Some ideas, more so than others, shine with a deeper truth. One such idea was articulated in 1967 by computer programmer Melvin Conway:
“organizations which design systems … are constrained to produce designs which are copies of the communication structures of these organizations.”
It is often found in good humour. The dysfunctional product: the system is a mess and the people who made it are a mess. Haha – that’s Conway’s Law for you! Apart from applied silliness, within this informal law lies the potential for a radical leap forward in the development of complex programmatic systems.
In 2011, Harvard business school published a paper: Exploring the Duality between Product and Organizational Architectures: A Test of the “Mirroring” Hypothesis. The collaborative work, conducted by MaCormack, Rusnak, and Baldwin, applied scientific rigour to Conway’s Law.
Under the name “The Mirroring Hypothesis”, the group analyzed patterns within software systems. The study separated software into two groups, those which are tightly coupled and those which are loosely coupled.
A tightly coupled organization is one with structured communication. This may be a corporate environment where employees dwell under the same roof, report to a manager - who reports to a manager - and who operate within an intentional structure in organized concert.
On the other hand, an open source project is a prototypical example of a loosely coupled organization. Communication includes random contributors, is unstructured, has no management, and no shared location. The contributors are motivated by completing a task and there is minimal - if any - social fabric.
Table from “A Test of the “Mirroring” Hypothesis”
|Tightly Coupled||Loosely Coupled|
|Goals||Shared, Explicit||Diverse, Implicit|
|Membership||Closed, Contracted||Open, Voluntary|
|Authority||Formal, Hierarchy||Informal, Meritocracy|
|Location||Centralized, Colocated||Decentralized, Distributed|
|Behaviour||Planned, Coordinated||Emergent, Independent|
A brilliant concept from the study is that of Propagation Cost. It is described as: “the percentage of system elements that can be affected, on average, when a change is made to a randomly chosen element.” In other words, the propagation cost is the overall percentage of a system that is impacted when a function or file is altered.
By comparing software systems which began as open source projects to those that were private, then opened, the paper establishes a clear link. Tightly coupled organizations create software with significantly higher propagation cost. In contrast, loosely coupled organizations build systems with a much lower propagation cost. These systems are more modular and therefore more approachable, or loosely coupled.
And so the take away is: systems reflect the communication structures within which they are developed.
A second study, The Influence of Organizational Structure on Software Quality: An Empirical Case Study by Nagappan, Murphy, and Basili of Microsoft pushed the idea further. The study revealed that:
”…organizational metrics when applied to data from Windows Vista were statistically significant predictors of failure-proneness.”
Fascinating! Organizational metrics predict failure-proneness. And the reverse is also true: code level failure metrics indicate poor organizational metrics. And what were these organizational metrics that they collected?
The following table has been adapted from the article, with the descriptions expanded for clarity and context. Note that high quality is low failure proneness, low quality is high failure proneness.
|Number of engineers||The greater the number of engineers that touched the code, the lower the quality.|
|Number of ex-engineers||Teams that lost team members had a decline in knowledge retention and therefore overall quality.|
|Edit frequency||The more edits to a component, the higher the instability and lower the quality.|
|Depth of Master Ownership||Components that were “owned” by a “master” – a single person with 75% or more of edits – had higher quality.|
|Percentage of Org contributing to development||The lower the overall percentage of the organization that contributed to development, the higher quality. Speaks towards small, focused groups.|
|Level of organizational code ownership||The more code contributors outside of the core working group, the lower the quality.|
|Overall Organization Ownership||The ratio of “masters” making edits to code compared to the total number of engineers. The better the ratio in favour of “master” contribution, the higher the quality.|
|Organization Intersection Factor||For each group that contributed 10% or more to a codebase, there was a decline in quality.|
In my interpretation of the study, an ideal development team is one that: is small, has had no churn, makes few edits, has an owner of 75% contribution history or more, does not need to work across an organization, and contains only one working group.
The article’s conclusion was thus:
“Our organizational measures predict failure-proneness in Windows Vista with significant precision, recall and sensitivity.”
And so our second key take away is: organizations can be analyzed to predict quality of programmed work and vice versa.
In summary, research is always helpful. But it is really dry and it is easy to miss ‘the forest from the trees’ when drawing conclusions from scientific rigour. Social dynamics are so rooted in chaos that one wonders about the repeatability and transferability of these studies.
But, we shall emerge from the weeds with two solid premises in tow and consider them to be true:
1) Systems reflect the communication structures within which they are developed.
2) Organizations can be analyzed to predict the quality of programmed work and vice versa.
Microservices by Conway
Phew. We needed some hard analysis to define parameters… But communication is ultimately a “soft problem”. It is emotionally rooted. Systemic attempts to control social dynamics within software engineering organizations is not met with enthusiasm. This is evident by the dire state of the ‘managerial’ concept.
We have product managers, tech leads, and scrum masters, but we do not have - nor seem to want - people managers, or any other form of social or spiritual advisement. Logically so, as it is evident that open source projects successfully operate without a long-term plan, organizational structure, or even social interaction. So what is the alternative?
Clever technical folk have tried to socially engineer communication by altering the technical system instead of the social system. To optimize for Conway’s Law and establish organizational structure as projects grow, proponents of microservice architecture presented a solution:
“When looking to split a large application into parts, often management focuses on the technology layer, leading to UI teams, server-side logic teams, and database teams. When teams are separated along these lines, even simple changes can lead to a cross-team project taking time and budgetary approval. A smart team will optimise around this and plump for the lesser of two evils - just force the logic into whichever application they have access to. Logic everywhere in other words. This is an example of Conway’s Law in action.” - Martin Fowler, Microservices.
The approach seemed wise. A microservice team will be small, likely have a “master”, and be made up of one working group. But when we think back to Microsoft’s ideal team, that group might still be impacted by churn, make many updates given cross-functionality, and work across an organization. Microservice proponents will point to ‘smart endpoints, dumb pipes’ and ‘you build it, you run it’ as social/technical maxims to address some of these higher level concerns.
Interesting approach. But it seems to create more dimensions within which we may apply Conway’s Law. Each microservice team is at risk of being constrained by its own communication structures, and so too the overall architecture constrained by the communication between each microservice team.
If we took an ‘empirical’ look at a team built around a microservice, what would we unearth as our indicators of quality? There will be variance, but what will govern it?
Compared to busy monolithic designs, loosely-coupled microservice teams proliferate systems that are more approachable and less failure prone. But it cuts the other way in that one now has a whole made up of independent pieces, each individually vulnerable to the drawbacks of poor communication. Are they better off because of this? I would argue no, they are not. No matter how we try to engineer the problem away, the fact remains: excellent communication is required for excellent systems.
The open source method has been successful in producing quality systems of a loosely couple nature. Is it then true that the right idea is to operate loosely, like an open source project? Or is there something that organizations can do to be better, to engineer a more effective social system that itself creates better systems?
Communication structures live in deep complexity and are fractal in nature. We can follow them to great depth, inward and outward, to a limitless degree: systems, influenced by systems, influencing systems. The microcosm hints at the macrocosm. As above, so below.
This not an engineering problem. It is a natural one. It is a social one. And it is not a problem at all. We walk the brightest path by improving our communication with others. The organization that learns to minimize ego and enhance verbal, non-verbal, emotional, technical, and spiritual communication will reach the highest of heights, building the brilliant things that today are constrained by contemporary discourse.