Suppose we lived in this completely alternate universe where nothing in academia was about status, and no one had any concept of style.  A universe where people wrote journal articles, and editors approved them, without the tiniest shred of concern for what "impression" it gave - without trying to look serious or solemn or sophisticated, and without being afraid of looking silly or even stupid.  We shall even suppose that readers, correspondingly, have no such impressions.

In this simpler world, academics write papers from only two possible motives:

First, they may have some theory of which they desire to persuade others; this theory may or may not be true, and may or may not be believed for virtuous reasons or with very strong confidence, but the writer of the paper desires to gain adherents for it.

Second, there will be those who write with an utterly pure and virtuous love of the truthfinding process; they desire solely to give people more unfiltered evidence and to see evidence correctly added up, without a shred of attachment to their or anyone else's theory.

People in the first group may want to signal membership in the second group, but people in the second group only want their readers to be well-informed.  In any case, to first order we must suppose that none of this is about signaling - that all such motives are just blanked out.

What do journal articles in this world look like, and how do the Persuaders' articles differ from the Informers'?

First, I would argue that both groups write much less formal journal articles than our own.  I've read probably around a hundred books on writing (they're addictive); and they all treated formality as entropy to be fought - a state of disorder into which writing slides.  It is easier to use big words than small words, easier to be abstract than concrete, easier to use passive -ation words than their active counterparts.  Perhaps formality first became associated with Authority, back in the dawn of time, because Authorities put in less effort and forced their audience to read anyway.  Formality became associated with Wisdom by being hard to understand.  Why suppose that scientific formality was ever about preventing propaganda?

Both groups still use technical language, because they both care about being precise.  They even use big words or phrases for their technical concepts:  To carve out ever-finer slices through reality, you need new words, more words, hence bigger words (so you don't run out of namespace).

However, since neither group has a care for their image, they use the simplest words they can apart from that, and sentences as easy as possible apart from the big words.  From our standpoint, their style would seem inconsistent, discongruous.  A sentence might start with small words that just anyone could read, and then terminate in an exceptionally precise structure of technically sophisticated concepts accessible to only advanced audiences.

In this world it's not just eminent physicists who - secure in their reputation as Real Scientists - invent labels like "gluon", "quark", "black hole" and "Big Bang".

Other aspects of scientific taboo may still carry over.  A Persuader might use vicious insults and character assassination.  An Informer never would.  But an Informer might point out - evenhandedly, wherever it happened to be true - that a supposedly relevant paper came from a small unheard-of organization and hadn't yet been replicated, or that the author of an exciting new paper had previously retracted other results...

If Persuaders want to look like Informers, they will, of course, restrain their ad-hominem attacks to sounding like the sort of things an Informer might point out; but this is a second-order phenomenon.  First-order Persuaders would use all-out invective against their opponents.

What about emotions in general?

Suppose that there were only Informers and that they weren't concerned about preventing invasion by Persuaders.  The Informers might well make a value-laden remark or two in the conclusions of their papers - after balancing the probability that the conclusion would later need to be retracted and that the emotion might interfere, versus the importance of the values in question.  Even an Informer might say, in the conclusion of a paper on asteroid strikes, "We can probably breathe a sigh of relief about getting hit in the immediate future, but when you consider the sheer size of the catastrophe and the millions of dead and injured, we really ought to have a spacewatch program."

But Persuaders have an immensely stronger first-order drive to invoke powerful affective emotions and lade the reader's judgments with value.  To second order, Persuaders will try to disguise this method as much as possible - let the reader draw conclusions, so long as they're the desired conclusions - try to pretend to abstract dispassionate language so that they can look like Informers, while still lingering on the emotion-arousing facts.  Formality is a very easy disguise to wear, which is one reason I give it so little credit.

Informers, who have no desire to look like Informers, might go ahead and leave in a value judgment or two that seemed really unlikely to interfere with anyone's totting up the evidence.  If Informers trusted their own judgment about that sort of thing, that is.

(Persuaders and Informers writing about policy arguments or moral arguments would be a whole 'nother class of issue.  Then both types are offering value-laden arguments and dealing in facts that trigger emotions, and the question is who's collecting them evenhandedly versus lopsidedly.)

How about writing short stories?

Persuaders obviously have a motive to do it.  Do Informers ever do it, if they're not worried about looking like Persuaders?

If you try to blank out the conventions of our own world, and imagine what would really be useful...

Then I can well imagine that it would be de rigueur to write small stories - story fragments, rather - describing the elements of your experimental procedure.

"The subjects were admininistered Progenitorivox" actually gives you very little information, just the dull sensation of having been told an authoritative fact.

Compare:  "James is one of my typical subjects.  Every Wednesday, he would visit me in my lab at 2pm, and, grimacing, swallow down two yellow pills from his bottle, while I watched.  At the end of the study, I watched James and the other students file into the classroom, sit down, and fill out the surveys on each desk; as they left, I gave each of them a check for $50."

This, which conveys something of the experience of running the experiment and just begs you to go out and do your own... also gives you valuable information: that the Progenitorivox or placebo was taken at regular intervals with the investigator watching, and when and where and how the survey data was collected.

Maybe this is the most efficient way to communicate that information, and maybe not.  To me it actually does seem efficient, and I would guess that the only reason people don't do this more often is that they would look insufficiently Distant and Authoritative.  I have no trouble imagining an Informer writing a story fragment or two into their journal article.

Robin says:  "I thus tend to avoid emotion, color, flash, stories, vagueness, repetition, rambling, and even eloquence."

I would guess that, to first order and before all signaling:

Persuaders actively seek out emotion, color, flash, and eloquence.  They are vague when they have something to hide.  They rehearse their favored arguments, but not to where it becomes annoying.  They try to avoid rambling because no one wants to read that.  They use stories where they expect stories to be persuasive - which, by default, they are - and avoid stories where they don't want their readers visualizing things in too much detail.

Informers avoid emotions that they fear may bias their readers.  If they can't actually avoid the emotion - e.g., the paper is about slavery - they'll explicitly point it out, along with its potential biasing effect, and they'll go to whatever lengths they can to avoid provoking or inflaming the emotion further (short of actually obscuring the subject matter).  Informers may use color to highlight the most important parts of their article, but won't usually give extra color to a single piece of specific evidence.  Informers have no use for flash.  They won't avoid being eloquent when their discussion happens to have an elegant logical structure.  Informers use the appropriate level of abstraction or maybe a little lower; they are vague when details are unknowable or almost certainly irrelevant.  Informers don't rehearse evidence, but they might find it useful to repeat some details of the experimental procedure.  Informers use stories when they have an important experience to communicate, or when a set of details is most easily conveyed in story form.  In papers that are about judgments of simple fact, Informers never use a story to arouse emotion.

I finally note, with regret, that in a world containing Persuaders, it may make sense for a second-order Informer to be deliberately eloquent if the issue has already been obscured by an eloquent Persuader - just exactly as elegant as the previous Persuader, no more, no less.  It's a pity that this wonderful excuse exists, but in the real world, well...

21 comments, sorted by
magical algorithm
Highlighting new comments since Today at 7:08 PM
Select new highlight date
Moderation Guidelines: Reign of Terror - I delete anything I judge to be annoying or counterproductiveexpand_more

Actually, there does seem to be at least one area of academia that at least partially looks kinda like this. Specifically, crypto research. I have an on and off interest in it, so have read a couple crypto papers and descriptions here and there, and many do seem to take a combination informal (even playful) and technical and formal mixture style of writing.

I mean, heck, think of even the way algorithms are discussed, the whole cast of characters... Alice and Bob, Eve the passive eavesdropper, malicious Mallory who manipulates the messages, and so on.

If most writers are using long words and convoluted sentences to convey authority, then a high-status academic could deliberately use simple language, using the handicap principle to convey "people take me seriously as an authority even though I use simple language".

And if lots of people start doing this, others might imitate them.

Eliezer, why so reluctant to analyze an actual equilibrium, rather than first order strategies ignoring so many important effects? My claims were about real equilibrium behavior, not some hypothetical world of clueless caricatures. And why so emphasize a few "writing" experts you've read over vast numbers of teachers of writing styles in law, engineering, accounting, academia, etc.?

Psy-Kosh, this is a fascinating point if true, considering that cryptography is a field with

(a) math with proofs

(b) lots of people trying to poke holes in the math

(c) strong real-world consequences

Compare: "James is one of my typical subjects. Every Wednesday, he would visit me in my lab at 2pm, and, grimacing, swallow down two yellow pills from his bottle, while I watched. At the end of the study, I watched James and the other students file into the classroom, sit down, and fill out the surveys on each desk; as they left, I gave each of them a check for $50."

I'm under the impression that old scientific papers were more like this. See e.g. Millikan (1926) on cosmic rays:

We chose for the first experiments Muir Lake (11,800 feet high), just under the brow of Mount Whitney, the highest peak in the United States, a beautiful snow-fed lake hundreds of feet deep and some 2000 feet in diameter. Here we worked for the last ten days in August, sinking our electroscopes to various depths down to 67 feet. Our experiments brought to light altogether unambiguously a radiation of such extraordinary pene- trating power that the electroscope-readings kept decreasing down to a depth of 50 feet below the surface. The atmosphere above the lake was equivalent in absorbing power to 23 feet of water, so that here were rays so penetrating that, if they came from outside the atmosphere, they had the power of pass- ing through 50 + 23 = 73 feet of water, or the equivalent of 6 feet of lead, before being completely absorbed. The most penetrating X-rays that we produce in our hospitals cannot go through half an inch of lead. Here were rays at least a hundred times more penetrating than these, and having an absorption coefficient but one twenty-fifth, instead of "about one-tenth of that of the hardest known gamma rays."8

(Emphasis as in the original.)

Another possibly relevant aspect of crypto is that it is fantastically, painfully difficult; due partly to the real world consequences, and partly to the inherent challenge of the maths involved.

As a first hypothesis, perhaps the degree of difficulty makes the entire field exclusive enough that people have less need to hide behind obfuscating verbiage and let their ability to handle the maths speak for itself; i.e., they feel they have less to prove, in the same manner as high-status academics in other fields.

This would imply some degree of inverse correlation between the inherent rigor and difficulty of a field and the formality expected from the average researcher. Are, say, physicists or engineers typically more informal than, say, psychologists or economists? (disclaimer: this is not an attack on practicioners of the latter fields, just an observation that they are more removed from easily tested real world consequences, meaning it's easier to bluff)

As I understood this post, there is an argument that:

  • posters who want to signal will use formality
  • only persuaders want to signal
  • the reader should be suspicious of anyone who writes formally

In general, IAWYC, but I think that we may be ignoring the fact that signal-recognition has legitimate uses. Standard example: if I read Eliezer's blog, I know in advance that he is a legitimate source of rational, logical arguments. If I read xxyy's blog (assuming this is not peer-reviewed, etc), I have no idea - given the volume of data published online - whether what I'm reading is in any way factual. Yes, I should analyze content, read reviews, and track down references, and such. But, this is very time consuming. It is reasonable to have some pre-selection mechanism.

If this pre-selection mechanism happens to be something as silly formal word-choice, the "right" font, or using some keyword within the first few paragraphs of a paper, then so be it. To clarify: signal-recognition is weak evidence that the paper is worth reading.

I think some of the comments act as though only things written by name-brand authors can be trusted. This is false: new science is done by new scientists every day. But, the presence of new scientists implies that there must be a signaling mechanism, and that this signaling mechanism cannot be morally wrong.

We shouldn't conclude that because someone uses signaling ("formality"/"expert style"/"eloquence"/etc) that they are attempting to persuade readers of their message. They may simply be trying to pass the first gate to readership.

Is it possible that (counter-intuitively) by attempting to disband recognized signals, we are simply being elitist?

