"The reason LLMs use the same cadence and cliches is that their RLHF does not emphasize writing well, the way it emphasizes, say, coding. And the reason for that should be pretty obvious: coding is where the big money is at."
Software can be checked for being 'written well' by compilers / linters etc. There is no equivalent for well-written natural prose. Spelling and grammar checkers haven't a clue about prose semantics.
I looked 'squeeze' up in the (paywalled) Oxford English Dictionary [0]. The are numerous variant forms including 'squoze' and 'squozen':
"Also 1600s squeez, squeaze, squease, squese, 1600s–1700s squeese. Also with dialect preterite and past participle 1800s– squoze, past participle 1800s squozen."
The OED cites quotations from the 1930s, 1870s, and 1860s that include 'squoze' and 'squozen', for example ‘Orange?’ repeated Waitress No. 1. ‘Do you want it squoze?’ in the Baltimore Sun, 1 Sep 1931.
I think the cherry-picked examples at the beginning completely ignore the massive number of examples throughout history where groups of people pushed back against mainstream society. A lot of them were hurt or died in the process but societies throughout history have been changed by people who refused to fit in.
The best strategy for the majority is homogeny, right? Just go with what others say to do 100% of the time offering no input yourself until the distribution of people becomes homogenous.
In my defence, I was coming at it etymologically, where (according to the OED), a perambulator is 'A person who walks or travels through or about a place' (hence 'walker') OR 'A carriage for a baby or young child, which may be pushed along by a person on foot. Now usually shortened to pram [...] a cradle mounted on four wheels'.
I can see his point that it is worth learning to think for yourself, but if you then try to publish your thoughts without understanding what else is out there (e.g. by reading about it), the results could be very embarrassing. This would seem to apply in both philosophy (which I'm assuming Schopenhauer was discussing) and (say) genre fiction.
Incidentally, having quickly scanned some of Schopenhauer's actual writing [0], I found statements such as "women are suited to being nurses and teachers of our earliest childhood precisely because they themselves are childish, silly and short-sighted" and women "are inferior to men in respect of justice, honesty and conscientiousness". If I'd come across these first I'd be very reluctant, rightly or wrongly, to uncritically accept his views on other subjects.
> If I'd come across these first I'd be very reluctant, rightly or wrongly, to uncritically accept his views on other subjects.
You don't want people to use this sort of criteria to judge arguments. Not just because it's ad hominem, but also for selfish reasons - your opinion that after a "quick scan" one should dismiss Schopenhauer's philosophical thought (his specialty) because he shared a common contemporary opinion on the nature of women (not his specialty) reflects badly on you. If somebody could reasonably dismiss an unrelated argument that you made based on that, your arguments would be ignored regardless of their merit.
Also, people would wonder if you have dismissed all of the writings and discoveries of sexists and racists; if you have then you are obviously uneducated, and if you haven't then your exceptions imply the shape of your own specific racism and sexism. It's a no-win situation for you.
My 'quick scan' involved reading the essay "On thinking for yourself" which was quoted from above. I specifically disagree with his general point here, essentially that you should avoid reading until your own ideas are fully developed. I do not believe this is universally true, especially in the context of this discussion, which is about becoming a better writer.
To get a feel for his other work, I read his essay "On Women", which is derogatory and sexist. Female philosophers such as Mary Wollstonecraft had already responded to similar patriarchal sexism (50 years before in her case) and as a philosopher one might have expected Schopenhauer to be familiar with similar intelligent women, and to have acknowledged their existence. His denigration of the capabilities of half the human race does not give me confidence in his other speculations.
[Edit] I also observe that Tolstoy stated "at present I am convinced that Schopenhauer is the greatest genius among men. ... It is the whole world in an incomparably beautiful and clear reflection" [0]. As someone who has slogged through the 27 chapters of philosophy in War and Peace, which are a case study of how to fail to understand how human society actually works, I am reluctant to uncritically accept claims from famous people that someone else is a genius.
Sydney Padua's awesome, mostly true, steampunk graphic novel "The Thrilling Adventures of Lovelace and Babbage" jokes about the error messages that might have produced by the difference engine. One example occurs when a character tries to stop a sequence that may or may not destroy a precious manuscript:
Annulation lever activated!
Do you wish to continue?
Engage STOP lever to continue. Engage CONTINUE lever to stop.
You'd also have things like bus doors now being on the wrong side (traffic-side), road curvature being wrong (e.g. motorway junctions and entry/exit ramps), let alone car mirrors and driver positioning.
Thiss is a very good example. Consider the difference between "he's hard working" and "he's hard work". My wife ran into this when some German co-workers were describing a colleague.
This also doesnt touch on translation also having an element of the audience in it as well. You see this in debates on what the best translation of the Odyssey is, with some preferring one version for being more "classical" in language and others preferring a more modern take. I would argue they can both be right, as its perfectly valid to choose turns of phrase that the intended audience will understand vs those which they wont. Language evolves over time.
Software can be checked for being 'written well' by compilers / linters etc. There is no equivalent for well-written natural prose. Spelling and grammar checkers haven't a clue about prose semantics.
reply