials" such as

The intended usage is to make explicit how the speaker came to think whatever follows. Of course, this is different from hedging with phrases like "I could be wrong". I remember thinking at the time that it would be useful to port the use of evidentials to some English conversations.

I use a lot of qualifiers in speech and in writing. I find that, in both informal and professional contexts, most people don't even notice them and interpret my sentences as if they weren't there. It regularly trips me up when people come be to me and say, "You said X, and now you're saying/doing Y!" and I say, "Well, no, actually I said [Qualifier]+X, and [evidence Z that we later encountered that led me to Y] is exactly the kind of reason why I didn't just say X." Sometimes even me saying "X, unless Z, then Y," isn't helpful, people just hear X anyway, and may try to hold me to X later.

So, my take is that the benefit of qualifiers is dependent on the audience you're talking too, and how well they understand why you're using them. I use them anyway because it's important to me to not pretend more confidence than I have, but also because I'm a perfectionists who sometimes twists his words with qualifiers so I won't later find out I was wrong and can rationalize it away. I'm working on that, but it's slow going.

Some people will misunderstand you whatever you do. That doesn't mean that you can't or shouldn't communicate more precisely with everybody else. Some people will notice. 

If I (correctly) expect to be misunderstood, then the precision is only in my head and not actually facilitating communication. It can still be useful, if it's recorded and I can call it back up later for some purpose, but otherwise it doesn't help me or whoever I'm talking to. At worst, it ironically brushes up against knowingly misleading people.

This is also my experience. Other misunderstood phrases are even simpler; I've told people "We need to do either X or Y" and have them come back later and say "X is impossible", and then be surprised when I asked about Y.

Useful writing tells people something true and important that they didn't already know, and tells them as unequivocally as possible.

How to Write Usefully by Paul Graham

Removing qualifiers from my speech and writing helps me think clearly because it forces me to make positive statements about reality. Compare the following three statements.

  • "China is the center of human history."
  • "I think China is the center of human history." ― If you actually () thought China is the center of human history then you wouldn't say "think". You would say "China is the center of human history". If you are unsure then you could say "I wonder if China is the center of human history." Saying "I think…" isn't exactly wrong, but it would be better to suspend judgment while you find out whether China is the center of human history. If you know the truth then you can state it with confidence. Using prefaces like "I think…" is putting your opinion before the facts. It changes the topic of a conversation from reality to your thoughts, often precipitating an epistemic train wreck.
  • "I believe China is the center of human history." ― The "I believe…" qualifier is even worse than "I think…" because believing in belief damages your epistemics.

A statement without qualifiers is the most meaningful because it is the most narrow. (More precisely, it forces entropy out of a probability distribution.) Qualifiers like "I think" signal lack of confidence. Such signals accurately calibrate statements when you truly do lack confidence. Signaling lack of confidence when you are confident obfuscates communication.

In my real life experience, people who demand unnecessary qualifiers are dangerous to consort with. (By unnecessary qualifier, I mean redundant words like "It seems to me…" when you know something to be true. Qualifiers like "one of the ___s" are fine if you really mean "one of the ___s".)

If " is true" then I will say " is true". If " is not true" then I will say " is not true".

If " is true" then I will say " is true". If " is not true" then I will say " is not true".

I just want to point out that "X is true" and "X is a fact" are themselves a type of qualifier, and it could be interesting to ask when/why a person says this rather than merely "X". Are these somehow better than other qualifiers? If so, why?

For example, many people are more inclined to talk about "facts" and "truth" when talking about contentious political issues, while rarely using those words at all when talking of practical matters like bi... (read more)

Many popular qualifiers don't have clear meaning. As lsusr noted, there is a sense in which the qualifier in "I think Z" is usually either empty of content or a lie. At the same time, applying charity to the uses of the more general "It's my opinion that Z," it's easy to see at least four good uses: (1) communicating a hypothetical without claiming any credence, (2) expressing a belief without offering a justification, (3) communicating uncertainty, (4) expressing uncertainty without offering a justification. But for those uses "opinion" is not a good qualifier. For hypotheticals, it's better to say "Consider the possibility that Z." For beliefs, it's "I believe Z, though I'm not going to discuss my reasons right now." For uncertainty, it's better to quantify the uncertainty, saying "It's not unlikely that Z, because J." And nobody tries to say things like "I believe it's not unlikely that Z, though I'm not going to discuss my reasons right now." (So perhaps just saying "opinion" is a sensible shorthand for all the possibilities, as long as the abominable ambiguity of the qualifier is understood.)

In the last case, not discussing reasons for choice of credence is similar to not discussing reasons for a belief. What you should put out into the world then is a claim about your own state of mind, as making sensible use of it requires distinguishing it from the underlying claim about the world. It's only when you do offer justification (or expect it to be apparent) does it make sense to insist on the underlying claim itself. Even then, it's better (more modular!) to communicate arguments, not conclusions (likelihood ratios, not credences).

