Stevey's Google Platforms Rant

Amazon's 2002 mandate to expose every team's data behind a network service reshaped the company into a platform-first organization. The edict required no direct data access, forcing teams to build service interfaces, registries, and discovery mechanisms. The resulting SOA revealed hidden costs: pager escalations bounce through dozens of services, quota and throttling become security concerns, and monitoring turns into automated QA. These concrete learnings-service registries, universal debugging sandboxes, and the need for strict throttling-are hard-won insights for any leader scaling a technical organization. The author contrasts this with Google's slower adoption of platform thinking. While Amazon turned internal tools into public services like EC2 and S3, Google still treats most work as product-centric, rarely exposing stubby services for cross-team reuse. The lack of a unified platform hampers scaling, forces duplicate effort, and leaves teams without consistent interfaces, leading to fragmented engineering practices. For technical leaders, the piece is a cautionary tale: mandating services can create massive operational overhead, but it also builds a reusable foundation that fuels growth. Leaders must balance the discipline of service contracts with the engineering burden, invest in discovery, throttling, and observability, and champion platform thinking to avoid siloed codebases and hidden technical debt.