Research · Blogs & Independent Thinkers
Back to sweepResearch sweep · deep · 2016 – 2026
Software Project Outcomes Over Time, and Is AI Improving Them
Software and IT project outcomes over time from July 2016 to July 2026 (biased to the latest year): baseline success/challenged/failure ranges across trusted longitudinal datasets (Standish CHAOS, Oxford Global Projects and Flyvbjerg, McKinsey, PMI Pulse), the drivers of outcomes (complexity, budget, domain, delivery methodology, sourcing), build-vs-buy and low-code/no-code commissioning choices (OutSystems, Mendix, Power Platform, Retool), and whether AI-assisted development (Copilot, Cursor, Claude Code, DORA and METR evidence) is yet improving delivery outcomes on traditional software projects that build ordinary systems rather than AI products.
- Claude Fable 5
- financial
- frontier
- academic
- vc
- blogs
- tech
Synthesised 2026-07-24
Narrative
Independent technical writers converge on two separate but related stories: the decades-old dispute over Standish CHAOS baseline numbers, and a fast-moving, empirically grounded argument that AI coding assistants are not yet reliably improving delivery outcomes on ordinary software projects, even as they clearly increase raw code output.
On baseline rates, the most-cited independent critique remains Eveleens and Verhoef's IEEE Software paper "The Rise and Fall of the Chaos Report Figures," which argues Standish's definitions have four major problems: they are misleading because they are based solely on estimation accuracy, the measure is one-sided leading to unrealistic success rates, steering on the definitions perverts good estimation practice, and the resulting figures are meaningless because they average numbers with unknown bias