The elegant, but unspoken solution
A quick write-up about the interlinking between engineering value, high performance teams and psychological safety
So far, throughout my career, I’ve been a software engineer, systems engineer, site reliability engineer, principal engineer and finally, engineering manager. I’ve built new user interfaces, checkouts, ansible definitions, and Kubernetes clusters and, more recently, been embedded in a large organisation in a team dedicated to improving the reliability of the checkout experience.
This sounds very impressive, but let me assure you — it’s a journey that’s been absolutely littered with failure. As a software engineer, I broke whole shops; as a systems engineer, I deleted critical data. As a site reliability engineer in a major production issue, I’ve been wrong, with strong insistence, while millions of dollars slipped away. I’ve caught Bitcoin miners running around my (non-production) systems, and I’ve soured relationships with colleagues. All of this is to say I’ve struggled quite a lot. As Site Reliability Engineer, I’ve also been exposed to many other teams' struggles.
Over time, I’ve developed a preoccupation with failure. This has led to the study of failure, which unexpectedly led to the study of success. There have been invaluable lessons here, but I want to discuss one on working with people.
On the value of engineering
Software development work's fundamental value is making users' lives easier. There are different ways to do this, but in my experience, the most remarkable deliverables result from some unique insight that understands the customer perspective, the product vision and the technology and combines them in some new and innovative way. Understanding (for example) that when buying clothes online, they struggle to understand their sizes but that recent gains in computer vision can be leveraged to make that easier. Or that customers are happy to watch a subset of content instantly if they do not have to leave the house.
To make these more immense capabilities available to customers, there still needs to be a lot of work done, but in each deliverable, there is often an insight that allows a lot of work to be done more efficiently, rather than deploying on Kubernetes, deploying on AppEngine. Rather than using Java, use Go or Python. Rather than using Redis as a document store, use DynamoDB. Small efficiencies lead some teams to substantially outperform others in delivering that value.
Insight comes from a few people with a large amount of context and the ability to combine that context in new and exciting ways — often in discussion with others. However, there can be a bit of a gap between how these colleagues gain and leverage their insight and how their leadership views the same work.
The decision-maker disconnect
Within any given organisation, there is frequently a set of people who hold substantially more power than others. They can be more formally allocated that power (for example, engineering management), or they can have it due to expertise or reputation (staff or site reliability engineering). Because of that power, these colleagues can either make or strongly influence decisions directly.
These colleagues all have some notion as to how the organisation works. They might have either explicitly prescribed it or might be imagining it. They use this understanding as a precursor on which to make decisions.
One of the more surprising things from the “human actors” research is just how disconnected a decision-makers model of how the org work is from how it works. Sometimes, the overlap between how the leader thinks the organisation gets work done and how it gets delivered is minimal! In practice, there are four ways to view “work”:
Work as imagined
Work as prescribed
Work as disclosed
Work as done
Only the last case is how work actually gets delivered.
Work as (not) disclosed
One of the challenges that decision makers or leaders within an organisation will face is the power dynamic between a decision maker (a “high power” person) and the person taken with executing that decision (a “low power” person) means that lower power colleagues can have a much more comprehensive range and more significant set of adverse consequences if they challenge that high power person. This could be as simple as losing esteem in the eyes of the decision maker or as complex as being criticized by that decision maker for disagreeing with their perspective.
As colleagues on the execution side of decisions deal with a broader range of leaders, they tend to encounter leaders who are more punitive in their approach and thus start to tailor their information to minimise the chance of that leader being unhappy. The more extreme the power differential between the decision maker and the executor, the more likely an executor will tailor their information to benefit that decision maker.
For an organisation that is predicated on insight, this can be disastrous. At best, a colleague may not contribute their ideas on how to solve a company goal. Still, at worst, the colleague will contort themselves into agreeing with the decision maker. The colleague views the work through a highly optimistic lens (“if everything goes right, I can make this work”), which, given that everything invariably doesn’t, leads to missed expectations and general unhappiness.
