Rationality and the English Language

The other day, some­one com­mented that my writ­ing re­minded them of Ge­orge Or­well’s “Poli­tics and the English Lan­guage.”1 I was hon­ored. Espe­cially since I’d already thought of to­day’s topic.

If you re­ally want an artist’s per­spec­tive on ra­tio­nal­ity, then read Or­well; he is manda­tory read­ing for ra­tio­nal­ists as well as au­thors. Or­well was not a sci­en­tist, but a writer; his tools were not num­bers, but words; his ad­ver­sary was not Na­ture, but hu­man evil. If you wish to im­prison peo­ple for years with­out trial, you must think of some other way to say it than “I’m go­ing to im­prison Mr. Jen­nings for years with­out trial.” You must muddy the listener’s think­ing, pre­vent clear images from out­rag­ing con­science. You say, “Un­re­li­able el­e­ments were sub­jected to an al­ter­na­tive jus­tice pro­cess.”

Or­well was the out­raged op­po­nent of to­tal­i­tar­i­anism and the muddy think­ing in which evil cloaks it­self—which is how Or­well’s writ­ings on lan­guage ended up as clas­sic ra­tio­nal­ist doc­u­ments on a level with Feyn­man, Sa­gan, or Dawk­ins.

“Writ­ers are told to avoid us­age of the pas­sive voice.” A ra­tio­nal­ist whose back­ground comes ex­clu­sively from sci­ence may fail to see the flaw in the pre­vi­ous sen­tence; but any­one who’s done a lit­tle writ­ing should see it right away. I wrote the sen­tence in the pas­sive voice, with­out tel­ling you who tells au­thors to avoid pas­sive voice. Pas­sive voice re­moves the ac­tor, leav­ing only the acted-upon. “Un­re­li­able el­e­ments were sub­jected to an al­ter­na­tive jus­tice pro­cess”—sub­jected by whom? What does an “al­ter­na­tive jus­tice pro­cess” do? With enough static noun phrases, you can keep any­thing un­pleas­ant from ac­tu­ally hap­pen­ing.

Jour­nal ar­ti­cles are of­ten writ­ten in pas­sive voice. (Par­don me, some sci­en­tists write their jour­nal ar­ti­cles in pas­sive voice. It’s not as if the ar­ti­cles are be­ing writ­ten by no one, with no one to blame.) It sounds more au­thor­i­ta­tive to say “The sub­jects were ad­ministered Pro­gen­i­torivox” than “I gave each col­lege stu­dent a bot­tle of 20 Pro­gen­i­torivox, and told them to take one ev­ery night un­til they were gone.” If you re­move the sci­en­tist from the de­scrip­tion, that leaves only the all-im­por­tant data. But in re­al­ity the sci­en­tist is there, and the sub­jects are col­lege stu­dents, and the Pro­gen­i­torivox wasn’t “ad­ministered” but handed over with in­struc­tions. Pas­sive voice ob­scures re­al­ity.

Judg­ing from the com­ments I get, some­one will protest that us­ing the pas­sive voice in a jour­nal ar­ti­cle is hardly a sin—af­ter all, if you think about it, you can re­al­ize the sci­en­tist is there. It doesn’t seem like a log­i­cal flaw. And this is why ra­tio­nal­ists need to read Or­well, not just Feyn­man or even Jaynes.

Non­fic­tion con­veys knowl­edge, fic­tion con­veys ex­pe­rience. Med­i­cal sci­ence can ex­trap­o­late what would hap­pen to a hu­man un­pro­tected in a vac­uum. Fic­tion can make you live through it.

Some ra­tio­nal­ists will try to an­a­lyze a mis­lead­ing phrase, try to see if there might pos­si­bly be any­thing mean­ingful to it, try to con­struct a log­i­cal in­ter­pre­ta­tion. They will be char­i­ta­ble, give the au­thor the benefit of the doubt. Authors, on the other hand, are trained not to give them­selves the benefit of the doubt. What­ever the au­di­ence thinks you said is what you said, whether you meant to say it or not; you can’t ar­gue with the au­di­ence no mat­ter how clever your jus­tifi­ca­tions.

