Free — no signup required

Introduction: The Missing "Why

2 min read

Organizational Amnesia

Have you ever joined a new project, looked at a piece of code or infrastructure, and wondered, "Why on earth did they build it this way?"

This is one of the most common experiences in software engineering. You see a Kubernetes cluster running on bare metal instead of a managed cloud service. You find a hand-rolled authentication system instead of an off-the-shelf identity provider. You discover a monolithic deployment pipeline that takes 45 minutes when a parallel approach seems obvious. In each case, the what is visible, but the why is invisible.

Often, the people who made those foundational decisions have moved on. They took the reasoning — the constraints they faced, the alternatives they rejected, and the trade-offs they accepted — with them. What remains is the artifact (the code, the infrastructure) but not the intent.

This leads to organizational amnesia: the systematic loss of decision-making context over time. It manifests in two destructive patterns. First, teams become afraid to change things because they don't understand the original intent — "maybe there's a reason it's this way." Second, teams repeat past mistakes because the knowledge of why something was tried and abandoned was never written down.

To cure this, we need "organizational memory." We need tools to record not just what the architecture is, but why it is that way. The engineering world has converged on two primary tools for this purpose: ADRs (Architectural Decision Records) for capturing decisions, and RFCs (Requests for Comments) for proposing and debating changes before they happen.

Key Point: The cost of organizational amnesia compounds over time. A missing "why" on day one is a minor inconvenience. The same missing "why" after three years of accumulated decisions built on top of it becomes a structural risk to the entire system. Recording decisions is not bureaucracy — it is risk management. Studies of large engineering organizations consistently show that the time spent re-litigating old decisions (or recovering from reversing them without understanding their purpose) dwarfs the time it would have taken to write a one-page record at the moment of the decision.

This is one of 18 chapters

Every chapter is free — Kubernetes, Terraform, SRE, distributed systems, and more — with fast daily review built in.

Sign up free