Looking to social science for help with metrics
Talkat CAST 2014 · on Wednesday, 2014-08-13· at 15:30· 10th fl Rosenthal Pavillion, Kimmel Center for University Life·
summary
Argues that software measurement borrows bad habits from early psychology, citing psychophysics experiments where subjects’ responses broke quantitative assumptions and exposed reliability and construct validity failures; recommends borrowing qualitative methods and Lean thinking and lists concrete flow metrics to track: work in progress, cycle time, lead time, touch time, slack, takt, variation, churn. Describes scripting deployments over three weeks to cut a build from 1 hour to 5 minutes, and urges using metrics as inquiry not control.
AI-written summary · about
chapters (12)
- 0:00Start
- 3:44Gustav Fechner
- 4:43Alfred Binet
- 6:22Titchner and the Torres Straits
- 7:34Reliability and Validity
- 8:34WH Winch working with averages
- 10:04The methodological imperative
- 12:38What is lean
- 19:46A philosophy of improvement
- 21:27Summary and open season
- 25:54Context driven measurement
- 34:28Traditional measurement, you've probably seen this before...
related talks
20mHow to Communicate Metrics to Build Engagement
41mRethinking Measurement for High-Performing Teams
18mFrom developer productivity to engineering impact
26mMetrics and Modeling – Helping Teams See How to Improve
43mThe Past, Present and Future of Developer Productivity Theories and Metrics
29mYour Metrics Are Lying: How to Stop Reporting and Start Learning
43mMeasuring Developer Productivity: The Good, the Bad and the Data
46mEngineering Intelligence via Engineering Metrics