In defence of bureaucracy by Mark Seemann
Objectivity, meritocracy, high bus factor. What's not to like?
Most people dislike bureaucracy, and to be honest, so do I. Even so, a subjective distaste is no argument. It may be wise to understand if bureaucracy entails any benefits, so as to not throw out the baby with the bathwater.
_-_14.jpg "Jarek Tuszyński / CC-BY-SA-3.0 & GDFL")
This essay does not attempt to make the point that more bureaucracy is better. Rather, it argues that some bureaucracy, wisely chosen, is better than no bureaucracy.
Origins of bureaucracy #
Consider the origins of bureaucracy. Believe it or not, it's an institution designed to get things done; to inject a degree of predictability and determinism into a situation that otherwise produces arbitrary results. While I'm no historian, I understand that the Sumerians, ancient Egyptians, Romans, Chinese, and others, all employed civil servants and a system of bureaucracy to control their empires.
In modern times, bureaucracy has served a similar purpose, replacing feudalism with rules-bound administration. Of course, as Seeing Like a State argues, part of the motivation was to centralize power, cutting out the middle-men (vassals). Even so, a by-product, which later became a goal of its own, was that bureaucracy is a rules-based order. It dampens the arbitrariness of a feudal lord by instituting universal rules for all to follow. During the nineteenth and early twentieth centuries, it became the primary means of making society more just.
Subjects could appeal to ideally objective rules to settle disputes, or to obtain permission to engage in various activities. In theory, a wealthier citizen could not by bribe influence decisions of those in power.
As rights were gradually expanded to include women, unlanded men, people of colour, etc., bureaucracy was a major instrument of implementation.
In parallel to this development, most organizations realized that hiring administrators based on merit, rather than on inheritance, ensured better execution.
Too much of a good thing #
Of course, when everyone complains about bureaucracy, it's because there's too much of it. It's like the Laffer curve related to taxation: Even if you don't like paying taxes, unless you're an anarcho-capitalist, you probably agree that some state institutions (e.g. courts, police, defence) are desirable. And unless you're a true communist, you probably believe in some degree of private property rights. Thus, neither zero percent or a hundred percent tax rates are desirable. The remaining problem is to decide on the optimal fraction in the open interval between 0 and 1.
The same applies, I believe, to bureaucracy. No bureaucracy entails the total absence of objective rules, meritocracy, traceability. Total bureaucracy, on the other hand, implies that nothing gets done. Again, the problem is to find the right balance. And this balance may not be an equilibrium. As external circumstances change, you may have to change how your bureaucracy works: How much of it, what the rules are, etc.
The worst system #
As has been incorrectly attributed to Winston Churchill, democracy is the worst form of government, except for all others which have been tried. Analogously, capitalism is the worst means of wealth distribution, except for all the other systems that have been tried.
It's natural to extend those notions: Bureaucracy is the worst form of administration, except for all the other systems that have been tried.
Of course, a counter-argument could be made that many societies work well without bureaucracy. Anthropologically interested readers will surely point to various well-functioning tribal societies, and tech bros will point to start-up companies. Such organizations surely exist, but they don't scale. Infamously, all successful start-ups ultimately add bureaucracy as organizations grow.
The mature choice #
Programmers tend to dislike bureaucracy as much as the next person. It can be difficult to see the value in moving Jira tickets around.
In light of the above, I'm not advocating bureaucracy for bureaucracy's sake. Much of it is, indeed, counter-productive, but some of it could actually boost your productivity.
For example, most programmers agree that interruptions are major productivity killers. Many interruptions are requests for status updates. How is the feature coming along? Are you working on this, instead of that? How soon can you be done?
I once lead a team of developers, in an organization that used Trello boards, and I tried to convince them that if they would reliably move tickets around on the board to reflect actual status, it could save them quite a few status requests.
Even such a modest request was, however, sullenly but actively ignored as beneath real programmers. The result: Frequent interruptions from stakeholders who wanted to know how far the project, or a particular feature, was coming.
I find such behaviour immature. By being proactive, you can surface information at the time you choose, and if done right, it can decrease the frequency of interruptions you experience. It does, however, require that you minimally play along with a bit of bureaucracy: Move those tickets from left to right. Write status updates. Radiate a bit of information.
Bureaucracy, the good parts #
Clearly, useless red tape exists. The goal of every software organization is to identify what actually works. This varies from organization to organization, and over time, so I don't claim to have the correct and complete list. If you're in doubt, however, I would recommend trying the following:
Write things down. And keep records in a place where everyone can find them. A Slack channel is probably not a good candidate. As I write in Code That Fits in Your Head, a sensible hierarchy of communication exists. It includes favouring readable code over comments, but comments over documentation. Specifically, you should prioritize writing better Git commit messages. And for broader decisions, keep Architecture Decision Records around.
The point of bureaucracy is not to move Jira tickets around for no reason. The crux is to put information where people look for it. If this is Jira, consider it a small price to pay for fewer interruptions.
Most organizations I've consulted tend to have an oral culture. Talking to each other is important: It's a fast, high-bandwidth mode of communication, and it often works well as a social lubricant. It may enable you to go fast in the short term. If, in the other hand, you keep no written record, you can't sustain the pace in the long run.
People cycle out of long-running projects. Key contributors move on to other jobs in other companies. Without a written record, you lack important information about the system. Consider good bureaucratic artefacts as a major counter-move against this problem. When done well, it increases a team's bus factor.
Conclusion #
Bureaucracy understandably has a bad reputation. It's usually done wrong, in which case it can induce unnecessary friction into processes.
Even so, consider what motivates it. Originally, it was a remedy against arbitrary rule of emperors, kings, feudal lords, or other people in power. It's a means to institute regularity and fairness into administration. It's meant to not only regulate, but also document, how administration is performed. When abuse of power occurs (which still happens), investigators can often use bureaucratic artefacts to uncover what most likely happened.
In software development, you may not be concerned with abuse of power, but keeping good records has many benefits. The challenge is not to be rid of bureaucracy, but to trim it to a size where only the good parts remain.
