The Things I Learnt about DevOps When My Car Was Engulfed by Flames

unnamed

This is a true story, based on a talk from DevOps Days London 2016:

It was a gorgeous sunny spring day, my family and I were driving through my home town of Bristol, ready for another weekend adventure. We were cruising along when my wife said quietly: “I can smell smoke”. Now, I’m a good mechanic, I used to restore classic cars. The car we were driving was modern and recently serviced. I checked the instruments, everything normal. “It must be outside”, I declared confidently. Two minutes later tentacles of smoke were curling around my ankles and my shins were getting remarkably warm.

We can learn something relevant to DevOps, and many other disciplines, from this experience… read the rest on InfoQ

 

Change, Disparity and Despair

Models? Meh.  We all know Box’s quote: “essentially, all models are wrong, but some are useful”, a quote generally heard seconds before someone presents an alternative model to the one you’ve just put forward.  However there’s an undeniable utility to models and diagrams, the way they convey concepts in a fashion that people can quickly understand and start to explore together.

One of my favorite diagrams, and theme of this post, is David Viney’s ‘J curve effect observed in change’.  Change in this instance is just about any planned change that impacts an organisation.  This is a distinct, but close relation to the Kubler-Ross individual change curve.

This J curve aims to show that for any desired improvement in capability (or fitness to achieve some purpose) there will be a decline in capability before there is an improvement.  This is virtually impossible to capture and graph accurately, but we can talk in general terms about transitions, duration in a state and trending.

Annotated J

The Danger Zone
When introducing Kanban David Anderson points out that time and depth an organization is comfortable in the trough reflects it’s appetite for risk.  Push change deeper or for longer, and the organization’s appetite risk is exceeded. End result: change agent gets fired.  Of course this danger zone applies to just about any substantial change, and practitioners of Scrum, DevOps, DSDM and so forth should be just as wary.  A further observation is that if a change is halted once in the trough things don’t magically return to the start state, and a second iteration through the curve is required – starting from a point of reduced capability.

The swan song
In a recent talk I added a hump near the start of the transition.  That’s the point where people hear about an incoming change.  Sometimes, if the instinct is to resist, this inspires efforts to prove original methods can and will work.  Alternative practices are implemented with renewed diligence, energy and fervor often leading to a short term uptick in capability.  This reinforces arguments that change is unnecessary, but will not yield the desired improvements in the long term.

Change Disparity?
While the J curve represents an organization’s progress during a period of change, it makes the assumption that people are moving – or adopting – at more or less the same pace.  In fact this is seldom true, and different adoption rates lead to significant gaps in understanding and approach.

This ‘change disparity’ hampers collaboration and can be as damaging as any silo or clique.

For simplicity let’s consider early and late adopters.  The reasons for being in those groups may vary greatly: work allocation, meddling by people with influence, environment, personality, good or bad luck.  Unaware of circumstances both groups get frustrated. There’s a temptation to say the late adopters are at fault and should hurry-up, but running too far ahead and expecting people to keep up, or ‘just get it’ regardless of circumstance seems no better.  This is common with technology zealots, characterized by a disparaging attitude towards people not using their tool of choice, or voicing concerns about it.  Of course this reaction actually discourages adoption, and serves to hinder the change they would like to bring about.

The awfulness of the situation reminds me of the Inuit game ‘Ear Pull’ in which two players face each other, linked by a string around their ears, and pull.  In opposite directions. You can almost feel the pain in this clip.

Ear Pull Leroy

Note the string does not cause pain by itself  – it’s pulling in an opposing direction, forcing another to follow at a rate they are not comfortable with.  If both players agreed a distance, or form of feedback, they could move together without discomfort.

This is something to consider when introducing change, new tools or ways of working.   This adoption gap, or Change Disparity, is easily overlooked but potentially damaging.  There are numerous solutions, but it all starts with recognition of the problem, when rates of change are outside productive limits, and willingness to do something about it.

Before we learn, must we first unlearn?

Sometimes I read something and think, this is awesome, how did I miss this one? Sometimes, I even get carried away and write more than 140 characters about it…

One such article explores the concept of ‘unlearning’, as a precursor, or catalyst, to learning. Learning feels like a common denominator across agile methods. But agile is not just about learning how to get better at building stuff, it’s about learning how to introduce and encourage change.

The article is ‘Unlearning Ineffective or Obsolete Technologies’ by William H Starbuck, currently a visiting professor at Cambridge. The article is an absolute goldmine, but Starbuck is also remarkable for having a CV that has to be one of the most simultaneously impregnable and impressive of all time.

The abstract grabbed me straight off: “Often before [people] can learn something new, people have to unlearn what think they already know.”

