Learning Culture Pattern
DORA research consistently finds that a generative learning culture is one of the strongest predictors of software delivery performance. Teams that experiment, share failures openly, and invest in continuous learning outperform those that don't — not just on cultural metrics, but on deployment frequency, lead time, and CFR.
What a Learning Culture Looks Like
A learning culture means:
- Blameless post-mortems — Incidents are learning opportunities, not occasions for blame
- Psychological safety — Team members raise concerns without fear of punishment
- Experimentation encouraged — Teams try new approaches and share what they learn
- Knowledge flows freely — Learnings are documented and shared, not hoarded
- Failure is expected — Complex systems fail; the goal is to learn and recover quickly
The Fawkes Dojo Learning Path
Fawkes implements a structured belt-based learning curriculum inspired by martial arts dojos:
| Belt | Focus | Duration |
|---|---|---|
| White | What is an IDP? DORA metrics fundamentals | 1–2 weeks |
| Yellow | CI fundamentals, Jenkins pipelines | 2–3 weeks |
| Green | GitOps with ArgoCD, Kubernetes | 3–4 weeks |
| Brown | Observability, SRE practices | 4–6 weeks |
| Black | Platform product thinking, advanced IDP | 6–8 weeks |
Each belt has a module, exercises, and an assessment. Belt-holders are expected to mentor those at lower belts.
Communities of Practice
Fawkes communities of practice (CoPs) bring together people with shared interests across team boundaries:
- Platform Engineering CoP — IDP components, Backstage plugins, ArgoCD patterns
- Security CoP — Threat modelling, security tooling, compliance
- Observability CoP — SLOs, dashboards, alerting best practices
- Dev Experience CoP — Developer productivity, golden paths, friction logging
CoPs meet fortnightly, post to Discourse, and maintain a reading list in the platform wiki.
Blameless Post-Mortems
After every P1/P2 incident, the on-call team writes a blameless post-mortem within 48 hours. The format:
- Summary — What happened and what was the user impact?
- Timeline — Chronological sequence of events
- Root cause — The contributing factors (avoid "human error" as a root cause)
- Action items — What will prevent recurrence?
- Lessons learned — What did the team learn?
Post-mortems are stored in docs/runbooks/post-mortems/ and shared in the weekly
engineering newsletter.