Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

New accounts on HN more likely to use em-dashes

marginalia.nu

Post reply on HN
Print view View original

81–90 of 643 posts

Re: New accounts on HN more likely to use em-dashes

#81

Post by d4mi3n »

asplake wrote: post #72
d4mi3n wrote: post #50

I'm still salty that I can't use em-dashes anymore for fear of my writing being flagged as AI generated. Been using them for years—it's just `alt+shift+-` on a Mac keyboard and I find them more legible in many fonts compared to the simple dash on the typical numpad. It's so sad to me that good typographical conventions have been co-opted by the zeitgeist of LLMs.

LLM adopting conventions (typographical or otherwise) is what they do, right? The idea that anyone should then have to change their behaviour is ridiculous, as is the whole conversation, really.

That's the rub though, isn't it? This feels like a form of self-censorship in response to some kind of shibboleth born of pattern recognition.

Re: New accounts on HN more likely to use em-dashes

#82

Post by shit_game »

gritzko wrote: post #29
marginalia_nu wrote: post #3

(author) I saw a 32:1 rate of EM-dashes last night when I just eyeballed the first 3 pages of /newcomments and /noobcomments. So I'm not sure how stable this is over over time.

This is probably the time to add some invitation system like GMail had in the beginning. Or make a shade for accounts <1yr. Or something else, before things get too mixed.

The issue with creating some hidden maturity heuristic for accounts is that it will be gamed just the same as any other, except that using age alone is the simplest heuristic to game. You can simply do nothing for incrimental periods of time and then begin testing aged accounts to roughly determine what the minimum age an account must reach to become "trusted".

Bot prevention is a very difficult constant game of cat and mouse, and a lot of bot operators have become very skilled at determining the hidden metrics used by platforms to bless accounts; that's their job, after all. I've become a big fan of lobste.rs' invitation tree approach, where the reputation of new accounts rides on the reputation of older accounts, and risks consequence up the chain. It also creates a very useful graph of account origin, allowing for scorched earth approaches to moderation that would otherwise require a serious (and often one-off) machine learning approach to connect accounts.

Re: New accounts on HN more likely to use em-dashes

#83

Post by OutOfHere »

jascha_eng wrote: post #46
CrzyLngPwd wrote: post #17

It's a predictable outcome, and it will get worse. What will/can HN do about it?

One solution is to get rid of anonymity online, enforce validation of identity. Every human only gets 1 account. And then we still ban people that use AI. Might take a bit but eventually we'll have filtered out all the grifters. If that's worth the cost... probably not?

Getting rid of anonymity is in time going to lead to getting rid of the platform, so do it if you're feeling suicidal. People seek real anonymity for good reason. Not everything should follow them in life or for life.

Re: New accounts on HN more likely to use em-dashes

#84

Post by MarioMan »

have_faith wrote: post #68
isoprophlex wrote: post #41

Earlier quoted context omitted.

The Big Chungus of dashes. Could this be the character that has the widest rendering?!

Unlikely in non-english languages (I seem to remember some super wide Arabic "single character" ones...?)

Last I’d checked, “﷽”is the widest Unicode character.

Re: New accounts on HN more likely to use em-dashes

#85

Post by bee_rider »

700 is actually a pretty good sample size unless you are looking at some tiny crosstab, or there’s some skew (which you won’t naively scale your way out of anyway).

It is also interesting to note that the comparison is between recent comments and recent comments by new users. So, I guess this would take care of the objection that em-dashes (a perfectly fine piece of punctuation) have just been popularized by bots, and now are used more often by humans as well.

Maybe there is a bot problem. Seems almost impossible to fix for a site like this…

Re: New accounts on HN more likely to use em-dashes

#86

Post by a4isms »

atourgates wrote: post #35

Shoutout to my English Major comrades who have been using em-dashes forever, and have had to stop so we don't sound like AI. If AI starts use the New Yorker style diaeresis (umlaut-looking thing when there are two vowels in words like coöperate) I swear I'm gonna lose it.

I worked for GitHub for a time. There was a cultural abhorrence of the diaeresis, it was considered reader-hostile and elitist. I refused to coöperate with that edict internally, although I grant that every company has the right to micro-manage communications with the public.

Re: New accounts on HN more likely to use em-dashes

#87

Post by bityard »

lapcat wrote: post #36
egypturnash wrote: post #6

— — — — — — — — — — — — — — — — — — — — — — — — — — — Don’t mind me, just skewing the results. — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — results. — — — — — — — — — — — — — — — — — — — — — — — —…

The use of em dashes is a human right. I ask that people not discriminate against em-dash users—we should be a protected class—and I refuse to abandon them. Perhaps I’ll have one engraved on my tombstone. He died doing what he loved—dashing.

I have always used double-dashes instead of emdashes, and it annoys me when software "auto-corrects" them into emdashes. Moreso since emdashes became an AI tell.

I also see AIs use emdashes in places where parentheses, colons, or sentence breaks are simply more appropriate.

Re: New accounts on HN more likely to use em-dashes

#88

Post by asplake »

d4mi3n wrote: post #81
asplake wrote: post #72

Earlier quoted context omitted.

LLM adopting conventions (typographical or otherwise) is what they do, right? The idea that anyone should then have to change their behaviour is ridiculous, as is the whole conversation, really.

That's the rub though, isn't it? This feels like a form of self-censorship in response to some kind of shibboleth born of pattern recognition.

Exactly

Re: New accounts on HN more likely to use em-dashes

#89

Post by marginalia_nu »

CharlesW wrote: post #20

A couple thoughts: (1) I don't recommend focusing disproportionately on one signal. They'll change, and are incredibly easy to optimize for. https://en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writing (2) I do recommend taking one minute to dash a note off to hn@ycombinator.com if you see suspicious patterns. Dang and our other intrepid mods are preturnatually responsive, and appear to appreciate the extra eyeballs o…

I have sent them an email a few days ago about the state of /noobcomments.

This wasn't really a intended as an "wow, dang is sure sleeping on the job", more than an interesting observation on the new bot ecosystem.

I also feel like there's a missing discussion about the comment quality on HN lately. It feels like it's dropped like crazy. Wanted to see if I could find some hard data to show I haven't gone full Terry Davis.

Re: New accounts on HN more likely to use em-dashes

#90

Post by OutOfHere »

The fear is that AI-generated comments will collectively promote an agenda, often a political or exploitative agenda, on a scale that humans can't match or hope to counter.

What could help is a careful clique hunting algorithm to accurately identify and delete the entire clique.

Post reply on HN