Optimizing for Time: Dark Matter
summary
Meta measures developer time and finds most work is overhead: one engineer used about 50 tools in a day, timelines compress into noisy bars, and a simple metric time spent coding per diff (time coding ÷ diffs) yields an example group TSD of ~21 minutes. Concrete fixes include autofocus (mute notifications while coding with an @notify override), reducing build latency, and a WTS workflow that groups events within 15 minutes and fills windows with tool time to map SEV resolution across a three‑day incident; an AI agent “SEV maze” then proposes root‑cause diffs.
AI-written summary · about
chapters (9)
- 0:00Introduction to dark matter in developer productivity
- 2:55Data analysis of an engineer's daily workflow
- 6:45Using data to improve developer productivity tooling
- 8:56Measuring efficiency with time spent coding per diff
- 13:23Analyzing workflow intent to reduce noise
- 16:15Measuring overhead with percent time coding metric
- 19:07Analyzing and optimizing time spent in meetings
- 22:08Framework for workflow time spent on specific tasks
- 27:31Measuring time by business value and future outlook
related talks
39mSPACE and Beyond - Measuring Developer Productivity
47mHow to win in software development with tighter feedback loops
47mHow to Optimize Non-Coding Time
46mDeveloper Productivity at Scale
02h 31mDeveloper Productivity Engineering - Extended Session
27mDeveloper Experience: It’s Time to Start Complaining
45mOne Metric to Rule them All: Cycle Time
46mLet Them Eat Code