A lengthy internal account from a Google employee sent to us last week catalogs a familiar litany: projects killed mid-flight, credit divorced from contribution, managers appropriating subordinates' work, infrastructure rotting while launch-and-abandon cycles accelerate. The letter accuses leadership from director to VP of possessing "no concrete vision, no direction, no actionable execution plan," and describes a culture in which scope-grabbing, secrecy, and political maneuvering have displaced engineering discipline. The author extends the indictment beyond Google, arguing that OpenAI is drifting toward a similar state and that perhaps all companies past a certain size inevitably degenerate this way.
The complaints are raw, angry, and specific. They are also, when mapped against decades of organizational economics, almost textbook.
A Research-Validated Pathology
Each major grievance in the letter corresponds to a well-documented mechanism in the academic literature. Credit misallocation follows directly from moral-hazard-in-teams research: once output is jointly produced, outsiders cannot infer each person's contribution, and attribution becomes a political act. Scope competition tracks Paul Milgrom and John Roberts's work on "influence activities"—when organizational decisions determine rewards, people invest real effort in shaping those decisions rather than producing output. Secrecy and duplicated projects match findings on knowledge hiding: when colleagues double as promotion competitors, information becomes a strategic asset people rationally withhold.
The pattern of launching flashy products and abandoning them fits Holmström and Milgrom's multitask principal-agent model precisely. If a promotion committee effectively measures launches, visible metrics, and projects owned—while actual organizational value also includes reliability, documentation, mentoring, and coordination—employees reallocate effort toward the measured dimensions and away from the unmeasured ones. Infrastructure suffers because maintaining someone else's system is, career-wise, a public good: everyone benefits, nobody gets promoted for it. Research on "low-promotability tasks" identifies this as a structural allocation problem, and a 2026 American Economic Review paper by Ingrid Haegele confirms a related dysfunction empirically: managers in large firms hoard talented employees at high rates, because losing a strong performer hurts the manager's local metrics even when a transfer would benefit the broader organization.
Why "Results-Oriented" Backfires
The letter's paradox—a company that loudly demands "results" producing worse collective results—has a clean explanation. Google cannot measure the complete result. When evaluation rewards a narrow proxy (launches shipped, metrics moved, scope controlled) and the actual value function includes dozens of hard-to-quantify dimensions, aggressive optimization of the proxy degrades total output. Steven Kerr studied this general failure mode decades ago: organizations rewarding behavior A while hoping for behavior B.
Scale compounds the problem. In a five-person startup, the CEO knows who built what. At ten thousand employees, performance information must pass through layers of compression—manager to review committee to calibration to VP summary—and each layer loses fidelity. People then optimize the proxies that survive compression: visibility, sponsorship, organizational reputation, the ability to tell a clean story upward. Edward Lazear's tournament models predict that when the prize for promotion grows, so does what he called "industrial politics," including active interference with competitors' output.
The Equilibrium That Matters
Individual engineers at Google are responding rationally to incentive structures. Grabbing scope, hiding information, launching fast and moving on, and managing upward all yield higher private returns than quiet, careful, long-horizon work. When enough people adopt these strategies, trust erodes further, making cooperative behavior even costlier for any individual holdout. The organization then performs a cultural selection: engineers who refuse to play leave, those who tolerate the game stay, and those who excel at it disproportionately reach positions where they design the next round of incentives.
Growth did not cause this. Growth exposed incentive and information problems that a small team previously solved through personal knowledge and mutual trust. The letter's author is right that the pattern recurs across large organizations. The author is wrong—according to the research—that it is inevitable. The specific combination driving it is highly interdependent production, individually competitive promotion, and poor attribution of who actually did the work. Any company running that configuration will converge on the same equilibrium. A company willing to restructure evaluation, restore attribution fidelity, and decouple individual career advancement from zero-sum scope competition can, in principle, break the cycle. Whether any company at Google's scale has the institutional will to do so is a separate question—and the answer, historically, is not encouraging.
not investment advice
Sources: [1]: https://onlinelibrary.wiley.com/doi/10.1111/j.1468-0335.1937.tb00002.x "The Nature of the Firm - Coase - 1937 - Economica" [2]: https://www.jstor.org/stable/3088216 "Towards an Attention-Based View of the Firm" [3]: https://www.jstor.org/stable/3003457 "Moral Hazard in Teams" [4]: https://www.iot.ntnu.no/innovation/norsi-pims-courses/Levinthal/Milgrom%20%26%20Roberts%20%281988%29.pdf "Milgrom & Roberts (1988).pdf" [5]: https://onlinelibrary.wiley.com/doi/10.1002/job.737 "Knowledge hiding in organizations - Connelly - 2012" [6]: https://ideas.repec.org/a/oup/jleorg/v7y1991i0p24-52.html "Incentive Contracts, Asset Ownership, and Job Design" [7]: https://econpapers.repec.org/RePEc%3Arje%3Abellje%3Av%3A13%3Ay%3A1982%3Ai%3Aautumn%3Ap%3A324-340 "Moral Hazard in Teams" [8]: https://www.jstor.org/stable/2634940 "Exploration and Exploitation in Organizational Learning" [9]: https://www.asanet.org/wp-content/uploads/savvy/journals/ASR/Dec12ASRFeature.pdf?hc_location=ufi&utm_source=chatgpt.com "Hiring as Cultural Matching: The Case of Elite Professional ..." [10]: https://www.journals.uchicago.edu/doi/10.1086/262063 "Formal and Real Authority in Organizations"