{"slug": "the-em-dash-report-who-started-punctuating-like-a-language-model", "title": "The Em Dash Report – who started punctuating like a language model", "summary": "An analysis of 80,913 posts and 59.7 million words from 42 blogs found that company blogs increased their use of em dashes after November 2022, with cloud and infrastructure blogs at 2.0×, AI labs at 1.8×, and developer tools at 1.3× their 2018–2022 rate per 1,000 words, while newsrooms still use the most at 3.42 per 1,000 words. The report, published by The Em Dash Report, tracks punctuation trends as a potential marker of machine-written prose but makes no claim about who used a model.", "body_md": "80,913 posts · 59.7M words · 42 blogs\n\n# Who started punctuating like a language model?\n\nThe em dash became the internet's favourite tell for machine-written prose. This counts them: every — in the body text of every post these 42 blogs publish, grouped by the month it went out. It makes no claim about who used a model. It is the punctuation, and the dates.\n\nCompany blogs moved together: cloud and infrastructure at 2.0×, AI labs at 1.8×, developer tools at 1.3× their 2018–2022 rate per 1,000 words. The newsrooms went the other way. Newsrooms still use the most at 3.42 per 1,000 words, as they always have.\n\n## Em dashes per post, 12-month trailing\n\nClick a name to show or hide it. Both charts share the selection and the address bar follows, so the URL always shares exactly what you are looking at. The dashed marker is Nov 2022, when ChatGPT launched — a date reference, not a claimed cause.\n\nEm dashes per post counts what a reader meets in one sitting, but it moves when posts get longer as well as when punctuation changes, and across this chart the average post grew 2.0×, from about 540 words in 2014 to 1,059 today. Over the same span the rate per 1,000 words went down 4%. Most of the climb above is longer posts, not denser punctuation. Switch to **per 1,000 words** to take length out of it.\n\n## Other things that changed\n\nA rise in one punctuation mark is thin evidence on its own. These are three other ways the prose moved over the same period. **Sentence variation** is the closest thing here to whether something reads like a person: human writing is bursty, a short sentence next to a long one, and a falling line means sentences are converging on one length.**Marker words** counts a narrow set that was rare in this corpus before 2023 — delve, tapestry, myriad, intricate and a few dozen others. None of these detects anything. Together they show whether a blog's writing changed shape, not who wrote it.\n\n## Words per post, 12-month trailing\n\n## The ranking\n\nOrdered by em dashes per 1,000 words since 2024. Rates divide summed dashes by summed words over the window, so a prolific month counts for more than a quiet one. Blogs with fewer than 40 usable posts are left out; [Method](/method/) lists them and why.\n\n| Blog | Group | Per 1k words, 2024–now | 2018–2022 | Change | Posts with one | Posts |\n|---|---|---|---|---|---|---|\n|\n\n[Anthropic](/blog/anthropic/)\n\n[Microsoft Azure](/blog/azure/)\n\n[Y Combinator](/blog/ycombinator/)\n\n[WIRED](/blog/wired/)\n\n[TechCrunch](/blog/techcrunch/)\n\n[The Verge](/blog/theverge/)\n\n[Together AI](/blog/together/)\n\n[Grafana Labs](/blog/grafana/)\n\n[Ars Technica](/blog/arstechnica/)\n\n[Replicate](/blog/replicate/)\n\n[Mistral AI](/blog/mistral/)\n\n[Tailscale](/blog/tailscale/)\n\n[Netflix Technology](/blog/netflix/)\n\n[Netlify](/blog/netlify/)\n\n[GitHub](/blog/github/)\n\n[Google Cloud](/blog/googlecloud/)\n\n[Google DeepMind](/blog/deepmind/)\n\n[Bun](/blog/bun/)\n\n[AWS](/blog/aws/)\n\n[Shopify Engineering](/blog/shopify/)\n\n[DigitalOcean](/blog/digitalocean/)\n\n[Cloudflare](/blog/cloudflare/)\n\n[Sentry](/blog/sentry/)\n\n[Docker](/blog/docker/)\n\n[Vercel](/blog/vercel/)\n\n[Deno](/blog/deno/)\n\n[Neon](/blog/neon/)\n\n[Meta Engineering](/blog/meta/)\n\n[Discord](/blog/discord/)\n\n[Railway](/blog/railway/)\n\n[PlanetScale](/blog/planetscale/)\n\n[Fly.io](/blog/flyio/)\n\n[Hugging Face](/blog/huggingface/)\n\n[Supabase](/blog/supabase/)\n\n[Nodecraft](/blog/nodecraft/)\n\n[JetBrains](/blog/jetbrains/)\n\n[TechRadar](/blog/techradar/)", "url": "https://wpnews.pro/news/the-em-dash-report-who-started-punctuating-like-a-language-model", "canonical_source": "https://emdash.report/", "published_at": "2026-08-05 17:48:26+00:00", "updated_at": "2026-08-05 18:07:14.710111+00:00", "lang": "en", "topics": ["artificial-intelligence", "generative-ai", "ai-research"], "entities": ["Anthropic", "Microsoft Azure", "Y Combinator", "WIRED", "TechCrunch", "The Verge", "Together AI", "Grafana Labs"], "alternates": {"html": "https://wpnews.pro/news/the-em-dash-report-who-started-punctuating-like-a-language-model", "markdown": "https://wpnews.pro/news/the-em-dash-report-who-started-punctuating-like-a-language-model.md", "text": "https://wpnews.pro/news/the-em-dash-report-who-started-punctuating-like-a-language-model.txt", "jsonld": "https://wpnews.pro/news/the-em-dash-report-who-started-punctuating-like-a-language-model.jsonld"}}