Semprini

Lies, damn lies and statistics

Enterprise Anarchy

Now, I'm not quite sure if that's a click batey title or not. I guess I see it as subverting expectations, because you see dear reader, I'll be arguing for the proposition that anarchy is a good model of operating IT in a large enterprise.

Bollocks! I hear you roar, or possibly chuckle to yourself if I'm not engaging in hyperbole. But none the less, I expect that you think it's a case of another architect, up in their ivory tower smelling their own farts.

Now, lets see if I can extricate myself from this hole that I've dug shall we?


As a shrewd reader you'll realize - this Semprini chap has a trick up his sleeve and today I'll be relying on the common misconceptions about what anarchy is.

I think Noam Chompsky has an elegant description about the anarchy way of thinking: "Anarchism is a tendency in human development that seeks to identify structures of hierarchy, domination, authority and others that constrain human development and then it seeks to subject them to a very reasonable challenge - justify yourself. Demonstrate that you're legitimate in some circumstance or conceivably in principle, and if you can't meet that challenge, which is the usual case, the structure should be dismantled."

This is not new thinking or untried, anarchist societies, like Catalonia or Manchuria were just as successful as contemporary states but anarchism is naturally not great at large scale government systems like defense so were soft targets for more aggressive ideologies. Therefore, if it's a valid way to organise a society - at least in-principle, then isn't it a plausible way to run an organisation?


I've been in the banking industry for a while now, which is notably more regulated than telco or energy. The standard response to this government oversight & regulation is internal policy, stage gating, governance forums, design authorities, information management certifications, privacy impacts, risk assessments, and so on. We then add all the usual drains on productivity like technology councils, steer co, cyber & security reviews, stand-ups etc. Lastly, we still all pretend that we're doing some form of Agile so have all those ceremony overheads too!

Anarchism does not mean no rules. In a previous rant I assert that we thrive when we have the right amount of freedom and individualism - not complete freedom by any stretch and anarchism aligns nicely with this.

There is only so much oxygen in an organisation to give us the energy for change. If we start with the hierarchy, middle management, councils, authorities, process and policy, this leaves engineering & testing gasping for air, with a massive load on their shoulders. The result of this is a glacial pace of change, people either find ways around the bureaucracy and bloat or burn out as they fight this monster of our own making.

But Semprini, you mystifying hunk of IT awesomeness, shouldn't an architect be providing guardrails for engineering? I hear you ask. Yes, indeed conveniently astute imaginary interlocutor. What I propose is to be skeptical about the validity of the edifices that we have erected, see them as a symptom of a problem rather than a solution, push back against the institutionalization of our thought and creativity. Some may indeed be justified, but lets take up the anarchy ethos to look a little deeper, a little more nuanced and ask if they are truly needed.

"There are no rules of architecture for a castle in the clouds." - G. K. Chesterton


Driving this tendency of slowly strangling innovation through hierarchy & process is what I think is a misunderstanding of risk. Just like the right amount of debt is a productive and profitable thing, the right amount of risk is essential for a high cadence of change and the opportunity costs are rarely taken into account. So often leaders don't trust their staff and try to reduce personal risk through external consultants and consensus processes.

Organisations are addicted to the adrenaline surge and the illusion of momentum from quick wins. These seemingly small victories, easily plucked from the low-hanging fruit tree, deliver an intoxicating rush. "I gotta have that hit on NPS" the addict may say and we put in a tactical solution to appease the addict. This short term outlook feeds the inability to accept risk, and consequently the imposition of draconian process while degrading the health of our IT ecosystem.

One of my least favorite process monstrosities is quarterly planning. If you are or have been involved in this, take a moment to think about the size and length of the meetings, how many senior managers and other staff are mandatory, put a hourly rate on everyone and marvel at the cost compared to the useful information each person actually received. Commonly, quarterly planning meetings cost over 100k and staff learn nothing over the 2-3 hours that they couldn't get in 5 minutes - if the information was readily available. And that leads me to the cure for this particular pointless exercise, personalisation. We shouldn't just strive for personalisation for our customers, but our staff too. With the right information at the right time, the edifice of quarterly planning (among others) can be dismantled - I.e. combine modern tech with anarchism to give the right info at the right time - and provide the right guard rails.


Process and policy may need to exist where there is choice. However, fewer IT institutions, better regulated can be achieved by being more opinionated about the way we implement IT. "Shift left" is great when things are standardised, like ingress/egress, access controls, infrastructure as code etc but we shouldn't use this ethos to force engineers to constantly reinvent the wheel.

We should track what percent of an engineers time is spent on the business problems and we should strive to always improve this ratio. We do this by making the best thing to do also the easiest -the pit of success.

This opinionated approach can remove or severely limit the need for things like technology councils.

Yet, we are all individuals. Everyone: "Yes, we are all individuals." Random peasant: "I'm not." - Life of Brian. We each thrive with different constraints. Hierarchy and process can allow flow state in people that get comfort from it and result in stress and anxiety if absent. Anarchism, I think, is about creating the right ecosystem for achievement and embracing these differences to allow the natural persona of our organisation to emerge and evolve.

Maybe a good way to think of individualism vs collectivism is like a zoologist as nature also never gives it to you straight - There is no such thing as a fish. I.e, trying to categorize things into boxes like "fish" is ultimately futile but as long as there is the right blur to the edges of the box, it can have huge utility.


As a data architect, I spend a lot of time pontificating on exact semantic meanings of things like "What is a Customer?" so this anarchism thing may seem at odds on the surface. However, I think a modern architect's role is to encourage change and individualism through sympathetic treatment of people process and technology. Anarchistic thinking helps us challenge the utility of processes to see a brighter north star.

I'm not an anarchist, socialist or capitalist. I view these things on a spectrum and the answer changes over time based on society. It's probably my Star Trek upbringing, but as a society becomes more egalitarian and more interested in self-actualization, the less control is needed. I think this holds true for enterprises too, and because most architects and engineers are already trying to reach their full potential, the only thing stopping us is legacy hierarchies and complex processes imposed by the short term thinkers.


Thanks to Stephen Clay, Alan O'Shanghnessy and Kent Matheson for their feedback and thought.

blog comments powered by Disqus