Hey Richard, thank you for your critique of my tweet. I actually agree with you that "prior" should sometimes be used instead of "assumption", and "stochastic" should sometimes be used instead of "random". If you read my tweet again, you'll see that I only advised against using the obscurer word in cases where you could just use the simpler word (i.e. where it wouldn't significantly change the meaning of the sentence).
I also agree with you that language isn't always about saying what you mean (though I think it should be). You say you want to signal your intelligence, and, well, you succeed in that -- I do find you intelligent -- not because you use clever words but because you use clear words to say things that are clever.
Thanks for the kind words. I'll say though that you can signal intelligence by both clear writing and obfuscating. It's actually hard to write like an academic, most people can't do it. Clear writing makes a person more understandable, but then the ideas are there to be judged on their own merits.
"A prior is something you have good reason to believe based on previous evidence. When you bring your prior with you to analyze a question, you are evaluating new data in light of it."
No. (Bayesian) prior implies no such thing. Prior can be based on anything - in statistics it is often a uniform distribution for a parameter, meaning think any value is equally likely. However, prior implies that you will be gathering evidence going forward and that you will change your prior accordingly. In fact, priors are often of little relevance because they are overridden by actual data. So when frequentists complain "so where do you get your priors", Bayesians tell them - "it doesn't matter because our conclusions are driven by the data, not priors". The point of Bayesianism is in the updating process, not in well thought out priors.
This whole post is about the idea that connotation is a thing that exists such that two words that superficially seem equivalent will put slightly different ideas in the reader's head.
The connotation of the word "prior" is that the person has put has put enough thought into their underlying assumptions that they could quantify their beliefs if asked. While it's true that nothing about the technical definition of prior requires that the prior be well-thought-out, there is still the connotation that "priors" have some rigor while "assumptions" do not. And that's all that matters.
There is the seperate interesting point that connotation depends very much on the person in question. For someone from a non-technical background, "prior" sounds mathy so it will be given an appropriate level of respect. But to someone who works formally in statistics, the word "prior" might have lost its glamour a long time ago such that there is no connotative difference between the word "prior" and the word "assumption". So this is also a lesson in knowing who your audience is. Since Richard writes mostly for smart laymen as opposed to statisticians, his point still holds.
The word "prior" never had any glamour. It is a much discussed weakness of the Bayesian approach, not the strength.
I mean, fine, as someone pointed out, people are now using word 'literally' to mean its exact opposite. So sure, use 'prior' to mean 'evidence based, well thought out assumption'. But perhaps at least don't rationalize this by supposed benefits that the use of technical language brings to the discussion?
Yes. Although his fundamental point - that there's a difference between what we can take away from someone writing "prior" rather than "assumption" - continues to hold. A prior implies a best-effort to obtain probabilities before further information is acquired, while an assumption is something much more likely to have had no thought put into it.
I agree that "prior" both denotes and connotes something rather different from "assumption", but I think Gurwinder may be reacting to widespread misuse of the former word. The rationalist sphere is overflowing with cargo cultists who love to dress up their perfectly ordinary human intuition (i.e. biases, spite and fundamental attribution errors) with a superficial veneer of Bayesian-style language. This was brought home to me by the recent kerfuffle over the EA startup Nonlinear (https://forum.effectivealtruism.org/posts/bwtpBFQXKaGxuic6Q/effective-aspersions-how-the-nonlinear-investigation-went), in which various figures involved were prone to framing their intuitive gut feelings as the product of careful, disinterested statistical reasoning: "But I think it would still have over a 40% chance of irreparably harming your relationship with Drew"; "Nonlinear's threatening to sue Lightcone for Ben's post is completely unacceptable, decreases my sympathy for them by about 98%". It's a linguistic device intended to give the appearance of solidity to pure wind, a verbal tic no different from the habitual misuse of the word "literally", or concluding a statement by tacking on the word "FACT" at the end in a tone of voice meant to sound definitive and unanswerable. What does it MEAN to say that doing such-and-such has a 40%+ chance of irreparably harming one's relationship with X? I'm in a rationalist WhatsApp group in which I saw someone use the phrase "I have pretty irrational priors on..." without so much as a glimmer of irony or self-awareness. When even the people using the word "prior" are openly admitting that their alleged "priors" are irrational, it seems reasonable to point out that "assumption" or "gut feeling" might capture the intended (or actual) sentiment better than "prior".
If you're discussing some empirical question, and you've done the work of digging through reams of data on said question before the discussion began, by all means use "prior". If you haven't (because you're lazy, or because no relevant data actually exists, or the question isn't really an empirical question at all) don't pretend you have - just be honest and say "assumption" or "gut feeling" or "intuition".
