Of course, I guess it's entirely possible—even accounting for OS—that this test remains statistically useful. It makes me kinda sad that my (very much human-generated) writing fails the Turing test....
(Windows probably has some way, but those are rarely discoverable.)
Released back in 2019 for Windows 10.
1. Install and configure this extra tool, which also by default enables a ton of other things you may not want, and may as well be a third-party tool even though it's technically built by Microsoft
2. Do a Google search and copy-paste (!)
3. Use a keyboard shortcut to bring up a symbol picker, then click on the tab containing the en and em dashes, then click to type them in
I mean, come on.
-- into —
If OP wrote their post on an iPhone, they would have inadvertently appeared as an LLM by their own test.
----
[2] Though having checked just now, the sequences for en-dash and em-dash don't seem to be working. Perhaps one of my custom macros is interfering somehow… (it is behaving overall, ellipsis just worked as did the following diacritic and other symbols: áèîöūñ±⁰¹²∞¡¿‽π⬚). I'll have to poke at it later and see what is ary.
(edit: apparently only on Mac, see reply below)
You just type two hyphens (--) and Word will convert it to an em dash.
Typing <word><hyphenminus><hyphenminus><word><space> yields an em dash.
Typing <word><space><hyphenminus><hyphenminus><space><word><space> yields an en dash.
That this has been true for some 3 or 4 decades makes me doubt all the comments that em dashes are a "tell" of LLM authorship. On the other hand, I guess when we confine this possibility to web content, I can see how people haven't used Office for web authoring lately, and whatever they do use (like web-based content management systems) don't tend to have this feature.
More importantly, typing just a single hyphen minus in this constellation triggers the autoreplace, too. (Typing the double hyphen is only necessary without spaces in order to distinguish between an intentional hyphen and an em dash.)
From TFA:
> August 1–August 31
From a top comment:
> Boston–San Francisco flight, 10–20 years
To achieve this using the replacement feature we're talking about would take something like <word><space><hyphenminus><space><word><space><alt+leftarrow><bksp><leftarrow><bksp><alt+rightarrow> which is ridiculous.
In professional typesetting, like a book, I sometimes see spaces flanking an em dash, however.
I almost never want that, and when typing "space, en dash, space", it happens quite easily and is usually impossible to tell visually.
In any case: non-breaking space (or otherwise suppressed linebreak, if you don’t use a space) is the rule.
(Side note: GTP says apostrophes should be used for pluralizing only for single letters to avoid confusion, but this seems more readable than "ens and ems" IMO.)
I also personally prefer en dashes, surrounded by whitespace on both sides, over em dashes. Apparently some WYSIWYG software interprets two hyphens as an em dash, while other will interpret that as an en dash, so I'd rather just use the real thing if possible to avoid the ambiguity.
I don't doubt there are publishing platforms that do it automatically as well, so I wouldn't count on seeing them as an indicator of generated output, even if it may be processed in some manner.
Money quote:
This issue does indeed have a history of provoking unhinged lunacy.<Multi_key> <minus> <minus> <period> : "–" U2013 # EN DASH
<Multi_key> <minus> <minus> <minus> : "—" U2014 # EM DASH
More in /usr/share/X11/locale/en_US.UTF-8/Compose
Or we are both just AIs, as a portion of HN comments are, commenting back and forth about other AIs.
If you use eg. a Japanese IME, you can also get it by typing a normal hyphen and selecting the em dash from the picker.
In your support, though, calling the extension “smartypants” really hints at the target audience :)
I was able to demonstrate my long use of them, prior to LLMs. And since I write in quarto markdown I don't need keyboard shortcuts.
Not sure when I started; my guess is that I got into the habit of using them in LaTeX when writing my thesis, and then at some point realized that they are easily reachable on standard macOS keyboard layouts (via "option" + "-").
Imagine being an NPC (a human bot), flattering yourself with the thought that people who understand the language are language bots...
1: https://www.thenationalliteracyinstitute.com/post/literacy-s...
1: https://en.wikipedia.org/wiki/Literacy_in_the_United_States
See, e.g., Boss Szabo's blog: https://unenumerated.blogspot.com/2018/03/the-many-tradition...
Two chained hypens, as was pretty much the norm back then.
And did you just call me an NPC?!? It's not a matter of "understanding the language" at all. It's a matter of convenience and of a sort of evolved convention.
I write poems a fair bit and use em dash a lot. (maybe too much and incorrectly)
Serbian and Croatian XKB keyboard layouts have had em- and en-dashes since early 2000s even if they were not standardized: AltGr (right Alt) + hyphen (to the left of right Shift) produces an em-dash, and press Shift on top, and you get an en-dash.
This is how long I've had them easily accessible on any keyboard (I even have them converted to MacOS keyboard layouts for use with Karabiner).
