[TimFogarty]

As somebody who used em-dashes a lot pre-ChatGPT, I have genuinely struggled with feeling I should change my writing style to appear more human. I would be happy with a double dash--but many programs autocorrect that to a full em-dash. So I'm left anxious that people will think I find them so unimportant I have offloaded communication with them to an LLM. So this post resonated with me.

I also like Will's "em-dash disclosure" on his about page:

> I like em dashes (—), en dashes (–), and hyphens (-), and I know how to type them. I also enjoy a well-placed ellipsis, but I didn’t know how to type one… until now. I believe that footnotes and sidenotes are superior to endnotes, appreciate the occasional fleuron, and at one point in my life, I knew what a colophon was.

> All of this is to say: the words, punctuation marks, misspellings, and opinions on this site are my own.

[macintux]

I have considered starting throwing more em-dashes into my writing, simply because I find the whole “this looks like LLM” to be a tiresome comment. Engage with (or dismiss) the material, not the pen.

[Jooror]

Have you considered throwing away your email spam filters? If so, I commend your willingness to engage with or dismiss the material

[furyofantares]

I've been pointing out LLM written stuff for months now, and often people ask how I determined it. When they do, I mention all the aesthetic things, and then I usually engage with the content and why the content is bad. In every case the content has been garbage. Usually it's a really bad infodump, in a singular tone, usually oversold, and you can't tell what was important to the original author and what's not. Often the some of the info isn't right. So it's like, infodump with extra labor to read that includes mistakes and masks what the author cared about.

It's just too easy to make garbage content that gets upvoted because it looks good if you skim it and serves as a good jumping-off ground for discussion. Engaging with the content of all the LLM-written garbage is a major waste of time and would make the site not worth it anymore to me.

Like it's already a major drain just to notice the aesthetic tells and then disengage. It's significantly more work to engage, and, AFAICT, around a 0% conversion rate to "oh shit I'm actually glad I read that."

[duskdozer]

Are you considering making the subject field for your personal emails "FREE VIAGRA" too? People try to filter out LLMs because they're often used like DDoS attacks on their energy.

[palmotea]

> ...simply because I find the whole “this looks like LLM” to be a tiresome comment. Engage with (or dismiss) the material, not the pen.

No. Engagement isn't free, and people need heuristics to figure out what's worth engaging with or not.

If people followed your advice, they'd waste their life conversing with dead-internet bots. And to what end? We're not machines mindlessly consuming and producing text. The our is often produced with a goal that's subverted if the consumer is a bot.

[abustamam]

> what's worth engaging with or not

I'd argue this entire HN discussion is proof that whether or not content is LLM generated, people can engage and have a meaningful discussion. I see lots of viewpoints in this discussion.

> And to what end?

The same could be asked of engaging with human commenters on HN :)

I comment on HN because writing is cathartic for me. If the person I'm responding to is a bot, or used a bot to generate it, it doesn't matter. I still stand by what I write. And other commenters can engage with what I wrote, regardless of the provenance of the text of the comment I responded to.

[palmotea]

> I'd argue this entire HN discussion is proof that whether or not content is LLM generated, people can engage and have a meaningful discussion. I see lots of viewpoints in this discussion.

People can talk to themselves too. The question is: is that what you want? Talking with a bot has a lot in common with talking with yourself.

> The same could be asked of engaging with human commenters on HN :)

One problem with software engineers is they oversimplify things based on technical factors, missing important characteristics. For instance: oversimplifying a text-based interaction into mere production and consumption of text, then claiming it doesn't matter what produces or consumes that text.

Well it does matter. For instance if you're looking for connection or community with other human beings, a bot is only giving of a false simulacrum of what you want. If you're engaging in politics, trying to convince others with persuasive case for your views, you've just wasted all your time if you were talking to a bot.

> I comment on HN because writing is cathartic for me. If the person I'm responding to is a bot, or used a bot to generate it, it doesn't matter. I still stand by what I write. And other commenters can engage with what I wrote, regardless of the provenance of the text of the comment I responded to.

Have you tried talking to yourself? That seems like it'd give you everything you need.

[abustamam]

Well sure. I have a blog that no one reads. That's textbook definition of talking to oneself in public.

> you've just wasted all your time if you were talking to a bot.

Not really. If the person I was responding to was a bot, it doesn't mean that humans can't read the thread and engage similarly. Most of the time when i respond to someone on HN, someone else responds to me with something compelling. I mean this thread is a prime example of that! Your comment I was responding to was a reply to someone else's comment. Let's say that person was a bot and neither you nor I are bots. We're still two humans connecting that wouldn't have otherwise connected if the bot hadn't posted that comment.

[palmotea]

>> you've just wasted all your time if you were talking to a bot.

> Not really. If the person I was responding to was a bot, it doesn't mean that humans can't read the thread and engage similarly.

Ok, so you've wasted some of your time, in proportion to how many bots you're interacting with. And the more prolific and capable they come, the more and more time you waste.

You seem to be walking a back your "engage with (or dismiss) the material, not the pen," statement. If you really stood by it, those humans wouldn't have to have to have read the thread.

[Kye]

"What's the current discourse on LLM writing tells as of today? Create a Markdown checklist."

Squishy brain heuristics can't last long enough to matter in this environment. Personally, I created a Claude skill to run this query (with some refinements) and check it against an article I suspect of being lazy AI writing. If it's good AI-supported writing, I probably won't suspect it, and I won't care if it is.

The people trying to fool you with lazy writing run the same list on their outputs to have the LLM fix it.

[jujube3]

You're absolutely right!

[halestock]

The pen is the material.

[zahlman]

> also enjoy a well-placed ellipsis, but I didn’t know how to type one… until now.

Made me look it up in my own environment. I had already set up a custom incantation for em and en dashes, although I really have no idea when to use the latter instead. Actually I never used to use em dashes, but now I do. I'd much rather deal with people who can intuit the quality of writing rather than relying on such blunt heuristics.

[abustamam]

Yeah I personally don't give it much thought. I write what I want to write, reread what I wrote, make sure it makes sense and briefly check for errors, then submit (at least as that's how I write HN comments, other venues may require more or less process).

Admittedly sometimes I'll pass my text through an LLM to check for obvious mistakes I may have missed. But the text itself was mine.

If that makes someone think I'm a bot, then maybe it's OK that we didn't engage anyway.

[alt187]

https://www.scottsmitelli.com/articles/em-dash-tool/

Discerning readers do not stop at the em dash. At least, I don't.

[patrickmay]

I use em-dashes, but I use them incorrectly, with a space before and after, because I think it looks better. I'm waiting to be flagged as an LLM.