LLMs prompted in 1860s telegraphese cut billed output tokens by 40-49% with full fidelity
Pytlicek · x · 2026-10-08
A reproducible blog experiment shows that prompting models to answer in "cablese" — the compressed dialect 19th-century telegraph operators invented to dodge $10-per-word fees — cuts billed output tokens by 40-49% across four model families while preserving full information fidelity. The author built the Telegraph Test benchmark to measure both compression and retrieval, and open-sourced the harness, frozen ledgers, and a one-click notebook at github.com/Travis42/telegraph-test.
More from Research
- 0 of 722 AI bio papers cured cancer, as biologists tell tech bros it's brutally hard — iScienceLuvr · 2026-10-08
- New multi-agent swarm writeup: smarter models benefit more from collaborative work — repligate · 2026-10-08
- COLM adds missing panel title: Math, Science, and Knowledge Production in the Era of LLMs — gregd_nlp · 2026-10-08
- Actionable Interpretability workshop at COLM 2026: four keynotes from Goodfire, Mila, Stanford — megamor2 · 2026-10-08
- COLM Paper: Meaningful Neuron Directions Extracted from MLP Weights Alone — megamor2 · 2026-10-08
- CoBPE tokenization cuts sequences 30% vs BPE, better LLMs at same compute — alisawuffles · 2026-10-08