On my Linux laptop, I confess to manually Googling them every time.
Hyphen -: -
En Dash –: alt -
Em Dash —: alt shift -
They have shortcuts for Í, Î, and Ï but not for many commonly used characters like arrows
sigh
Thanks for the suggestions
You can remap Fn/Globe directly to it if you want. It's also accessible from the Input menu bar item if you show that.
> Both of those open the same GUI which combines emojis, stickers, and unicode symbols—preferring the first two categories over the last. To type out a unicode symbol it takes at least three clicks on top of me starting to type in the name of my symbol
Are you using the expanded Character Viewer window[0], or the default collapsed Emoji & Symbols pane[1]? Because the expanded Character Viewer lets you customise and reorder the categories[2] (though that doesn't affect search), including adding a full Unicode view[3]. And they both default to the search bar when opened (though the Character Viewer opens unfocused for some reason).
[0]: https://imgur.com/hTtrbcA
[1]: https://imgur.com/3L31DQu
¹ https://software.sil.org/ukelele/
² https://codeberg.org/datatravelandexperiments/kps-keyboard-l...
Examples:
en and em are on -
Below are maybe Swiss specific?
~ is on N
@ is on G
| and \ and / are on 7
√ is on V
¥ is on Y and € is on E
∑ on W ( ∑ is a rotated W :)
etc.
I appreciate that the designer of the layout clearly attempted to make some kind of mnemonic connection to the degree they could. Makes it easier to discover and remember the key-combos, even without a cheat sheet.
modifiers:
opt-e+letter é (acute/aigu)
opt-`+letter è (grave)
opt-i+letter û (circumflex)
opt-u+letter ü (umlaut)
opt-n+letter ñ (for the mañana)
thank you for teaching me √
Its not wrong for en-dashes (and en-dash set open—with space on either side—is generally an alternative to an em-dash set closed.) And its not wrong on the trailing side of an em-dash used in dialogue to show an abrupt stop mid-sentence if the stop is followed by a new sentence. And there's a few other particular uses, but, generally, setting an em-dash open is wrong.
> but newspapers often space them.
I've never seen a newspaper set em-dashes open, but I have seen them use en-dashes set open instead of using em-dashes at all. Given the space premium in print newspapers, em-dashes set open, which would consume enormous horizontal space, would, other concerns aside, be an odd choice.
I, of course, used proper dashes in typeset documents, at least after I'd learnt about them in Knuth's The TeXbook. I have found myself occasionally use them in ASCII contexts just as ---. But I've never sought out the proper unicode character.
It is maddening that the whole world uses typewriter keyboards with some facelift in the era of Unicode and even blasphemous full color emoji font rendering. What has changed in decades? Windows logo key, power keys, media keys, IE and Outlook logo keys — all Microsoft's fancies.
So initially IBM made some ad hoc decisions on what keys would be suitable for a single user office computer (as opposed to data input and admin terminals they had). Then everyone copied that, because sending unexpected scan codes could lead to bad things (random BIOS and program code couldn't care less about your ideas of forward compatibility). Then Windows became the “basic system” installed on most computers. Microsoft really pushed forward the internationalisation at the time, making a lot of national layouts and code pages (sometimes contradicting the national standards, for better or for worse). Then everyone copied what they decided. What's more important, even single byte code pages had the basic typographic symbols, anyone could've been using them for three decades, but they were not added to most physical keyboard layouts.
I wonder if that was because they wanted Word to seem more sophisticated than it was, and to make people think it was a requirement for “proper documents”, or because programmers still treated all non-ASCII symbols as free data markup constants that would “never appear in a regular text”.
Didn't it match ASCII and possibly typewriter keyboards?
For other combos — see /usr/share/X11/locale/en_US.UTF-8/Compose
See also: System Settings > Keyboard > Key Bindings > Position of Compose key
option-[-] for en dash –
shift-option-[-] for em dash —
Compose--- produces —
Compose--. produces –
Lots of other characters like áăǎ°±€ are available through compose: https://whynothugo.nl/journal/2024/07/12/typing-non-english-...
> Absolutely proper and correct use of em dashes, en dashes, and hyphens is, to me, the most obvious tell of the LLM writer.
Or just someone who likes to use the right characters. There was a report a few months back about how writing from autistic kids keeps getting mislabelled as LLM simply because they use the correct specific terms.
Please stop associating being precise with being an LLM.
Or, I mean, it does SOMETHING. I've never checked, and just always assumed I was getting the em dash.
I mean if it's an obvious break from their normal style, sure. But by itself? Every time I hear this argument, it just seems like sour grapes from poor writers.
Absolutely proper and correct use of em dashes, en dashes, and hyphens is, to me, the most obvious tell of the LLM writer. In fact, I think that you can use it to date internet writing in general. For it seems to me that real em dashes were uncommon pre-2022.