A Defense—or an Elegy—for the Em Dash

(Photo by satwik arora on Unsplash)

LinkedIn is a realm of nightmares.

Soundcloud rappers turned sales gurus, spitting bars about bagging SQLs. Content strategists lauding the realism of generative AI, meanwhile posting videos of writhing, twitching humanoids defying Euclid’s geometry as they attempt to pet a cat. Newlywed husbands cherishing their seaside nuptials—by musing about how it relates to B2B marketing.

I work in B2B marketing—no, I’ve never married on the beach—and visiting LinkedIn is an unsettling necessity of my job. In the absence of agrarian tasks, it’s my version of shoveling manure. My nostrils are numb at this point to the piles of thought leadership plopping into my feed each morning. But a couple of weeks ago, I saw a post that finally broke my spirit.

Someone was slandering the em dash.

For reference, the em dash is longest of the dashes (—) and my favorite bit of punctuation. A hyphen (-) is useful for compounding words, sure. An en dash (–) is pragmatic for showing ranges, of course. But the em dash? The em dash is rebellious and decadent. It’s the instigator of tangents, non-sequitors, and absurdities. And unlike parentheses, the em dash regards a derailment of thought not as an optional aside, but as indispensable zigzag on an otherwise dull track of semantics. It can even replace the comma for a twist of rhetoric that demands more flourish. It happily indulges every new phrase, no matter how errant or wild. The em dash is a stalwart friend to those of us whose brains can’t think in straight lines.

And according to this LinkedIn poster, it now belonged to robots.

In the post, she said that generative AI models often use the em dash because it’s reflective of existing style guides, whereas most human authors won’t use it, mainly because the keystrokes required to make it are cumbersome. For her, an em dash was like a Voight-Kampff test, a betrayal that the writer behind the words was less organic than robotic. The implication was that, if someone wanted to avoid this perception, they might want to avoid the em dash entirely.

Now, I’m being hyperbolic when I say this is slander, but as a frequent user of the em dash, it necessarily puts me on the defense. I am not a robot, at least not to my knowledge. I could easily be accused of a glitched awkwardness at business meetings and cocktail parties, but I assure you, that’s the result of pure, bio-based neurology. If anything, I’d like to think it makes me more human.

But written text is a mode of communication that I can’t afford for people to misinterpret as robotic. For one thing, it’s my job as a marketer to forge an emotional connection between customers and a brand, something that I hopefully do with integrity—and that I’ve done with no small amount of em dashes in my copy writing. If customers smell the insincerity of a large language model, whether or not I’ve used one, then the bond with the brand is broken, and I’ll need to write a new resume. (Let’s hope no one thinks that’s the result of an LLM, either.)

More importantly, though, writing is my preferred conduit for words in general. As someone who has issues regulating his attention, writing and editing gives me a fenced area to wrangle and domesticate my thoughts—whereas spoken words feel like a herd of wild horses, and I’m supposed to somehow lasso them with Silly String. If people no longer trust the origin of written words, then I’ve lost the craft by which I’ve always felt my thoughts could best be understood.

The em dash is a casualty of the generative AI era, but it’s not the most consequential one, despite my affection for it. The greater casualty is written words in general. This ancient vessel, the word, was designed to float meaning from one brain to another. Porous though it is, it’s still the best method we had for transferring ideas from person to person without those ideas capsizing entirely. And even if those words can be strung together into untruths, you could be reasonable certain before the advent of genAI that those words had embarked from the port of a human mind.

Not anymore. In every medium from printed books to instant messaging, the existence of genAI has drained the perceived veracity of words. There’s no longer a full assurance that what’s speaking to you is human, or if it’s the tortured amalgamation of a million different voices, fashioned by an unthinking algorithm into a vaguely canny echo of one. This unsettling reality leaves us whipping out our magnifying glasses—sleuthing for clues like em dashes—vainly hoping to snoop out the robots so we can maintain the internet as a place of real connection.

I won’t stop using the em dash, no matter how ridiculous the keystrokes. And I won’t stop using my own neural processors to write, no matter how imperfect the results. I’m not a Luddite about LLMs: so long as the results can be reverse-engineered, there is great potential in LLMs as an assistive tool. But to delegate the writing process entirely to them is to deprive ourselves of the reason writing exists in the first place. Writing is a tool that makes us think deeper, dream bigger, reason harder, and feel stronger. It forges minds in fire, and it blazes trails between them.

The more we trust LLMs with our writing, the more we lose than just em dashes. We will blunt the tool we’ve used for millennia to make us more human.

And the more like hell LinkedIn will become.