“ What does it MEAN to say that doing such-and-such has a 40%+ chance of irreparably harming one's relationship with X? ”
It means that if you were going to bet, you would buy at a price less than 40 sense on the dollar and not higher. Putting probability on events is good and more people should do it in more situations.
It doesn't mean "if you were going to bet...". It means that you actually took the bet. Personal probabilities are not based on introspection. Instead, bets you actually take reveal your degrees of belief. There is nothing magical or rational about merely framing your beliefs in probabilistic language.
I agree with Gurwinder. I agree with George Orwell’s advice in “Politics and the Abuse of the English Language”. The two example sentences are much better with non-fancy words. The fancy words cause unnecessary effort and add little necessary meaning. This isn’t fictional literature nor technical writing aimed at a specific audience.
I worked with a guy once who had this self-improvement technique -- he'd learn a new word every day, and try to find ways to use that new word in a sentence. One of the senior trial lawyers in our firm said that young lawyers should instead try to forget one word every day.
I've also come around to 'prior' after reading SSC subreddit, and I agree that Bayesian prior sounds pretentious from most people, with the exception of Harry Potter Evans Verres.
But don't use it if you're not sure the person will understand. It is still jargon.
> Yet prior is not an obscure term. If you haven’t heard of it, you probably are not that well-read.
Well, I'd say that "prior" is well-known among people who have a grasp of the conceptual structure of statistics. But I'd also say that is less than 1% of the population. Where you place the cutoff for "well-read" may vary, though.
I recently noticed that I've been overusing the word "prior(s)", saying it automatically, even when it is not quite what I mean. I've been idly groping around for better phrasings.
1) I like to use "base rate" when I mean to refer to the actual frequency of something in real life, especially on topics wherein base rate fallacy lurks. https://en.wikipedia.org/wiki/Base_rate_fallacy
2) I'll sometimes refer to "the stereotype in my head" or "my default assumption" when I want to refer to the things that come to mind automatically, have not undergone any math or rectification, and are especially prone to human bias.
3) I try to reserve phrases like "on priors" for when I think I could unpack my claim into a clear and principled picture if asked, especially one with rough degrees of credence/surprise. As far as I am aware, "prior" in this sense started out as a truncation of "prior probability".
I worry that "prior" will undergo semantic diffusion; more and more people will use it to make their claims sound cooler, when in fact they are actually doing (2). I think I have already contributed to this somewhat.
Though, "prior" as a noun is definitely an obscure term. I've hardly ever seen it used outside of rationalist discourse. I'm confident that most educated people know the word "prior," but only in the meaning of "previous." This claim could be tested with a mechanical turk survey.
“Everything should be made as simple as possible, but no simpler” (maybe Einstein…?).
And words have meaning. And it’s important to be precise. So it’s no problem, and in fact should be preferred, for writers to be precise in their writing, when that is what’s called for…even if that means using bigger or more complicated words. Of course, it should be geared to your audience as well, so you need to strike a balance btw reading the room Vs not always catering to the lowest common denominator.
“…but knew I had to get the paper up to a certain word count for it to be published.” The best and perhaps hardest writing is concise. Most nonfiction books and many academic papers could be much shorter (or not published at all). I blame the editors.
“A sentence should contain no unnecessary words, a paragraph no unnecessary sentences, for the same reason that a drawing should have no unnecessary lines and a machine no unnecessary parts.”
Each of your sentences is ungrammatical (thereby confusing) as it stands.
"prior" is adjectival / adverbial, not a noun (a prior), unless being used as an elided anaphoric reference .. eg, a prior (criminal conviction).
One "makes an assumption", one does not "have" it. Grammatically correct usage would be the plain "I assume... will be close"... the assumption points to a future event/state.
The recent podcast between Bill Maher and Bill Burr had some interesting moments on this topic, where Maher used fancy words and Burr wasn’t having any of it. How would you interpret those moments?
“Prior” can be a noun or an adjective. When I first read that sentence I assumed that “prior”was used as a modifier and I was looking for the noun. I was not assuming it was used as a noun. This is why I like assumption better. Assumption may not be the best word, but it is clearer within that particular sentence structure.
There is also a counter to sloppy writing and that is writing that mistakes sophistication with clear thought and understanding.
Then use "Bayesian prior" which is clear in the sense that it is precisely, mathematically, defined. Paul Graham once said that maths is the philosophy of precisely defined things, which I think is a good definition.
Hey Richard, thank you for your critique of my tweet. I actually agree with you that "prior" should sometimes be used instead of "assumption", and "stochastic" should sometimes be used instead of "random". If you read my tweet again, you'll see that I only advised against using the obscurer word in cases where you could just use the simpler word (i.e. where it wouldn't significantly change the meaning of the sentence).
