Looking in more detail at joint activity, we find it is made up of multiple joint actions. Those joint actions each have phases and a “choreography”.
That choreography can also be changed or influenced by whether or not the parties can signal to each other and use coordination devices. Coordination cost is the burden on each action that occurs, high or low, because of the participants trying to choreograph their moves.
One way to understand the choreography of joint action is to look at the things that make it up. The three parts of joint action are entry, a body of action, and exit. Additionally, joint actions can nest within each other, but this simply a way of further breaking down and in analyzing how it is that the teams coordinate. I think knowing this and having this lens to view our own coordination efforts can be helpful.
The synchronization of entry and exit points of the different phases in complex activity is a major challenge.
Scripting action and choreography
These phases can be pretty much scripted in many situations and not at all in others. For example, if there are legal constraints then it may be very scripted. Or, going back to the surgical team, how they exactly do this surgery is perhaps less scripted. The authors give the example even less scripted work being a design team making a new company logo.
It’s important to note that the entire activity can be scripted to varying degrees as well as each of the individual phases. I think that’s an important distinction as well, often we might have a very complicated activity to do, like responding to an incident, but certain procedures might be very scripted. For example, determining the health of a node in the system.
As we are likely familiar with as software practitioners, the authors touch on this idea that there are a few different motivations for standardizing procedures: the desire to prevent interactions where they would close off later choices and also the ability to have better predictability anticipate others.
No matter how scripted we make things though, we will always still need to make adjustments and clarifications. Perhaps even to drastically revise the plan as a situation proceeds.
How does signaling work?
How the joint activity evolves can change dramatically based on the way the different members of it are able to signal to each other. Signaling is just a label to encompass all the ways that the individuals might communicate information to each other. They might signal what they intend to do, what they’re having a hard time with, what they wish other people would do anything like that.
The authors use the literal example of signaling in a car to show how important it is, specifically on the freeway.
It’s important to note that signaling is only successful if other people noticed the signals.
So if another team member signal something but no one notices then that signal did not work. Signaling also comes with a responsibility for the individual in the signaling to determine how interruptible the other people are.
We’ve likely all experienced a mismatched judge of this, perhaps someone asking for updates when there are none available during incident while you’re trying to do a critical task. If you’ve experienced this, then you already know that careless signaling, without looking at what someone else is doing can cause performance to degrade.
We touched a bit on how different people do this in the voice loops issue
. Mission controllers would listen to what other people are listening to and then judge what they had to say against what was being worked on.
Coordination devices are not physical devices, but different ways and mechanisms of signaling that participants can use to help make themselves more predictable. Some typical signaling devices are: agreement, convention, precedent, and situational salience.
Of these, I found salience to be the most interesting.
“salience has to do with how the ongoing work arranges the workspace so that next move becomes apparent within the many moves that could conceivably be chosen.”
The authors use an example of surgery where exposing a certain part of the body can very clearly communicate what steps can be taken next.
I was brought back in reading this to my time in emergency medicine with scenes where you may notice certain gear present or absent; in a state of readiness or not.
For example, if you didn’t even see a gurney nearby and you were not the first on scene, you might know that the patient had been leaning towards non-transport or perhaps it was interpreted to be less serious since they weren’t immediately getting someone on a gurney. Also, lack of a backboard might indicate lack of a suspicion of trauma, and indicate more toward a medical cause.
The authors also interestingly note that salience is likely to be the main mode that long-standing and highly practiced teams used to coordinate
An important concept that the authors that is the joint action ladder where the rungs from bottom to top are attending, than perceiving, then understanding, and finally action.
A signal must be attended to by the recipient, then perceived, then understood, and then acted upon.
Accomplishing these allows them to have “climbed the ladder”. Typically it can be possible to skip rungs as higher once can take care of lower ones, but going the other way can be dangerous. If I assume that someone is understanding, but they didn’t even attend to the message this can be a problem. But if they demonstrate their understanding, then I can pretty safely assume that they have attended to it.
As common ground degrades, coordination costs rise. So then that means that the more effort spent in improving and building common ground will allow more efficient communication, less involved signaling, and lower coordination cost in the moment.
Conversely, if time or other resources try to be saved by not investing in signaling or coordination, as a way of reducing these costs, then break down becomes more likely.
Loss of common ground
It’s important to realize that no matter how much effort we put in breakdowns are going to happen.
“no matter how much care is taken, breakdowns in common ground are inevitable: no amount of procedure or documentation can totally prevent them”
I think this is really important for us as software practitioners. Often times, lists of action items are generated that presume this is the case. Or things will be said like,“if only we had more process or documentation”. Sure, sometimes these things do help, but that isn’t always the case.
Anytime common ground is being built up in some ways, it is also being lost in others. This is because each person can be interpreting the data available differently and have different access to it.
Common ground is lost partly because a situation is dynamic and as it unfolds they are making different interpretations draw different conclusions about what is happening what others know and what they’ll do next.
“Baseline state is one in which people are detecting and repairing problems with common ground — and not attempting to document all assumptions and duplicate the contents of each person’s mind.”
The authors’ research showed that people lose common ground mostly for the following reasons:
- Team members may not have experience working together
- The might have access to different data
- They may not understand why the leader is giving the directives
- They might be unaware that they have differences in stance (e.g. some team members could have a higher work load or other, competing priorities)
- Their communication could be disrupted suddenly and they may not have the skill to fix it
- They may fail to monitor confirmation of messages and get confused about who knows what.
The author’s noticed that as they were looking through all these different forms of communication and coordination: driving, coaching football, etc. they kept seeing a certain type of breakdown, so often that they named it the “fundamental common ground breakdown”. The last one, confusion over who knows what, is the fundamental common ground breakdown. It can be shown in four parts.
- Person A thinks Person B know something
- Person B doesn’t know this and doesn’t know they’re supposed to know it
- As a result they don’t request information
- That lack of request gets interpreted by person A as person B having the knowledge
They of course fail to notice and resolve this mismatch, which then as the event goes on creates a cascading set of incorrect assumptions that cause problems. Typically, this will be discovered when they experienced surprise, in this case a “coordination surprise”. When something occurs that doesn’t make sense to them anymore based on what they believed and what they thought they knew and as a result, now they will the digging deeper.
- Joint activity is made of phases, coordinating the beginning and end of those phases is especially difficult
- Long standing teams use different signaling devices than newer teams
- Their communication also becomes more dense and coded
- Degradation of common ground causes coordination costs to rise
- Having more procedure isn’t going to prevent some amount of breakdown