We’re familiar (and often lazy) with concepts like keeping an open mind, and perhaps techniques like DeBono’s thinking hats which invite other perspectives. Deliberate unlearning though, seems counter intuitive and somewhat destructive, especially if the ultimate aim is to learn more.

The article is packed with great antidotes to reinforce the messages, from how a navy spent weeks bombing aquatic mammals they believed were submarines to exploding steam boats and Sony Walkman development.

Frankly I’d recommend you stop reading this now, and read full the paper, but if you don’t have the time available, allow me to offer a summary:

Starbuck suggests that there are three key points to recognize in order to encourage learning.

1.Learning cannot happen without unlearning – current beliefs are blinkers, something is required to demonstrate that people should no longer rely on their current beliefs and methods – “Expertise creates perceptual filters that keep experts from noticing technical and social changes”
2. Organizations make it difficult to learn without first unlearning. Policies and practice are often created from individual’s beliefs, and these polices mesh together to form a kind of structure, in which it is difficult to change a small part. This creates a self-perpetuating situation discouraging change, where it becomes hard to change anything without dismantling the whole system.
3. Unlearning by people in organizations may depend on political changes. I think the key point here is that unlearning may need to be enabled by people changes. The motivation may be political or something more mundane, the change in influencer is the significant part. This is because information is interpreted by people, influential people create ways of working, culture and policies. Any modification to these may be seen as a threat to the individual and suppressed, rather than exploring suggested change. Starbuck suggests this is why senior managers are prone to overlook, and miss-interpret bad news.

I hear things that support these views time and time again, phrases like “our Agile culture was going no where until so-and-so joined, or so-and-so finally left”. Other disruptions seem to foster unlearning – particularly stronger collaboration and a better appreciation for the challenges of other teams, something very visible in the DevOps movement.

Starbucks goes on to identify methods, or viewpoints, to encourage unlearing.

Dissatisfaction – A common reason for doubting, and reconsidering current approaches, he observes that this can take a long time, presumably requiring a high level of discontent before people are motivated to seek change.

“It’s only an experiment” – There is a mind trick that goes on when we are in experimental mode; we take calculated risks, and we are more observant, we want to evaluate outcomes, rather than preferring a particular one. Often there is less to loose if the results aren’t as predicted. Starbuck puts it: “[Experiments] create opportunities to surprise”. As a side note, Cynefin recognizes the value of this, and promotes safe to fail experiments, nice post here.

“Surprises should be questions marks” – In other words when something surprises us, we should not dismiss it, or categorize it as an interesting anomaly, but look to see if it challenges any of our beliefs or assumptions.

“All descents and warnings have some validity” – Starbuck admits that this is a little over zealous, and there are sources of dissent that don’t provide value, never the less in many cases there is something to gain. Often these comments are attempts to warn or inform, and merit attention.

“Collaborators who disagree are both right” – or rather, there are elements of truth in both arguments. In these situations the art is discovering how the seemingly contradictory elements can both exist. This doesn’t mean creating a compromised win-win situation. It means challenging assumptions and seeking new models until there is understanding.

“What does a stranger think strange?” – Strangers haven’t been exposed to, or adopted, your ways of working, and therefore are more likely to challenge and make valuable observations. In my opinion this is yet another reason to pay close attention to new hires, especially if they are new to the industry or fresh from college.

“All casual arrows have two heads” – If I’ve interpreted it correctly, this indicates that we should change the way we consider flow, by recognizing that there are two directions for each path and we should seek out overlooked feedback routes. Starbuck illustrates with a great example: Mass vehicle manufacturing was once be based on accumulating inventory. Materials were shaped into components, components into cars and customers selected cars from the vehicle inventory. That’s one direction for a causal path. Taiichi Ohno saw the opposite direction and created Toyota’s just in time system, where the absence of inventory to serve customer demand stimulated flow.

“The converse of every proposition is equally valid.” – This pithy phrase is almost immediately caveated to indicate that not all propositions have a valid converse. I guess the aim is to train ourselves to explore the converse, a neat method of flipping our perspectives. Are leaders really leading their people, or just servants to them?

Summing up then, Starbuck puts forward a set of useful techniques to help us overcome our inherent biases and tendency to filter what we consider threatening or bothersome. Even if you don’t agree with the techniques it’s a useful reminder, and the goals are worthwhile. These techniques may avoid some other more catastrophic event, like being fired or going out of business, being the trigger for unlearning. The term unlearning is convenient but perhaps a misnomer, nothing is discarded, it is more a recognition that current beliefs, ways of working ,or processes, no longer serve us; that it is time to seek alternatives.