This issue of Resilience Roundup is free for everyone! If you want to get more analysis, access to the podcast, and upcoming new features you can subscribe here.
Communication Strategies from High-reliability Organizations: Translation is Hard Work
This is an article by Emily Patterson, that while short, is a very useful overview of some of her previous work, especially her work on handoffs, some of which I’ve analyzed before.
“In human factors engineering, the basic unit of analysis is composed of practitioners supported by tools meeting the demands of a particular setting”
The primary functions of those practitioners that we are focused on is analysis and planning or replanning.
Those functions require communication because analysis and replanning are distributed, both across humans and computers.
She identifies 3 main heuristics that are useful when designing systems to help coordination.
3 heuristics for designing for coordination
Here the goal is to reduce complexity where possible. Patterson uses the example of shift change handoffs, where having multiple people from many specialties and changes in where the patient is all add lots of increased complexity. Minimizing these, where it doesn’t take an untenable trade off like very high financial or other cost is ideal.
For example (as detailed in Being Bumpable), patients were proactively moved as soon as possible after a procedure was done, so that they could avoid having a time where they were instead just waiting for a bed
Here the goal is to make hidden events and activities visible. If they are not visible then those activities that have a high cost of coordination cannot be found and addressed.
Once found, one way to address them might be to move from direct to indirect coordination.
The “traditional” way to make this shift is very familiar to us in software. Typically some sort of dashboard is built and displayed in some common space where it can be pointed to and referenced in discussions.
Of course, as we know (and Patterson warns), designing good dashboards isn’t easy. Also, this can be harder to do when people are not in the same physical space. Still, in those cases having some sort of shared reference can still be helpful, but it may not be as impactful in reducing the costs of coordination.
Some things that might be on these dashboards are:
- What activities others are doing.
- In our case, these “others” could be people or automation.
- Deviations from typical plans.
- or us it might be things like change freezes, derivations in “normal” metrics, etc…
- Stances of stakeholders around important decisions.
- This one doesn’t translate quite as well for us. Paterson suggests the “‘aside’ notes regarding family dynamics” from charts, but we don’t really have that. This could be a note from other operators though.
- Constraints and side effects from contingency plans.
- This could include things discovered from incident reviews or other investigations.
The goal here is to focus attention:
enable peripheral detection of unexpected events and actions while performing primary tasks.
This is exactly the sort of thing provided by voice loops in mission control.
This one is a bit harder for us to replicate with audio, but Patterson points out that it may be possible using visual. Audio would be ideal though, since we already have so much visually vying for our attention.
Translation, not copying, is critical
When working with teams, Patterson translated her knowledge of human factors into the domain, she didn’t just copy it over. While they could have just copied it directly, that would have overlooked differences in the domain and might have unintended consequences.
Resident physicians are trained very differently from mission controllers or ambulance dispatchers or say SREs. They have different technology available to them as well. While mission controllers readily have voice loops, telling a bunch of doctors that they should setup such a system but would be a bit ridiculous, not to mention onerous and expensive.
But instead, if you take that idea that one of the things voice loops help with is the ability to contact someone when you don’t know who you should be contacting exactly you can arrive at a solution that would work for doctors. In this case, one that’s also very familiar (and not necessarily ideal) for us, a team pager carried by whomever is the first contact for the team. This instead is cheaper and more feasible for the group to implement, while gaining some (or all) of the intended benefit.
That’s not say this is easy, after all, Patterson gave the article the subtitle of “Translation is Hard Work” for a reason.
- Knowledge from other domains needs to be translated into the one being worked in, not just copied over.
- If its just copied over it can cause opportunities to be missed or even make things worse.
- 3 heuristics can be used to help design for coordination:
- Reducing complexity where possible.
- Revealing hidden events.
- Focusing attention.
- Analysis and planning (or re-planning) require coordination since they are distributed, both across multiple people and computer systems.
- Some direct coordination can be replaced with indirect coordination artifacts like dashboards that can be referenced in conversations.
- There is a lot vying for our visual attention though, so it can be harder to make as much headway on lowering costs of coordination if you stick to only the visual.