本文由 AI 分析生成
建立時間： 2026-03-28 來源： https://churchofturing.github.io/the-enterprise-experience.html
Summary
“Church” (a developer with a decade of startup/SME experience) reflects on their first year at a large enterprise after deliberately choosing to “sell out.” Observations cover: invisible tool ownership, spectacular monetary waste, competency variance due to no-termination culture, urgency theater, metric-worship as proxy for security, six people with the title “head of architecture,” and the selection paradox of senior leadership. Deliberately sardonic, but balanced — the author has no regrets.
一位在新創和中小企業工作十年的開發者進入大型企業一年後的觀察：無主工具、驚人浪費、無解雇文化造成的能力差異、假緊急、以指標掩蓋安全感、六位「架構主管」，以及高層選拔的悖論。諷刺但平衡。
Key Points
- Orphaned tools: tools outlive their owners; when owners leave, tools become “rogue” — costing thousands/month, unsupported, slowly breaking
- Waste at scale: retirement-fund sized projects killed before launch for minor overruns; Raspberry Pi-appropriate workloads on AWS; hundreds of people debating a $100/month SaaS
- Competency inconsistency: no performance-related terminations → inconsistent new hires → self-perpetuating baseline problem; surreal situations where technical leads can’t use computers
- Schrödinger’s urgency: some urgency is real (production incidents), some is artificial (arbitrary dates committed without telling the team), some exists only if you engage with it
- Metric theater: security team exports vulnerability Excel sheets twice daily; the number of “bad things” goes down; engineers get called “Dependabot” — process in service of dashboard, not security
- Senior leadership paradox: hired to project confidence and make immediate change; rewarded for exactly the traits that prevent learning domain complexity; repeat failures guaranteed
Insights
The “Dependabot” story is the clearest example of Goodhart’s Law in action: “When a measure becomes a target, it ceases to be a good measure.” The security process optimizes for the vulnerability count metric, not actual risk reduction. Engineers who point out that a CVE has no exploit path are obstacles, not contributors.
The senior leadership selection paradox has a structural explanation: boards and C-suites select for narrative confidence because that’s what shows well in a few hours of interviews. Domain knowledge and epistemic humility don’t signal in that context. The same selection pressure that produces “this time it’ll be different” speeches also produces quick wins that justify the hire before the real problems surface.
The contrast with startups isn’t just cultural — it’s incentive structures. Startup employees see direct consequences of waste; enterprise employees have enough abstraction layers that the consequences land on someone else’s budget.
Connections
Raw Excerpt
If the selection function for senior leadership is tuned to a certain personality type, you’re doomed to repeat your mistakes.