I also agree with you that language isn't always about saying what you mean (though I think it should be). You say you want to signal your intelligence, and, well, you succeed in that -- I do find you intelligent -- not because you use clever words but because you use clear words to say things that are clever.
Thanks for the kind words. I'll say though that you can signal intelligence by both clear writing and obfuscating. It's actually hard to write like an academic, most people can't do it. Clear writing makes a person more understandable, but then the ideas are there to be judged on their own merits.
"A prior is something you have good reason to believe based on previous evidence. When you bring your prior with you to analyze a question, you are evaluating new data in light of it."
No. (Bayesian) prior implies no such thing. Prior can be based on anything - in statistics it is often a uniform distribution for a parameter, meaning think any value is equally likely. However, prior implies that you will be gathering evidence going forward and that you will change your prior accordingly. In fact, priors are often of little relevance because they are overridden by actual data. So when frequentists complain "so where do you get your priors", Bayesians tell them - "it doesn't matter because our conclusions are driven by the data, not priors". The point of Bayesianism is in the updating process, not in well thought out priors.
This whole post is about the idea that connotation is a thing that exists such that two words that superficially seem equivalent will put slightly different ideas in the reader's head.
The connotation of the word "prior" is that the person has put has put enough thought into their underlying assumptions that they could quantify their beliefs if asked. While it's true that nothing about the technical definition of prior requires that the prior be well-thought-out, there is still the connotation that "priors" have some rigor while "assumptions" do not. And that's all that matters.
There is the seperate interesting point that connotation depends very much on the person in question. For someone from a non-technical background, "prior" sounds mathy so it will be given an appropriate level of respect. But to someone who works formally in statistics, the word "prior" might have lost its glamour a long time ago such that there is no connotative difference between the word "prior" and the word "assumption". So this is also a lesson in knowing who your audience is. Since Richard writes mostly for smart laymen as opposed to statisticians, his point still holds.
The word "prior" never had any glamour. It is a much discussed weakness of the Bayesian approach, not the strength.
I mean, fine, as someone pointed out, people are now using word 'literally' to mean its exact opposite. So sure, use 'prior' to mean 'evidence based, well thought out assumption'. But perhaps at least don't rationalize this by supposed benefits that the use of technical language brings to the discussion?
Yes. Although his fundamental point - that there's a difference between what we can take away from someone writing "prior" rather than "assumption" - continues to hold. A prior implies a best-effort to obtain probabilities before further information is acquired, while an assumption is something much more likely to have had no thought put into it.
I agree that "prior" both denotes and connotes something rather different from "assumption", but I think Gurwinder may be reacting to widespread misuse of the former word. The rationalist sphere is overflowing with cargo cultists who love to dress up their perfectly ordinary human intuition (i.e. biases, spite and fundamental attribution errors) with a superficial veneer of Bayesian-style language. This was brought home to me by the recent kerfuffle over the EA startup Nonlinear (https://forum.effectivealtruism.org/posts/bwtpBFQXKaGxuic6Q/effective-aspersions-how-the-nonlinear-investigation-went), in which various figures involved were prone to framing their intuitive gut feelings as the product of careful, disinterested statistical reasoning: "But I think it would still have over a 40% chance of irreparably harming your relationship with Drew"; "Nonlinear's threatening to sue Lightcone for Ben's post is completely unacceptable, decreases my sympathy for them by about 98%". It's a linguistic device intended to give the appearance of solidity to pure wind, a verbal tic no different from the habitual misuse of the word "literally", or concluding a statement by tacking on the word "FACT" at the end in a tone of voice meant to sound definitive and unanswerable. What does it MEAN to say that doing such-and-such has a 40%+ chance of irreparably harming one's relationship with X? I'm in a rationalist WhatsApp group in which I saw someone use the phrase "I have pretty irrational priors on..." without so much as a glimmer of irony or self-awareness. When even the people using the word "prior" are openly admitting that their alleged "priors" are irrational, it seems reasonable to point out that "assumption" or "gut feeling" might capture the intended (or actual) sentiment better than "prior".
If you're discussing some empirical question, and you've done the work of digging through reams of data on said question before the discussion began, by all means use "prior". If you haven't (because you're lazy, or because no relevant data actually exists, or the question isn't really an empirical question at all) don't pretend you have - just be honest and say "assumption" or "gut feeling" or "intuition".
“ What does it MEAN to say that doing such-and-such has a 40%+ chance of irreparably harming one's relationship with X? ”
It means that if you were going to bet, you would buy at a price less than 40 sense on the dollar and not higher. Putting probability on events is good and more people should do it in more situations.
It doesn't mean "if you were going to bet...". It means that you actually took the bet. Personal probabilities are not based on introspection. Instead, bets you actually take reveal your degrees of belief. There is nothing magical or rational about merely framing your beliefs in probabilistic language.