When it comes to qualifiers it's good to actually use them to communicate information instead of just making statements defensible. 

If you preface every statement with a qualifer then the qualifiers that you use communicate no information. If you however preface those statements that are less certain with qualifiers, then you communicate information.

When editing a post, it's worthwhile to ask for every qualifier what value it adds and scrap it when it doesn't add value.

 Of course, the answer to "Should we use qualifiers in speech?" is almost certainly, "It depends". But despite that, it still seems like it'd be pretty useful to figure out what the "default" or "your go-to" should be.

This could be easily reformatted into "The answer to "Should we use qualifiers in speech?" is almost certainly, "It depends". It's still useful to figure out what the "default" or "your go-to" should be."

The added words don't provide any value and habitually adding unneccessaru words means that the qualifiers that actually do convey information are more likely to be ignored.

This could be easily reformatted into "The answer to "Should we use qualifiers in speech?" is almost certainly, "It depends". It's still useful to figure out what the "default" or "your go-to" should be."

I agree with the broader point, but in this case I wasn't certain enough in what I was saying to omit the "seems like" part. Maybe the truth is that it's very much a case-by-case thing and defaults aren't in fact much use.

To the extend that you wanted to express some uncertainity it didn't get through to me when I read it. I think part of the problem is that the nature of your uncertainty isn't expressed.

I think they are basically epistemic security theater. 

Then why did you begin your comment with "I think"?

I think the "I think" could perhaps be considered (basically) part of a joke, from a certain point of view.
Oh, that's possible. deluks917, my apologies if I missed your joke and then tried to make the same one more clumsily.
2Adam Zerner3y
I assume that the claim deluks917 had in mind is that heavier usage, eg. more than what a normal person would use in everyday life, is basically security theater, but below that threshold it's fine.
Me too. I just saw the opportunity for a bit of hopefully-amusing snarkiness :-).

I think they can be used that way sometimes. And when they do, it is harmful. I recall hearing someone on LessWrong comment recently about how rationalists often don't say things "loudly" enough.

But I also think that there are a lot of times where they are used appropriately. At least here on LessWrong and in other groups of respectable people. So saying that they're basically epistemic security theater seems like a big stretch to me.

Know your audience. Does your audience generally read carefully and parse statements precisely? The message you send to LWers or to a group of lawyers might be different from the message you send to people who are less highly verbal or only patient enough to listen to the first few words.

If you've been told you come off as arrogant, which I expect is true for a lot of people here, more qualifiers can mitigate that. 

“I am inclined to think—” said I.

“I should do so,” Sherlock Holmes remarked impatiently.

Arthur Conan Doyle, "The Valley of Fear"

Sounds a bit like E Prime

I have used this in writing for good results. I have not tried to use it in spoken communication but I predict it to take more effort to get started there. 

In writing, I take a hard look at any dubifiers I notice, and only let them stand if they are really necessary. I find that often (a quantifier I have let stand!) they result from mere timidity rather than justified, significant, and relevant uncertainty. In speech too, if I'm quick enough to make these decisions on the fly.

I especially avoid multiple dubifiers, like "It seems to me like there's a chance that probably it might be a good idea to maybe try and see if it's possible to..." As deluks917 said, epistemic security theatre. Or in that concocted example, epistemic security farce.

I like qualifiers that give information on the person's epistemic state or, even better, process. For example:

  • "Technological progress happens exponentially (70%)"
  • "My rough intuition is that technological progress happens exponentially"
  • "Having spent a year looking into the history of technological progress, I concluded it happens exponentially: [here are specific data sources, or links, or some sort of pointer]"

Given that I don't start thinking that anyone can report directly the state of the world (rather than their beliefs and understanding of it), "From what I understand, technological progress happens exponentially" does not provide much information (particularly because there is large interpersonal variation in the strength of hedging a given qualifier is supposed to convey).

Sometimes I feel forced to add qualifiers because it makes it more likely that someone will keep engaging with me. That is, I am confused about something, and they are helping refine my model. By adding qualifiers, I signal that I'm reasonable / understand I should be uncertain of my conclusions / am open to updating.

Given that I don't start thinking that anyone can report directly the state of the world (rather than their beliefs and understanding of it), "From what I understand, technological progress happens exponentially" does not provide much information

When interpreted literally, I agree. But "from what I understand" connotes a sense that you're not super confident eg. because you're talking about something that you're not an expert on.

Yeah, that's what my parenthetical was supposed to address Perhaps you are able to get more reliable information out of such statements than I am.

When I'm as sure as I can be about something, I won't use qualifiers. For my areas of interest, I'll try to get to this stage, which has the added benefit of making my language more concise. If you're unsure, you should qualify, but if you qualify a lot, why are you talking? (However, for some question domains, things can't be known, like predicting election outcomes, in which case, it's fine to get on a soapbox and qualify.) It's a mistake to cut out your qualifiers if you haven't done the hard work of figuring out all the details.

