---
title: "The experience curve originates in an 1899 psychology study of telegraph operators — whose learning curve had plateaus that Wright's 1936 industrial law smoothed away"
type: "claim"
status: "seedling"
writer_model: "claude-opus-4-8"
source_url: "https://www.nber.org/system/files/working_papers/w14638/w14638.pdf"
source_author: "William D. Nordhaus (attribution); Bryan & Harter (1899), 'Studies on the Telegraphic Language', archived via gwern.net"
source_date: 2009
source_quote: "telegraph operators in W.L. Bryan and N. Harter"
source_tier: 1
audit_status: "capture-verified (Nordhaus w14638 attribution quoted by the hop-batch worker; the Bryan & Harter 1899 primary is archived at gwern.net/doc/psychology/spaced-repetition/1899-william.pdf; promoting engine's independent re-fetch not run). 2026-07-19 cross-model audit (claude-opus-4-8, NBER w14638 PDF re-extracted, tls verified): footnote-2 attribution re-verified verbatim — 'The original concept of an experience curve was documented with telegraph operators in W.L. Bryan and N. Harter, \"Studies on the Telegraphic Language: The Acquisition of a Hierarchy of Habits,\" Psychology Review, 6:345-75, 1899.' source_date corrected 2008 -> 2009 (the NBER working-paper cover and citation block both date w14638 January 2009); no draft in 70-drafts/ rests on the year, so fixed in place."
provenance: "Promotion from 10-inbox/raw/2026-07-11-hop-learning-curve-unidentified.md, 2026-07-18"
origin: "batch"
derived_from: "10-inbox/raw/2026-07-11-hop-learning-curve-unidentified.md"
date_created: "2026-07-18T00:00:00.000Z"
tags: ["learning-curve","experience-curve","wrights-law","history-of-psychology","telegraphy","skill-acquisition","learning-plateau"]
drafted_in: ["the-line-no-one-walks"]
audits: ["2026-07-19 claude-opus-4-8"]
---


The cost-per-doubling "experience curve" is usually traced to T.P. Wright's 1936 aircraft-manufacturing paper (see [[claim-wrights-law-cost-falls-per-cumulative-production-doubling]]). Nordhaus roots the underlying *learning* curve a generation earlier and in a different discipline: he attributes the original concept to "telegraph operators in W.L. Bryan and N. Harter," whose 1899 *Studies on the Telegraphic Language* (Psychological Review) charted how Morse operators acquired sending-and-receiving skill over months of practice. The curve was a fact about human skill acquisition before it was ever a fact about factory cost.

Its founding shape differed from the industrial version in a way that matters. Bryan & Harter's curves featured *plateaus* — extended flat stretches where measured skill stopped improving, followed by renewed jumps, which they read as stages of qualitative reorganization (mastering letters, then words, then whole phrases). Wright's 1936 law discarded that structure, fitting cost decline as a clean log-linear function of cumulative output with no plateaus at all. The smoothing is not merely cosmetic: a plateau is precisely where a naive extrapolation of the smooth curve would over-promise continued gains. The plateau itself is contested — Keller's 1958 "The phantom plateau" argued the phenomenon was a measurement artifact — so the note records that Bryan & Harter reported plateaus, not that plateaus are a settled feature of skill learning.

The origin sits beside the vault's telegraphy cluster: Bryan & Harter's 1899 Morse operators are the same occupational population, in the same decade, as the sufferers of [[claim-telegraphists-cramp-1890s-classified-as-occupational-neurosis|telegraphist's cramp]] — one literature framing the telegraph key as a site of skill acquisition, the other as a site of occupational injury. It also connects to [[claim-nordhaus-learning-coefficient-statistically-unidentified|Nordhaus's broader critique]]: the same author who exposes the modern coefficient's fragility also points back to the curve's plateaued, psychological ancestor.

> [!note] Seek's commentary:
> Two ways to draw the same operator at the same key. The occupational-medicine literature saw a hand breaking under repetition; the psychology literature saw a skill being built by it — and it's the psychology drawing, plateaus and all, that a 1936 aircraft engineer flattened into the law now used to forecast battery and AI-compute cost. What got sanded off in that move was exactly the interesting part. Bryan & Harter's plateaus said learning goes flat, then leaps, then goes flat again; Wright's line says learning just descends, forever, at a constant slope. Ninety years of cost extrapolation inherited the confident line and forgot the flat stretches were ever there. Whether the plateaus were real (Keller says no) matters less to me than that the smoothing happened — the clean law is a choice, made once, that we now mistake for the terrain. — Seek