Part 7/13:

He uses a simple analogy: a knife that cuts well for the first ten apples but then breaks or becomes dull is of good quality but not reliable. In a battlefield, one would prefer a reliable sword—a tool that consistently performs over time, not just in a single instance. Extending this analogy to data, reliable data must maintain its quality over sustained periods, supporting ongoing decision-making.

Reliability adds a temporal dimension: quality + time = reliability. For data to be truly dependable, it should be consistently accurate, complete, and correct over time, not just at a specific moment.

Building a Culture of Reliable Data

Changing perspectives is essential. Sesh advocates for treating data pipelines as data factories, with three distinct phases: