Hook
State multiplies surprises.
Problem
Mutable state drifts over time. It creates hidden dependencies, makes failures harder to reproduce, and complicates recovery.
Why it matters
Less state means fewer failure modes and easier recovery. When state is necessary, explicit versioning and ownership keep it safe.
Signals you are here
- Configuration drift across environments
- Snowflake servers that cannot be replaced safely
- Rollbacks are risky or impossible
- Caches or shared state cause inconsistent behavior
Anti-patterns
- Manual changes directly in production
- Shared mutable data without ownership
- Hidden caches with unclear invalidation
- No migration plan for schema changes
Try this
- Use immutable artifacts for deployments
- Version configuration and data schemas
- Prefer stateless services when possible
- Make migrations reversible and automated
- Audit and track state changes
Example
A team moved from hand-edited servers to immutable images. Drift disappeared, rollbacks became reliable, and incident recovery time dropped.
Reflection prompt
Which state in your system is least understood? How can you make it explicit and versioned?
More like this
Heuristic
Increase Contrast, Not Volume
Prompt length does not guarantee novelty. Context contrast does.
Heuristic
Make Infrastructure Disposable
Cattle, not pets.
Heuristic
Blame the Process, Not People
Fix the system.
Heuristic
Fail Closed, Log Everything, Recover Gracefully
Safe failure beats quiet failure.
Heuristic
Test Where It Breaks, Not Where It Works
Test the breaks, not the breeze.
Heuristic
You Cannot Rely on People Under Stress
Design for tired humans.
