Everyone treats higher coding throughput as shipped product, but the mit sloan study found…
everyone treats higher coding throughput as shipped product, but the mit sloan study found autocomplete adds 40% more coding activity and async agents 180% more, with no matching rise in final outputs. an enterprise study measured pr throughput doubling to 2x in nine months while review capacity stayed flat.
the bottleneck moved, it did not disappear.
Context
The MIT Sloan article of September 2, 2026 describes a paper by Mert Demirer, Leon Musolff and Liyuan Yang using data on more than 100,000 GitHub developers. Autocomplete tools increased coding activity by 40 percent, the cumulative effect including sync agents was 140 percent, and the additional use of async agents brought it to 180 percent. That boosted coding led to 50 percent more projects and 30 percent more releases than for developers without AI. The NBER working paper 35275 describes a matched event study design on GitHub developers with usage telemetry. A separate arXiv case study of July 2, 2026 follows a mid-sized B2B company's 2x mandate announced in June 2025, with a panel of 802 developers and 196,212 pull requests from January 2024 to April 2026. Per-capita merged pull request throughput reached 2.09 times the pre-mandate baseline in April 2026, per-reviewer load roughly doubled, automated review overtook human review, and merge and revert rates held steady.
The post's 180 percent for async agents is the cumulative figure that includes autocomplete and sync agents, and the 40 percent is autocomplete alone. The note's no matching rise in final outputs is a simplification, since the article reports 50 percent more projects and 30 percent more releases, a smaller gain than in coding activity and not zero. The enterprise study is one company, with adoption not randomly assigned, and the authors call it strongly implicating an adoption-and-use channel and not exact causal attribution. It reached 2.09x over the period from the mandate to April 2026, and the exact nine-month window was not matched. Its finding on review is that per-reviewer load roughly doubled and automated review took over a large share, so review capacity stayed flat is not what the abstract says. The bottleneck moved, it did not disappear is the author's conclusion, and the MIT Sloan article says the researchers expect AI to address bottlenecks as it improves.
Watch next
- The NBER paper's own stage-by-stage figures and replication on other firms.
Sources
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 4 October 2026 at 22:33 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 →