I'm pretty sure Eliezer is familiar with 4chan, given his explicit reference to it a couple weeks ago...

The weird hive mind culture and memetic cesspit that is /b/ is actually interesting as well, in its own way, if one can tolerate looking at it (I heard that "lieking mudkipz" helps).

There seems to be an implicit idea here that being a persuader is a bad thing.

But even the informer is persuading people of something.

After all, to communicate at all, you MUST induce some sort of state transition in the recipient's brain.

And both the informers and the persuaders in your presentation are attempting to induce such a state transition; they differ only in what state they're attempting to induce, who they're trying to induce it in, or how effective they're willing to be at inducing that transition.

If the informer is uncertain and wishes to convey that uncertainty, great! Then they should use every available persuasive tool to persuade people to be uncertain! (As opposed to half-heartedly persuading them to be certain.)

However, if they are NOT uncertain, but are instead just trying to be "fair" or "evenhanded", then they're wasting their time on status-signalling.

Of course, it doesn't feel like status signaling, it just feels like it would be "unfair" of them to "trick" people into agreeing... because people should just "rationally" end up agreeing with them, not be persuaded.

In other words, the problem is viewing persuasion as evil from the outset... which then leads to conscientious (i.e. low-status!) people bending over backwards not to do it.

Thus, only high-status people end up persuading. And those who persuade, end up high-status. I expect that this isn't a coincidence: persuasion works better from high-status to low-status, and low status probably inhibits persuasive ability (under the guise of being "fair" or "not manipulative"). In order to become more effective at persuasion, I had to de-inhibit myself from assuming status in places where I didn't previously have it.

So, my suggestion: if you're in the least bit worried about whether you should use every possible tool of persuasion at your disposal... don't. It's merely an indication that you haven't updated your self-perception of status!

(Mind you, it's best to go the opposite way when taking information IN... because high status also inhibits the intake of contrary information.)

Two other examples of sciences where the vocabulary is less formal than is typical: Astronomy: "star", "black hole" (as compared to e.g. deoxyribonucleic acid) Genetics: names of genes have included "fruity", "shaven baby" and "killer of prune"

, it may make sense for a second-order Informer to be deliberately eloquent if the issue has already been obscured by an eloquent Persuader

If you actually want to inform, being deliberately eloquent helps the information stick, helps to inform, so even without Persuaders in the world, you should still wish to be eloquent.

The "completely alternate universe where nothing is about status" is called 4chan (in b4: 4chan != /b/, also other chans), you can observe thousands of people posting on various subjects, and one thing they don't care about is social status because they're all anonymous (other than tripfags, pseudonimity combines worst of both worlds) so you cannot even have social status on chans.

You should try it Eliezer, full anonymity removes many of the biases you see in every other place in the world. Certainly chans have all kinds of other problems, but different a very bias profile is something that you might appreciate. Some of it is obvious, some effects are rather subtle and take time and participation to notice and appreciate. One thing I noticed is that people weren't significantly less truthful on chans - there is a total lack of concern for truth so outright making things up is a fair game, but all the coloring of reality to protect social status goes away, I think it pretty much evens out.

Go lurk and post a bit, not because chans are a better mode of communication, but because they're very different in a way that you will have a hard time finding anywhere else.

Why wouldn't the Informers inform the public what he Persuaders are trying to do (i.e. they're not providing unbiased information, they just want you to believe them)?

Paul Crowley:

One trivial example of signalling here is the way everyone still uses the Computer Modern font. This is a terrible font, and it's trivial to improve the readability of your paper by using, say, Times New Roman instead, but Computer Modern says that you're a serious academic in a formal field.
I don't think that these people are signaling. Computer Modern is the default font for LaTeX. Learning how to change a default setting in LaTeX is always non-trivial.

You might argue that people are signaling by using LaTeX instead of Word or whatever, but switching from LaTeX to some other writing system is also not a trivial matter.