I agree with Gurwinder. I agree with George Orwell’s advice in “Politics and the Abuse of the English Language”. The two example sentences are much better with non-fancy words. The fancy words cause unnecessary effort and add little necessary meaning. This isn’t fictional literature nor technical writing aimed at a specific audience.
In that essay, Orwell also says that sloppy imprecise writing is both a cause and a result of sloppy imprecise thinking.
Saying "assumption" when you mean "prior" is both sloppy and imprecise.
I worked with a guy once who had this self-improvement technique -- he'd learn a new word every day, and try to find ways to use that new word in a sentence. One of the senior trial lawyers in our firm said that young lawyers should instead try to forget one word every day.
I've also come around to 'prior' after reading SSC subreddit, and I agree that Bayesian prior sounds pretentious from most people, with the exception of Harry Potter Evans Verres.
But don't use it if you're not sure the person will understand. It is still jargon.
> Yet prior is not an obscure term. If you haven’t heard of it, you probably are not that well-read.
Well, I'd say that "prior" is well-known among people who have a grasp of the conceptual structure of statistics. But I'd also say that is less than 1% of the population. Where you place the cutoff for "well-read" may vary, though.
Lambasting people who use a fancy word for "random" is stochastic terrorism.
I recently noticed that I've been overusing the word "prior(s)", saying it automatically, even when it is not quite what I mean. I've been idly groping around for better phrasings.
1) I like to use "base rate" when I mean to refer to the actual frequency of something in real life, especially on topics wherein base rate fallacy lurks. https://en.wikipedia.org/wiki/Base_rate_fallacy
2) I'll sometimes refer to "the stereotype in my head" or "my default assumption" when I want to refer to the things that come to mind automatically, have not undergone any math or rectification, and are especially prone to human bias.
3) I try to reserve phrases like "on priors" for when I think I could unpack my claim into a clear and principled picture if asked, especially one with rough degrees of credence/surprise. As far as I am aware, "prior" in this sense started out as a truncation of "prior probability".
I worry that "prior" will undergo semantic diffusion; more and more people will use it to make their claims sound cooler, when in fact they are actually doing (2). I think I have already contributed to this somewhat.
https://www.jillesvangurp.com/blog/2006-12-15-semantic-diffusion.html
Strong points.
Though, "prior" as a noun is definitely an obscure term. I've hardly ever seen it used outside of rationalist discourse. I'm confident that most educated people know the word "prior," but only in the meaning of "previous." This claim could be tested with a mechanical turk survey.
“Everything should be made as simple as possible, but no simpler” (maybe Einstein…?).
And words have meaning. And it’s important to be precise. So it’s no problem, and in fact should be preferred, for writers to be precise in their writing, when that is what’s called for…even if that means using bigger or more complicated words. Of course, it should be geared to your audience as well, so you need to strike a balance btw reading the room Vs not always catering to the lowest common denominator.
"Prior is even shorter than assumption, so that takes away another argument against using the fancier word."
...and hunch is even shorter than prior (which makes another argument against using the goofier word)
Fewer syllables, anyway; same letter count.
“…but knew I had to get the paper up to a certain word count for it to be published.” The best and perhaps hardest writing is concise. Most nonfiction books and many academic papers could be much shorter (or not published at all). I blame the editors.
“A sentence should contain no unnecessary words, a paragraph no unnecessary sentences, for the same reason that a drawing should have no unnecessary lines and a machine no unnecessary parts.”
― William Strunk Jr., The Elements of Style
Each of your sentences is ungrammatical (thereby confusing) as it stands.
"prior" is adjectival / adverbial, not a noun (a prior), unless being used as an elided anaphoric reference .. eg, a prior (criminal conviction).
One "makes an assumption", one does not "have" it. Grammatically correct usage would be the plain "I assume... will be close"... the assumption points to a future event/state.
The recent podcast between Bill Maher and Bill Burr had some interesting moments on this topic, where Maher used fancy words and Burr wasn’t having any of it. How would you interpret those moments?
I agree: "prior" is more accurate, therefore better, than "assumption". It's more accurate.
“Prior” can be a noun or an adjective. When I first read that sentence I assumed that “prior”was used as a modifier and I was looking for the noun. I was not assuming it was used as a noun. This is why I like assumption better. Assumption may not be the best word, but it is clearer within that particular sentence structure.
There is also a counter to sloppy writing and that is writing that mistakes sophistication with clear thought and understanding.
Then use "Bayesian prior" which is clear in the sense that it is precisely, mathematically, defined. Paul Graham once said that maths is the philosophy of precisely defined things, which I think is a good definition.