‘Stock phrases’, in the sense I am using it here, refers to established phrases (in the more common, more specific sense), noises, gestures, etc.; they form a canon of well-known signifiers for messages one might want to convey, like the verbalisation ‘I am happy’, or the gesture of nodding in agreement. They can be very useful, because they save communicators the time, effort, and distraction of forming descriptions from existing phrases. Sometimes a stock phrase has been honed so finely that to try to recreate its precise meaning from scratch would not be possible in any practical period of time. As with language in general, novel or less common combinations of stock phrases are more liable to be misinterpreted. (For example, winks, nods, and other individual gestures are generally less ambiguous than chains of gestures.)

To put it another way: Compression is useful because some amount of upfront time and effort (learning meanings of stock phrases) can save a lot of time and effort later (having to construct new stock phrases repeatedly from scratch).

Two considerations that arise from this are over-reliance on the existing canon of stock phrases, and the skill of originating successful new stock phrases. 

With the former, stock phrases are used even in situations where it would be better to construct a phrase not already in the canon. It is very tempting to round off a complex sentiment into the nearest available stock phrase, because it is so much more convenient—they are available. For example, saying ‘I’m an atheist’ can be a lot more convenient than saying, ‘I put an effectively-zero, but non-zero, probability on the existence of God’. And in some contexts, the former might genuinely be just a useful approximation. But in some other contexts, it can lead to spending an hour arguing with someone before they realise that you don’t rule out God entirely like they were arguing against, and you realise that they have been disagreeing with you ruling out God entirely, rather than you not believing in God with high probability. (Of course, this might not mean the argument is over since there will probably be remaining disagreement. But it might shorten the argument by a frustrating hour.)

Over-reliance on stock phrases can also not only fail to communicate to others, but actually alter the shape one’s own aliefs or beliefs actually take. For example, identifying oneself with a label as a convenience, when one does not actually endorse all the implications of that label, can cause one to begin to advocate for those other implications, even if one did not originally. “I’m an X now, guess I have to believe Y/advocate for Z.” Sometimes this is to avoid censure by other people who identify with that label, whose approval one desires, and this might be a stable decision under reflection. But sometimes it’s as simple and undesirable as the social anxiety of, “If I stop using this label because it doesn’t describe me well, then people might point and laugh at me for seeming to change my mind.”

Originating successful stock phrases is important because of how dependent we are on them—as we should be. Neither extreme—doing everything from scratch on the spot, nor only using the most common stock phrases in the canon—is best; the optimum lies in between these extremes. Therefore we must depend on stock phrases to some extent, and moreover we need to depend on them often enough that we should get good at creating new ones to suit our circumstances, and ensuring that they spread to the relevant people with whom we shall need to use them.

Some things that help:

(1) Training the skill of noticing similarities between attempts to communicate, so that opportunities to generalise a new stock phrase are not missed. A common cue for this would be a feeling of dissatisfaction or frustration that one had not communicated exactly what one had meant and had been misunderstood, and the feeling of ‘I feel like there is a general meaning or class of experience here that I have in mind, but the other party does not realise this, but until I point them to it, we are kind of talking past each other.’

(2) Getting good at coming to catchy, memorable phrases or names. This need not be a solo effort; seeking others’ assistance or going to people who are particularly good at this are also options. Should we have a Phrase Lab here where we can post requests for assistance propagating useful phrases? Vote here!

(3) Surrounding oneself with or having access to people who are good at absorbing, using, and propagating useful phrases. Or at least avoiding people who are actively bad at these things; some people are scornful of new phrases (possibly a status thing; originating widely-used phrases gains status, so endorsing or using a phrase can feel like someone gains status relative to onself), and some people are snobbish prescriptivists (again partly a status thing) and will shoot down novel suggestions on principle. I suspect that an underestimated factor in the Bay Area success story is the unusually high openness to phrases and jargon, which allow deeper exploration of ideas and systems than the more general population’s stock phrases allow.

(4) Related to the above, but worth stating standalone: Surrounding oneself with or having access to people who are good at telling you when your phrases are good, and also when they’re crap. It is good to be motivated when you do good and notice useful categories or clusters, and also good to be warned when you are crystallizing a disuseful patterns. Similarly, people who are willing to say, ‘I think this phrase has made everything look like nail. We should reconsider our usage of it,’ once a phrase has taken off are to be valued.

(Related to my comment on (3): Although there are other factors in the gap, people perhaps underestimate how much of the gap between, say, Eliezer and Yvain and the average LessWronger-who-is-not-a-LessWrong-celebrity comes from their ability to crystallize, describe, and promote useful phrases. For those of us who are not so good at doing all three in one go and need more assistance, LessWrong could probably be more welcoming in seeking assistance or feedback on not-yet-complete phrases or crystallizations. This might not seem like a big advantage, but bear in mind that intelligence correlates very, very well with manipulating patterns, which is what phrases help with, and that while the leverage of using a phrase once is not very high, two or three decades of iterative use of phrases and the resulting positive feedback loop might explain more of the gap than one might initially think.)


11 comments, sorted by Click to highlight new comments since: Today at 8:50 AM
New Comment

From time to time I hear LW or the x-rationalist culture criticised for overusing jargon. It is true that, all else being equal, introducing jargon that creates barriers to entry is undesirable. But generally not all else is equal, and it raises a red flag to me that those who make this criticism rarely give examples of bad core LW jargon or do much beyond pay lipservice to the potential benefits of jargon. It is easy to appear Wise and Above It All by alluding to something like, “Ha ha! Silly nerds! They overuse jargon and alienate people,” or a contrarian “Here’s something LW does that is unusual, and the broader effects of which they’re oblivious to, and this is a stereotypical failure mode,” without actually addressing the object-level analysis of whether the alleged problem is actually a net-negative thing or where its optimal amount of usage lies. This isn’t necessarily what’s taking place, but the aforementioned red flag is evidence of this.

Think about referential distance. If you have a choice between a common-speech phrase and a jargon one, the common-speech one is probably a better choice: it will be understood by a broader audience, and will better serve to spread useful ideas into the mainstream — to raise the sanity waterline, as Eliezer puts it.

The expression, "When all you have is a hammer, everything looks like a nail" is a reasonably common expression. So it is probably fine to refer to it, at least to a technically-educated English-speaking audience. (Although it sounds like it might be ancient, it was apparently coined by mid-20th-century psychologist Abraham Maslow — yeah, the "hierarchy of needs" guy.)

In the Berlin Community weekend I noticed one instance of bad phrasing. System I and System II are bad names for cognitive processes. The numbering is arbitrary and there a good chance that we could find a way to label the systems that more accessible.

It's not quite arbitrary. 1 < 2, and system 1 is more fundamental than system 2. Animals have a system 1 but no system 2.

Still, agree that the names are bad. You shouldn't need to think about the rule to tell them apart.

They roughly correspond to subconscious and conscious, not sure why Kahneman felt he needed separate connotation-free names.

There are only two hard things in Computer Science: cache invalidation and naming things.

-- Phil Karlton

There are often plenty of different words that one can use with different connotation.

A lawyer works in a firm. Google is a company. Then we have words like business and enterprise. Shops and startups.

Should we have a thread for useful crystallizations to be honed into fully-fledged memes?


What would it be called? Phrase lab? Meme lab?

It might be better to do this organically & continuously, rather than in a one-time thread. That is, we could keep an eye out for reasoning-related patterns that haven't been labelled yet, and then coin names for them as & when we notice them (maybe here, maybe in the Open Thread, wherever).

(As it happens, I have a couple of such patterns I've thought about posting about, but I see no point in writing posts until I have clear & uncontroversial examples to illustrate them. Though I did post an RQ about one of them.)

Milhouse Lab, obv.