Safety in adversity
In his article “Why do things go right”, Sydney Dekker highlights several critical properties of organisations that are more successful:
Diversity of opinion and the ability to voice dissent
Keeping a discussion on risk alive
Deference to expertise
Ability to say stop
Broken down barriers between hierarchies and departments
Not waiting on audits or inspections to improve
Pride in workmanship
The challenge of any decision-maker (or anyone responsible for the management of people) is to try and figure out how to encourage diversity, dissent, discussions of risk, the ability to say stop and so on. The decision maker must first make the environment psychologically safe to promote these behaviours.
Psychological safety is:
the belief that you won’t be punished or humiliated for speaking up with ideas, questions, concerns, or mistakes. At work, it’s a shared expectation held by members of a team that teammates will not embarrass, reject, or punish them for sharing ideas, taking risks, or soliciting feedback.
Leaders within an organisation influence psychological safety through their actions. Whether by encouraging discussion or punishing dissent, leaders set the tone for what is tolerable within their organisations. Correspondingly, this leads to those insightful engineering outcomes — or late projects.
A safe space
In his article “Psychological Safety in operations teams”, John Looney recommends concrete ways leaders can make their organisations psychologically safer. These include:
Creating space for people to take chances
Making it obvious when the team is doing well
Making your communication clear and your expectations explicit
Making your teams feel safe
In my experience, the most critical habit here is “listening with the intent to understand”. I’ll ask questions about how colleagues feel about their work in 1:1s, their primary challenges, and how they feel it fits into the strategic whole. I’ll listen, try and restate their perspective and ask them to confirm it before we proceed. Then, I’ll try to answer that question using their language and reconcile it against what I’ve seen and what other stakeholders might be considering. This gives them a much larger context to operate with and the ability to ask further clarifying questions.
In broader environments, we can set this expectation through example, asking questions that are deliberately naive so as to set the expectation other questions also have this capability. We can ask for input from our colleagues, with the explicit direction we’re asking them because of their perspective — even if they do not share (or even fully understand) what else is happening. We can listen and be kind in our interactions.
Lastly, there are environments in which more conversation is inherently permissible. Going for a walk, meeting at a conference or meetup, having a glass of wine after work or deliberately manufacturing a different context to discuss a topic can make it easier for people to take more risks.
By taking these actions, decision-makers can encourage their executors to take more chances when proposing, discussing or contributing to a group discussion. More discussion allows a greater diversity of opinion and dissent and the ability to recognise and surface expertise. This, in turn, will enable us to manufacture insight.
Safety with limits
The only caveat to an organisation that encourages dissent is that such an organisation, with controls around decision-making processes, can avoid getting stuck in analysis paralysis. In my experience, while decisions should be discussed freely, there comes a point when a leader needs to make and own a decision.
This is made a much smoother process if there is a way of retrospecting on previous decisions made, as well as learning whether or not there was a different decision that could be made in future. This allows dissenting colleagues to “disagree and commit” and for either that dissenting colleague or others within the organisation to learn from the results of previous decisions.
In Conclusion
My career so far has been varied but pockmarked with failure. Most recently, I’ve made failure my study of choice, and through this, I have encountered the analysis of success. This has led me to understand the value of engineering being fundamentally measured in customers' happiness, which is most easily found through some unique insight into an existing set of problems and technologies that can be applied in a new way.
This can be challenging to implement within organisations as decision-makers can become disconnected from those that execute their decisions. The power dynamic within an organisation means that unless they’re invested, those leaders might never discover how what they expect doesn’t bear up against reality, and what happens is much more fraught with risk. Some organisations are routinely more successful, and those organisations prioritise discussions, have diverse opinions and allow dissent. To turn our organisations into these successful examples, we need to cultivate an environment of psychological safety that will enable colleagues to raise their diverse opinions or participate in discussions.
If we do this, we’ll have a much larger potential space to find our fundamentally elegant insights and a group of peers able to understand and execute them, delivering on that most valuable improved customer experience.