In case anybody was wondering, just add this to your LaTeX file:


It's actually pretty trivial. I always do this, though it took me a while. Also, in the papers I see online, they've usually done this.

I've read countless papers on crypto, and they mostly seem pretty formal to me - what are people comparing them to? Is it really worse in other fields? There is some variation - DJB's style is distinctly less formal than other authors - but my perception is that papers on for example network engineering seem a lot less formal than crypto papers. I think there's plenty of room to improve the readability of crypto papers by encouraging less formality.

One trivial example of signalling here is the way everyone still uses the Computer Modern font. This is a terrible font, and it's trivial to improve the readability of your paper by using, say, Times New Roman instead, but Computer Modern says that you're a serious academic in a formal field.

Maybe part of why crypto is less formal is that it's a new field, and formality takes time to build up.

As for the general point, Root-Bernstein talks about how scientists who make major breakthroughs are fascinated by their instruments. I bet that Informers will write more-- and more clearly-- about why they chose a hypothesis, how they chose to do a particular experiment, and the physical and theoretical obstacles to getting a conclusive result.

As for eloquence, if I'm right that the researcher's mental states are a part of the story, then there's room for Informers to use language which is vivid as well as clear.


there will be those who write with an utterly pure and virtuous love of the truthfinding process; they desire solely to give people more unfiltered evidence and to see evidence correctly added up, without a shred of attachment to their or anyone else's theory.
They're implicitly attached to the theory that this process really does find the truth, and they may be attached to the idea that it is the best or one of the best processes for doing so. On a slightly more abstract level, is there a difference between Informers and Persuaders?

For example, a Keynesian and an Austrian economist may not be at all attached to their theories, but are attached to their very different truth-seeking methodologies (though perhaps the term 'truth-seeking' is giving them too much credit).

As an aside, I find Robin's posts to be much easier to understand and follow than Eliezer's, and I don't think this has anything to do with the complexity of the arguments. Robin's style seems to just be simpler and more concise, making it easier for me to spot the premises and logic of his arguments. I think this is a benefit of more formal types of arguments in general, at least to my brain.

nazgulnarsil: the easiest way to spot scientism is to look for value statements being conflated with factual statements

And if you read something where you can't tell whether the writer is trying to make a value statement or a factual statement, the writer probably doesn't appreciate the difference between the two.

Eliezer: I'm not saying it's perfect informers, but plenty (not all, but plenty (of that which I've seen, at least)) of the writings include a more informal and playful "feel" to them.

Really, near as I can make out, it relates to the historical facts with regards to public crypto research having an almost "naughty" feel to it, specifically with gov'ts trying to originally discourage that sort of work, etc. But for whatever the reason, the net effect is, well, as I said, the standard "cast of characters" for one thing.

Some other examples may be, say, an annonymous communication protocal that is presented in a paper called "The Dining Cryptographers", and a later version that is meant to be resistant to interference/noise/troublemakers called "The Dining Cryptographers in the Disco"

And the implementations of these protocols are actually referred to as DC nets, which can either mean, well, the title, or be a reference to the author, David Chaum.

Oh, I just remembered something. Now this I have rather less familiarity with, but I seem to recall hearing or seeing mentioned somewhere that graph theory writings tend to have a different feel, that graph theory proofs tend to feel more like "stories being told" as opposed to formal proofs. But this is just something I've heard, and so I don't really know how true that is.

yes, the easiest way to spot scientism is to look for value statements being conflated with factual statements. This is done unintentionally in many cases, the persuaders can't help it because they can't distinguish between the two. 1) you falsify the data that someone thought was factual that they used to support their values. They take this as an attack on said values. 2) you point out errors in the train of logic between factual statements and values, and/or point out that there is no valid logic train between their values and facts. 3) you make a factual statements and it is confused for a value statement. This happens because we're taught to value truth and this valuation occasionally glitches. People assume that because you say something is true that you are also saying that it is good. 4) vice-versa of the above. you make a value statement and people take it as a factual statement. this is the goal of a persuader.

I'm sure there are other common examples.