The em dash is dead – long live the spaced en dash

I love using em dashes, arguably the most versatile of all punctuation, to the point that my friends make fun of me for it. But sadly, when you encounter text on the modern Internet with properly formatted em dashes—i.e., U+2014 em dash—it's more likely than not that it's LLM generated. Insofar as writers have started avoiding them, I consider em dashes one of the greatest casualties of the LLM revolution.

Obviously, this doesn't matter when you're reading The Wall Street Journal, or a post by someone you already know to be fastidious in their work; but when sifting through which technical blog post or so-far-interesting online comment to read to the end, it becomes an easy heuristic, especially when other aspects of the writing are LLM-adjacent.

A common shorthand is to use one or more hyphens- like this -- or like this.¹ I think this is a good option when in a pinch or limited to ASCII, but it's a little unsightly, and the one-character case can easily be confused with ordinary hyphenation. Fortunately, there's an alternative: the spaced en dash. It looks like this – and it's really cute!

While an unspaced en dash is universally used for indicating ranges (2:30–3:30), a spaced en dash is a valid alternative for an em dash used to set off information in sentences. Wikipedia has a spaced en dash template, which even makes the first space non-breaking so that lines don't start off with an en dash (I usually don't go that far). That page notes:

The spaced en dash is generally preferred in modern 'Commonwealth English' typography, while older British and recent American typography generally uses an unspaced em dash for the same purpose.

Most LLMs are biased toward outputting formal American English, so it's no surprise that they generate em dashes aplenty and few spaced en dashes. Although I find the generous width of the em dash to be wonderfully indulgent, I think spaced en dashes still get the job done, while being just as unambiguous and readable.

My personal policy for now: I'll keep using my beloved em dash in long-form writing, where there's plenty of context to reduce the reader's suspicion. But otherwise, I'll be using spaced en dashes.