Physics · Dynamics
ExogenousStress
An external event or change in requirements that the current shape does not handle and that forces it to change
Definition
Mechanism
A system (software architecture, organism, organisation) has a current shape that copes with a certain set of conditions. An exogenous stressor is an event from outside that falls outside that set: a supplier fails, a power outage, a new pathogen, a changed regulation, a requirement nobody predicted. The system survives only if it can change shape fast and cheaply enough to absorb the stressor. Stressors are often unpredictable in probability but can still be imagined as possible, and a design can be prepared by working out which of its parts would need to change for each stressor. Exogenous stress is the external counterpart of endogenous stress, the internal load a system carries from being prepared for it.
Invariants
- The stressor originates outside the system
- Surviving it requires a change of shape, not only correction back to the current shape
- A stressor can be planned for as possible without knowing its probability
Risks
Failure modes
Only planning for stressors whose probability can be estimated, so stressors that are possible but cannot be estimated are missed
Treating survival of one stressor as proof of robustness to all
Relationships
Root pattern: builds on nothing in this library.
Used by
References
Mind
The internal load a system carries from being partly prepared for possible external stressors
Mind
An implementation that is at once one working model and a partial implementation of several others, so that some functions can be completed or removed with a small change
In a co-evolving environment an agent must keep changing just to hold its position, because the yardstick of fitness moves with every rival's change
Mind
Stop integrating stressor handling into a design once it survives a fresh round of unseen simulated stressors, or once further integration costs more than it buys; tested by experiment, not proven from a probability distribution
Reference
