Write Like It's 1866: LLMs Relearn Telegraphese
What happened
Researchers tested how modern large language models can mimic the terse style of 19th‑century telegraph messages, called cablese, by instructing them to drop articles, filler words, and use lowercase. They compared two historic compression methods—codebooks and telegraphese—and found about a 10% reduction in token count versus plain prose. The experiment used GLM‑5.3‑Flash for both writing and reading the compressed records, showing that downstream models answered questions from these records as well as from plaintext, with recovery ratios around 1.00‑1.10. The most effective instruction was to use lowercase, which yielded up to 48% savings for some models.
Why it matters
For everyday readers, this shows that AI can efficiently compress text without losing comprehension, potentially reducing data usage and costs in bandwidth‑limited contexts. It also suggests that older writing conventions still influence modern AI behavior.
Reality check
An automated, sceptical read of the sourceIncremental. The work demonstrates a modest efficiency gain using known compression tricks rather than introducing a fundamentally new capability.
Is this a big deal?
Get stories like this every morning
The day's AI news in plain English, in your inbox at 8:30 IST. Free.
More AI news
Show HN: New Cartesia Text-to-Speech ModelIndustry news · 13 Nov 2026Chicago schools to Google – we'd rather lose Search than be forced to accept AIIndustry news · 13 Nov 2026Introducing ChatGPT WalletOpenAI · 13 Nov 2026Shield helps technicians hand off legacy portal work with OpenAIOpenAI · 13 Nov 2026How AI Is Helping Care Teams Get Patients the Right Care, SoonerNVIDIA · 13 Nov 2026This page was collected from fiveminutesforward.com and summarised automatically by AIPulsee's pipeline. See how it runs.