Essentials of resilience, revisited

Essentials of resilience, revisited

This is a chapter by David Woods from the Handbook on resilience of socio-technical systems where he revisits his list of what things are essential to resilience and expands it to include two more elements.

Some of it may sound familiar. I find it can help to hear more about the component parts, as sometimes they can be subtly, but importantly different. Also, being able to think about these essentials, helps us evaluate our own systems. How poised to adapt are they? Are we supressing or encouraging initiative? What about reciprocity?

The essentials are:

• Poised to Adapt
• Brittleness
• Trade-offs

and the newly added:

• Initiative
• Reciprocity

Poised to adapt

Resilience can be said to be “potential for future adaptive action when conditions change.” This means systems will face surprise, that is guranteed. How it faces them will vary. To be able to change in the face of surprise requires some ammount of prep work in advance of the surprise.

Woods reminds us that adaptation isn’t always about changing the plan or model, but is the potential to do so if a change would better fit the new situation or environment. Likewise, adaptive capacity doesn’t mean a system is always changing, but is recognizing whether or not it makes sense to keep working or planning in the normal way.

Woods highlights for us the definition of adaptive capcity, interstingly the only definition called out in such a way:

Adaptive capacity is the potential for adjusting patterns of activities to handle future changes in the kinds of events, opportunities and disruptions experienced, there- fore, adaptive capacities exist before changes and disruptions call upon those capacities.

Brittleness

Brittleness deals with what happens to a system as it approaches the edge of its design for performance. A term that is used for this a lot is “performance envelope”.

Every system, no matter how much planning went into designing it, has boundaries in what it was designed to handle. It has places where the performance is likely to be consistent and places where that’s less likely.

This is because none have infinite resources nor do they operate in an unchanging, frozen world. As a result they all have a “performance envelope.”

Brittleness deals how a given system performs near and at that boundary. Does it suddenly fail? Or perhaps it fails more slowly, able to continue for a time?

Though we speak of the boundary lines in a very concrete way, where exactly they are is rarely certain and it is always changing because the environment the system operates in is changing too.

Some boundaries are fuzzier than others though, some sharper. But it’s not like a property line where we can have it assessed and know where it is for all time.

The opposite of brittleness is graceful extensibility. The term Woods coined as a mix of graceful degradation and extensibility (things that should be familiar to us in software).

If a system has low graceful extensibility then it will run out of resources or capability to respond as it faces increased difficulty and may fail very suddenly.

Conversely if a system has high graceful extensibility then it is able to anticipate problems and continue to function. Graceful extensibility is more than just “less bad” or less negative though. Systems that have high graceful extensibility may even be able to take advantage of new opportunities that present themselves as a result of a challenge.

Trade-offs

All adaptive systems face trade offs. They could be things like:

• Acute vs chronic goals
• Safety vs production
• Efficiency vs thoroughness

or others, but they’re always present. How these trade-offs are managed is what influences whether or not a system can “demonstrate resilient performance.”

This is where questions of how organizations should respond to warning signs occurs. If organizations always respond to any sort of warning sign by stopping production or otherwise sacrificing some short term goal, then they are unlikely to be able to achieve their goals and continue as an organization. But on the other hand, if warning signs are ignored all together, then the organization is likely taking on much more risk than is intended. The later can occur especially when there are signs of success previously.

This is also where “sustained adaptability” comes into play. The ability to adapt to multiple changes over time and cycles, not just one or many changes now. This might even mean adapting how the adaptations take place.

Initiative

This is a new one and Woods gives two points right off the bat to explain why it’s needed.

1. Some part of the system (a person, group, department, etc…) must have some amount of initiative if it is to contribute to graceful extensibility.
2. How the different roles and parts interact affects how that initiative is used or displayed, and that is influenced by the potential for surprise in a given area.

He gives the example of the military, where when the potential for surprise is high, the interactions between roles supports initiative. This is also true of ERs and likely true in high consequence incidence response (that you dear reader may have participated in).

There is a lot of research about initiative, but much of it is social sciences and other studies of human systems. Not much exists for “engineered systems”. So Woods defines initiative as having 3 parts.

1. The ability to adapt when a plan doesn’t fit anymore, at least from the perspective of the person or group who would be doing the adapting.
2. “The willingness (even the audacity)” to change the activities that were planned in the face of roadblocks or to take advantage of opportunities that arise that better fit the overall goal.
3. When initiative is taken, the person or group adapts by itself, using the information available at the time, without waiting for explicit permission (and without asking for it).

Initiative matters in our “human adaptive systems,” because initiative itself is very similar to an ability to adapt, especially when considered with the need to anticipate.

This is in contrast to just doing something by the book or following a plan without any anticipation or changes.

But there can be something like too much initiative or the initiative taken can be too broad. If this happens, then we run into one of the other ways that adaptive systems fail, working at cross purposes.

Initiative can also be destroyed, especially by pressures to do things by the book or exactly as planned. The difference between where one might occur instead of the other can be determined by how the risk of surprise is perceived.

Again using the example of military action or emergency medicine, “initiative is pushed down to the unit of action.” This is what allows for additional adaptive capacity to be obtained, even beyond whatever was made for in the plans.

On the other end of the spectrum, is where the risk of surprise seems low, where the world appears to be stable. Organizations that operate here are more likely to invest a high degree of trust in the models and as a result explain bad outcomes as a failure to follow the plan.

This means that the role of resilience engineering is to find some system or architecture that can help balance the potential for surprise, as it changes, against the use of initiative.

Reciprocity

Here we also learn from social science, taking those lessons and applying some bits of control engineering, mixing in what we learned of brittleness to see that reciprocity is an essential component of resilience.

Reciprocity is when one part of the system or “unit” (perhaps a person), relaxes their immediate goals, sacrificing something in the process (time, resources, efficience, etc) to help another achieve some larger goal for both. This is done with the understanding that the second person will do the same at some point in the future.

Reciprocity isn’t guranteed of course. The second person or unit could choose not to do the same in the future. Also, the first could decide since the sacrifice needs to happen now, but the payoff is fuzzy and later, that it isn’t worthwhile. Of course when this happens, risk of saturation increases. If reciprocity across units isn’t taking place, then ability to respond decreases. If no other lever can be pulled, no other intervention or resources called up, then system will approach it’s boundaries, where brittleness will come into play.

Takeaways

• The essentials of resilience with the two new added are:
  • Poised to adapt
  • Brittleness
  • Trade-offs
  • Iniative
  • Reciprocity
• These are things that we benefit from understanding, encouraging, or managing. Not all of them can we or do we neccessarily need to create in order to foster resilience.
• Some of the items, like iniative, can be created or encouraged, but can also be destroyed by things like pressure to work by the book or according to a strict rule set.
• Iniative can be applied too broadly, when this happens you have one of the ways that adaptive systems fail, working at cross purposes.
• With the addition of iniative and reciprocity, we can learn a lot from social sciences where some of these things may seem more obvious as they’ve been working in these areas for a long time. Of course initiative is important in human systems.