A gate built for quality becomes a bunker the day someone needs it to keep their job.
A team sends a design to the architecture review board and gets it back with one comment: “I have architectural concerns.” Perhaps the design is bad and the reviewer has just saved the company a year of pain. Perhaps it routes around a platform the reviewer’s own team owns, and approving it would make that team less necessary. The sentence is the same either way, and so are the delays, the requests for more evidence and the appeals to consistency. From outside there is no telling which reviewer you have.
The usual story about decaying organizations is that they grow, bureaucracy piles up, and the wrong people end up in charge, so the cure is better people. The review board suggests another account. Any point that controls the flow of work also controls credit, blame, hiring, promotion and access, and in a bad year it controls who survives. The bottleneck becomes a throne the moment people need it for protection more than the organization needs it for throughput.
How a filter becomes a shelter
A healthy chokepoint is a quality filter, which says no because the work demands it. A captured chokepoint is a survival asset, which says no because saying no preserves the operator’s position. Both versions still think, but the captured one thinks for itself.
Capture is hard to challenge because it rarely requires lying. The reviewer translates private interest into public language such as risk, maintainability, security or alignment, and the language may well be true, since almost every design carries some real architectural risk. A captured chokepoint only needs to select the true reason that preserves control.
Why would a decent engineer drift into this? Greed is the least likely reason. If employment, immigration status, social standing, retirement security, family obligation or professional identity depends on staying inside the machine, then control over a chokepoint becomes insurance. Suppose the reviewer is on a work visa and her team was recently described as “under evaluation”. A pure meritocracy, for her, means a floor that can disappear tomorrow. The rational response is a coalition that makes disappearance less likely: allies on the board, and a platform that only her team understands.
Why nobody fixes it
Everyone may see that the board is slowing the product, yet nobody can defect safely. The reviewer cannot loosen control without losing protection. Her allies cannot criticize her without losing their own shelter. A team that routes around the board is marked as hostile, and management cannot dismantle the gate without paying the whole political cost at once. That is an inadequate equilibrium: ugly, widely recognized, and stable. A would-be reformer thinks he is improving throughput, while the reviewer experiences the reform as eviction.
Repeat this at the hiring loop, the promotion panel and the compliance desk, and each gate keeps working for a different beneficiary. The user is not the real client of a captured chokepoint. The real client is the operator’s future self: still employed, still credentialed, still respectable, still legally present, still hard to remove. Enshittification is what product quality looks like after every chokepoint has been repurposed into somebody’s survival bunker. Appeals to craftsmanship fail here because craftsmanship is competing with survival.
Not every gate is a throne
The opposite mistake is to treat every “no” as self-dealing. Some reviewers really are protecting the product, and a company that tears down its gates will ship the bad designs they existed to stop. The goal is to keep chokepoints aligned with reality.
That alignment does not come from exhortation, because chokepoints are survival assets before they are process assets. Any point that controls flow will be captured unless the surrounding system makes capture less valuable than quality. That takes low enough precarity, high enough trust, clear enough ownership and strong enough standards that the operator can afford to serve the work instead of sheltering behind the gate.
It also explains why firing the reviewer rarely helps. The captured chokepoint is usually an adaptation to an external threat, and if you remove the operator without removing the threat, someone else will capture the same gate. There is no culture change in this sense, because the culture is the memory of what people had to do to survive. So the useful question about the review board is not whether the reviewer believes her concerns. She probably does. The question is who becomes exposed if the gate opens, and what pressure further upstream made holding it shut the rational thing to do.