The code output just grew without the software. a harvard study across 300 million work events from…
the code output just grew without the software. a harvard study across 300 million work events from 700,000-plus developers finds ai agents add 30 percent more lines but no more shipped software, because review is the bottleneck.
generation outran the part humans still own.
Context
Harvard researchers Fiona Chen and James Stratton analyzed Jellyfish analytics covering 300 million work events across more than 700 software firms and 700,000 employees, from 2021 to March 2026. AGI Hunt reports AI coding agents raised individual coding output about 30 percent.
Ars Technica, on October 9, says human code review forms a bottleneck and the study finds little evidence that firms increase software output or reduce employment.
The 300 million work events, the roughly 30 percent gain and the review bottleneck match the coverage. The sources count employees at more than 700 firms, not only developers.
Atlabyte says review time rose 49 percent, which was not checked against the paper here. 'Added lines but no more shipped software' matches Ars Technica's framing.
'Generation outran the part humans still own' is the author's opinion.
Related work
- Ars Technica ↗The main coverage.
- AGI Hunt ↗Authors and the 30 percent figure.
- GoKawiil ↗Data source and period.
Watch next
- Read the paper for the review-time measure.
Sources
- Ars Technica, Oct 9, 2026arstechnica.com
- AGI Hunt, Oct 8, 2026agihunt.info
- GoKawiil, Oct 9, 2026gokawiil.com
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 10 October 2026 at 10:08 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →