W12 - Delivery Changes After Organizational Expansion and Antifragility

Last week an article in the company forum titled "Changes the Chinese Internet Will Undergo from 2025 to 2030" went viral beyond internal channels. As an organization grows, its responsiveness slows; I have firsthand experience with changes over recent years.

Four years ago, when hiring was tight, I had eight product requests on my "ones" queue at once, juggling throughput across them. Delivery processes were nimble, discretion was high, and communication paths were short, so I could flexibly adjust each request’s development, integration, and testing schedule. Mistakes happened in the rush and reviews were sparse, but most issues were controllable and minor — a hotfix would resolve them.

Today, a delivery environment that can run two requests in parallel is considered demanding. With larger organizational scale, stability and institutional knowledge have become higher priorities. Adding control gates, driving convergence, and building mechanisms have been the main themes in recent years. I attended a release plan review last week and the tone was essentially "nothing can go wrong." Many of these constraints applied to services of that scale offer poor cost-effectiveness. Most of our services are low-traffic yet represent the bulk of delivery work; taken together, some measures simply aren’t worth the expense. I used to repeat "stability first"—it reflected both job requirements and lessons learned—but now with more diverse perspectives we need a dialectical review. There are improvement ideas: tiering services, removing preventive controls that don’t meet expectations, and leveraging AI to boost efficiency.

I heard a fund manager describe strategies for profiting from risk in a VUCA environment, crediting the design to Antifragile. That book offers many practical perspectives that sharpen everyday decision-making. Many human-made policies, mechanisms, and regulations actually act as "fragility drivers;" complex mechanisms can trigger unforeseen chain reactions. The visible gains from "man-made complexity" are usually small and tangible, while the side effects can be severe yet invisible.