How to Observe Software Systems (Quality Software, Vol. 3)writing

measurementself-publishedsoftware-managementobservation
2010-01-01 · 1 min read · Edit on Pyrite

The third volume of the Quality Software series on leanpub, published in 2010, addresses what Weinberg calls first-order measurement: the foundational skill of learning to see what is actually happening in a software project, as opposed to what is reported, assumed, or wished. The material draws heavily on qsm-vol2-first-order-measurement-1993, reorganized and distilled for the Leanpub audience.

The volume's premise is that most software management failures have an observational dimension — managers and organizations act on models of reality that are significantly wrong, and they persist in those wrong models because they lack the discipline or the methods to check them against actual evidence. Weinberg argues that observation is a learnable craft, not simply a matter of adding more metrics or reporting requirements.

First-order measurement in Weinberg's framework means measuring what is directly observable — what people actually do, what artifacts actually contain, what events actually occur — before inferring higher-order constructs like "team productivity" or "project health." This discipline of grounding inference in direct observation connects to Weinberg's broader general-systems-thinking commitment to distinguishing map from territory, and to the quality-software-management-framework's insistence that you cannot manage what you cannot observe.

The treatment of observation here also has a human dimension. People in organizations respond to being observed and measured, and those responses are themselves data. Weinberg addresses the dynamics of observation in organizational contexts — how to gather honest data in environments where honesty may feel risky — connecting to themes developed more fully elsewhere in the Quality Software series and in secrets-of-consulting-1985.