Learning that Transforms Systems
David Leigh
Embedded Files
“Resilience requires recognizing when our model is wrong—and being able to revise it quickly.
“Resilience requires recognizing when our model is wrong—and being able to revise it quickly.
- David Leigh
- David Leigh
Listen at:
In this episode with return guest David Leigh, we explore one of the most important pillars of Season 3: how learning transforms systems - and more specifically, how resilience is built not through tools or processes alone, but through the interplay of diverse perspectives, psychological safety, and evolving mental models.
David shares stories, examples, and insights that apply equally to engineering teams, organizations, and society.
Key Insights
1. Resilience = Responding to Surprises
1. Resilience = Responding to Surprises
David frames resilience as the ability to respond effectively and adaptively to surprises that challenge our assumptions about how systems work.
2. Diversity of Perspective Is a Technical Requirement
2. Diversity of Perspective Is a Technical Requirement
From high‑pressure incidents to release planning, diverse mental models allow teams to detect risks earlier and resolve crises faster.
IBM’s GitHub Enterprise SRE team routinely paused incidents to gather different theories
Quiet voices often contribute critical insights
Collaboration improves when responders understand each other’s mental models
David reinforces: resilience is a team sport.
3. Psychological Safety Makes Learning Possible
3. Psychological Safety Makes Learning Possible
Teams must create a culture where:
People can state what they don’t know
Experts recognize the limits of their expertise
Quiet contributors feel safe voicing concerns
Feedback is framed as a path to improvement, never as blame
This openness is the antidote to overconfidence and blind spots.
4. Mental Models Have a Half‑Life
4. Mental Models Have a Half‑Life
Systems evolve. Teams change. Environments shift.
Therefore:
Expertise expires
Automation based on old runbooks becomes dangerous
Assumptions quietly degrade
David encourages us to assume:
“Our mental models are always wrong. The question is how wrong, and in ways that matter now.”
5. Pre‑Mortems: A Low‑Cost, High‑Impact Learning Tool
5. Pre‑Mortems: A Low‑Cost, High‑Impact Learning Tool
One of David’s strongest recommendations:
Before a major change, gather a diverse group
Pretend the launch has already failed
Ask everyone to write all the reasons why
Group, rank, and discuss
This practice surfaces hidden risks, gives voice to quiet experts, and builds shared understanding.
6. Automation Isn’t a Cure—It Creates New Challenges
6. Automation Isn’t a Cure—It Creates New Challenges
Automation reduces toil, but also:
Obsoletes human expertise
Encodes outdated assumptions
Introduces new failure modes
Creates “silent drift” between how systems work and how we think they work
Teams must treat automation as living code, not a static safety net.
7. Stop Chasing Recurrence Metrics
7. Stop Chasing Recurrence Metrics
David critiques the traditional ITIL notion of “preventing recurrences”:
Systems change too quickly
People rotate
No failure repeats exactly
Metrics incentivize the wrong behaviors
Better signals of resilience include:
Team confidence
Clarity on who to call
Psychological safety
Quality of cross‑functional learning
Things to listen for:
Why resilience is fundamentally about preparing for the unexpected
How mental models age, erode, and become outdated—and why acknowledging that is a strength
The power of diverse representation in incident response
Why psychological safety is the foundation for resilient socio‑technical systems
How practices like pre‑mortems and tabletop exercises build cultural readiness
The limitations of automation, metrics, and “preventing recurrence” mindsets
And why learning is not simply knowledge acquisition, but a force that shapes systems themselves
Page updated
Google Sites
Report abuse