Quoting Charity Majors

Majors is not making a productivity argument. She is making a cost-structure argument: when code generation approaches zero marginal cost, the scarce resource becomes judgment about what code to write and whether to keep it, which is a fundamentally different hiring and process problem than throughput.

This connects directly to the Zhipu GLM-5.2 coverage from The Decoder on June 17th. That story documented open-source models approaching parity with closed-source leaders on sustained coding benchmarks, which is precisely the supply-side pressure Majors is describing. If capable coding models are proliferating under MIT licenses, the abundance she warns about is not a future condition but an accelerating present one. The public sentiment data from The Verge the same day adds a second pressure: two-thirds of Americans already feel AI is moving faster than institutions can absorb, and engineering teams generating more code with less friction are a concrete example of that gap materializing inside organizations.

Watch whether engineering orgs that have publicly committed to AI-assisted development (Shopify and Duolingo have both made statements this year) begin reporting measurable increases in defect rates or architectural rework costs within the next two to three quarters. That would be the first empirical confirmation that the abundance paradox Majors describes is showing up in production outcomes, not just theory.