A writer knows that read­ers will not stop for a minute to think. A fic­tional ex­pe­rience is a con­tin­u­ous stream of first im­pres­sions. A writer-ra­tio­nal­ist pays at­ten­tion to the ex­pe­rience words cre­ate. If you are eval­u­at­ing the pub­lic ra­tio­nal­ity of a state­ment, and you an­a­lyze the words de­liber­a­tively, rephras­ing propo­si­tions, try­ing out differ­ent mean­ings, search­ing for nuggets of truthi­ness, then you’re los­ing track of the first im­pres­sion—what the au­di­ence sees, or rather feels.

A nov­el­ist would no­tice the scream­ing wrong­ness of “The sub­jects were ad­ministered Pro­gen­i­torivox.” What life is here for a reader to live? This sen­tence cre­ates a dis­tant feel­ing of au­thor­i­ta­tive­ness, and that’s all—the only ex­pe­rience is the feel­ing of be­ing told some­thing re­li­able. A nov­el­ist would see nouns too ab­stract to show what ac­tu­ally hap­pened—the post­doc with the bot­tle in their hand, try­ing to look stern; the stu­dent listen­ing with a ner­vous grin.

My point is not to say that jour­nal ar­ti­cles should be writ­ten like nov­els, but that a ra­tio­nal­ist should be­come con­sciously aware of the ex­pe­riences which words cre­ate. A ra­tio­nal­ist must un­der­stand the mind and how to op­er­ate it. That in­cludes the stream of con­scious­ness, the part of your­self that un­folds in lan­guage. A ra­tio­nal­ist must be­come con­sciously aware of the ac­tual, ex­pe­ri­en­tial im­pact of phrases, be­yond their mere propo­si­tional se­man­tics.2

Or to say it more bluntly: Mean­ing does not ex­cuse im­pact!

I don’t care what ra­tio­nal in­ter­pre­ta­tion you can con­struct for an ap­plause light like “AI should be de­vel­oped through demo­cratic pro­cesses.” That can­not ex­cuse its ir­ra­tional im­pact of sig­nal­ing the au­di­ence to ap­plaud, not to men­tion its cloudy ques­tion-beg­ging vague­ness.

Here is Or­well, railing against the im­pact of cliches, their effect on the ex­pe­rience of think­ing:

When one watches some tired hack on the plat­form me­chan­i­cally re­peat­ing the fa­mil­iar phrases—bes­tial , atroc­i­ties , iron heel, blood­stained tyranny, free peo­ples of the world, stand shoulder to shoulder—one of­ten has a cu­ri­ous feel­ing that one is not watch­ing a live hu­man be­ing but some kind of dummy . . . A speaker who uses that kind of phrase­ol­ogy has gone some dis­tance to­ward turn­ing him­self into a ma­chine. The ap­pro­pri­ate noises are com­ing out of his lar­ynx, but his brain is not in­volved, as it would be if he were choos­ing his words for him­self . . .

What is above all needed is to let the mean­ing choose the word, and not the other way around. In prose, the worst thing one can do with words is sur­ren­der to them. When you think of a con­crete ob­ject, you think word­lessly, and then, if you want to de­scribe the thing you have been vi­su­al­is­ing you prob­a­bly hunt about un­til you find the ex­act words that seem to fit it. When you think of some­thing ab­stract you are more in­clined to use words from the start, and un­less you make a con­scious effort to pre­vent it, the ex­ist­ing di­alect will come rush­ing in and do the job for you, at the ex­pense of blur­ring or even chang­ing your mean­ing. Prob­a­bly it is bet­ter to put off us­ing words as long as pos­si­ble and get one’s mean­ing as clear as one can through pic­tures and sen­sa­tions.

Charles San­ders Peirce might have writ­ten that last para­graph. More than one path can lead to the Way.

1Com­ment at http://​​less­wrong.com/​​lw/​​jb/​​ap­plause_lights/​​f1t.

2Com­pare “Se­man­tic Stop­signs” and “Ap­plause Lights” in Map and Ter­ri­tory.