One row per occupation that no longer exists, and what killed it: 197 hand-curated jobs across 13 walks of working life, from gong farmers, knocker-ups and garden hermits to switchboard operators, flight engineers and the ships' radio officers whose Morse watch ended in 1999. The closing act of the creative-destruction trilogy beside Famous Firsts and Dead Companies. There is no Wikipedia list for this; the spine is curated in the scraper and EVERY row is fact-checked against the occupation's own article before it ships (the article must read as obsolete and support the killer keywords; unsupported entries are excluded and reported). Killed_By names the killer with Cause_Type separating technology (136 jobs) from regulation (25: resurrectionists fell to the Anatomy Act, privateers to the Declaration of Paris) and social change (36). Killer_Debut_Year derives Years_Outlived_Killer, how long the work survived the invention that doomed it: median 33 years, from the Pony Express riders killed the same year the telegraph arrived to the fullers who outlived the fulling mill by six centuries. Eras ship as written ("Middle Ages", "1950s") beside numeric year twins, antiquity honestly NA. Peak employment and annual pay are mined from each article's own prose with the receipt preserved (Pay_As_Written, basis, currency, year; day and week wages annualized by stated conventions), so coverage is honestly sparse rather than invented. Dominant_Gender, an Employed_Children flag (35 jobs ran on child labor) and Survives_As (gone, niche, ceremonial, hobby, tourism, rebranded) complete the picture. Built with a 15-test known-facts suite.
197 rows in one CSV file, with 24 columns in total: Job, Description, Category, Started_Era, Started_Year, Died_Era, Death_Year, Years_Existed, Killed_By, Cause_Type, Killer_Debut_Year, Years_Outlived_Killer, Peak_Employment, Peak_Employment_Year, Pay_As_Written, Peak_Annual_Pay, Pay_Currency, Pay_Basis, Pay_Year, Dominant_Gender, Employed_Children, Survives_As, Job_URL, Killer_URL.
It is built by programmatically scraping en.wikipedia.org, last pulled on 2026-09-10. Datasets are versioned; older versions stay downloadable.
Yes. Every CodeSights dataset is completely free as a CSV download; a free account is all it takes. Anyone can preview the data without signing in.
Yes. The exact Python scraper that built it is viewable on the dataset page by any signed-in member, so every number is reproducible.
Automated scraping leaves room for error and the underlying sources change over time, so no version is guaranteed accurate or complete. If a number matters, verify it against the original source.