Of course, qualifiers should be common because people talk about stuff they don't know all the time, and they want you to engage with them. It'd be a bit weird to say "I don't know" and then walk away for 99 percent of your interactions. In these situations, I'll try to include qualifiers. Sometimes I'll forget, and state something as if it were a fact only to be categorically shown to be wrong two seconds later. I hate this. So, for me, qualifiers are worth it. But even if you're not embarrassed when the truth is literally the exact opposite of what you just said, qualifiers are good. They help you delineate between what you know to be true and what you think is true, which is useful for your own thinking. They also communicate your actual beliefs.

Even better than this binary distinction is using credences.

I use qualifiers rarely in writing, and only when I wish to emphasize my own lack of confidence or conviction. I think I am even less likely to use them in speech.

There are some situations where altering your language to drive down overconfidence (or the opposite) seems to be a good idea - namely ones where you don't want to trust your intuition or reasoning very much (or when you realize you are too epistemically humble). According to Sapir-Whorf, intentionally injecting or excluding qualifiers in a statement can change a speaker's/listener's/reader's perceived level of confidence in the statement, as you alluded to in your question.


Politicians (much like many smarter-than-average children) like to use qualifiers so they don't have to commit to anything, or so that two sides of a debate can each interpret their statement to map a candidate onto their side. Take this slightly exaggerated example where, I started with "X is happening" and inserted three qualifying fluffs ("as it has been presented" vs first-person observation, "seems to" vs definitely, "look like" vs some form of "to be"):

 "The situation as it has been presented to me seems to look like X" - Candidate

If you believe in X, the candidate confirmed that they also believe in X, so you can support them. If you believe in Not X, but you wanted to support the candidate anyway, this formulation allows you to tell yourself that they:

  • don't really strongly believe in X; or,
  • maybe they will change their mind once they are in office/have all of the facts/etc.; or,
  • they really believe in Not X but they have to try to win an election so they need to give off the impression that they support X to not piss off special interests/the party elite/primary voters/etc.

The third option in particular tends to be transparent mental gymnastics to anyone else but the believer. It's a way to resolve or bury cognitive dissonance created when someone supports a candidate they don't 100% agree with.

Suggestion: continue to use qualifiers, but encourage writing them as acronyms so that they take up less space. Same meaning but quicker and less annoying.

I like that idea. Coordinating seems difficult — how do you create the acronyms and get people to know what they mean? — but if we were able to get past that I'm pretty sure I'd enjoy the acronyms.

There are some acronyms that are already in common usage.
4 comments, sorted by Click to highlight new comments since: Today at 3:48 PM

I don't have a full answer, but here's what seems important to consider - in my experience, the baseline for the level of confidence in speech that is associated with competence and authority is a lot lower in intellectual circles like LessWrong, compared to the general public. 

This is because exposure to rationality and science usually impresses into someone that making mistakes is "fine" and an unavoidable component of learning, and that while science has made very impressive progress there is still a lot to learn and understand about the world. On the other hand, the real world and social opinion usually very closely associate mistakes with failure and the ensuing moral penalties and lowered status. 

Based on that, if you rely a lot on qualifiers while speaking to someone who's not as exposed to or interested in intellectual thought, they may write you off as confused or unsure - they will expect "smarter" people to give them definite verdicts. So if you're trying to socially maneuver someone into agreeing with your reasoning based on competence, forgoing qualifiers is probably a good idea.

Agreed. That makes me think back to the following story.

It was my first job as a programmer. The PM would ask me questions about whether X is doable for me or how long it would take me to do Y. I would give my honest response, which often would be that I can't do it without help or that it would take me a while. Then one day the tech lead sat me down and said that it's important for me to project confidence. It frustrated me a lot because in theory the best thing to do for the company would be for me to provide accurate information and then try to make the best decision based off of that accurate information. Now I realize that the tech lead probably wasn't thinking about that and probably was just trying to look out for me, knowing that my lack of confidence would end up hurting me.

I'm curious, after your work experience since then, if you still think the tech lead's comment was about the information you were conveying, or in your tone and specific wording choice. Like if instead of, "I can't do it without help or that it would take me a while," he knew you'd do better rephrasing it as something like, "In order to do that, I'll need to allocate X days/weeks, as well ad [any other resources or people you expect to need, for how long, for what parts]," followed by (depending on the company's practices) "Would you like me to write up a proposal for that?" or "How should I prioritize that relative to my other projects?"

In my experience, "Projecting confidence" can also often be achieved by speaking clearly and precisely, stating things in the affirmative more than the negative, and claiming enough status to be worth the investment necessary to do what's asked of me, without compromising the informational content aside from emotional valence. It also makes it easier for my bosses to act on my answers without putting in planning work themselves.

I feel pretty confident that he was referring to the information I was conveying. Ie. that it should be "yes" instead of "no". I think he was trying to protect me, because if I don't tell people what they want to hear I'll develop a bad reputation.

Thanks for explaining how you would rephrase it/go about it. That makes a lot of sense and I feel like something clicked for me after reading it.