Both digital transformation and devops are organizational and cultural transformations more so than they are technology changes – although in both cases, technology plays a large part in driving the organizational change necessary to achieve the business value for either effort. Just how the organizational changes and technology changes work together, however, is a difficult question.
Fortunately for us, this question is an old one – 47 years old, in fact. In a 1968 paper, computer scientist Melvin Conway wrote, “Any organization that designs a system will inevitably produce a design whose structure is a copy of the organization’s communication structure.”
Conway produced no evidence of this statement, and his paper was initially rejected as a result – but to this day, we refer to this statement as Conway’s Law.
Law, however, is decidedly an overstatement. Observation is actually more accurate. In fact, the Wikipedia page for Conway’s Law states that “Although sometimes construed as humorous, Conway’s law was intended as a valid sociological observation.”
Whether the statement be humorous observation or law, however, today Conway’s Law plays a central role in our efforts to break down organizational silos in order to improve business velocity and better meet the needs of customers – the purported goals of devops and digital transformation, respectively.
For Conway’s Law to be a useful tool, however, we need a better causal story. Can changing our organizational structures impact our technology? Or more importantly, how does technology impact our organizational structures? And why?
Melvin Conway was an early computer scientist whose most interesting accomplishment other than the law itself may have been his part in creating the compiler for the original 1984 version of Mac Pascal.
His eponymous law, however, apparently languished until sometime before 1996 or perhaps even earlier than 1991, when open source guru Eric Raymond included Conway’s Law in his Jargon File. He restated it with the snarky example, “If you have four groups working on a compiler, you’ll get a 4-pass compiler.”
Our story then gets more interesting when the Harvard Business School attempted to put some meat on the bones in a 2008 study. They use an objective measure of the modularity of software to find evidence for Conway’s Law. They conclude:
It’s important to note two important updates to the context of Conway’s Law by the time of the 2008 study: an organization’s “communication structure” becomes the structure of the organization itself, while “system design” becomes “product architecture” – with the constraint to product architecture being the scope of the study, rather than any sort of judgment on the scope of Conway’s Law itself.
What the 2008 study is essentially lacking, furthermore, is a discussion of the underlying causal principles behind the law. Instead, it is more or less taken for granted that siloed technology teams will create modular systems with the modularity aligning to the team structure simply because that’s the way people behave when they’re in teams.
The reason this question of causality is so important is because it goes to the heart of how we might use Conway’s Law to actually improve things. In particular, will changing organizational structures enable us to build better software?
Jonny Leroy and Matt Simons of ThoughtWorks explored this question when they coined the term “Inverse Conway Maneuver” in an article in the 2010 issue of the Cutter IT Journal. They state:
Conway’s Law … can be summarized as “Dysfunctional organizations tend to create dysfunctional applications.” To paraphrase Einstein, you can’t fix a problem from within the same mindset that created it, so it is often worth investigating whether restructuring your organization or team would prevent the new application from displaying all the same structural dysfunctions as the original. In what could be termed an “inverse Conway maneuver,” you may want to begin by breaking down silos that constrain the team’s ability to collaborate effectively.
The ThoughtWorks article takes the context of Conway’s Law from observational to normative: not satisfied simply to reflect on the way things work, they take the important step of opining on how things should – and should not work. They also intentionally focus on the negative: rather than simply describing how to build good software, they note that Conway’s Law is about building dysfunctional software.
It’s no surprise that ThoughtWorks’ focus is on changing organizational structures in order to build better software – after all, ThoughtWorks is a software development organization, and most modern software development thinking, including both the Agile and Lean movements, focuses on organizational change in furtherance of better software.
For digital transformation efforts, however, we must reverse this discussion. Technology change is driving changing customer preferences and behavior, which in turn are driving organizational change across increasingly software-driven enterprises.
The causality question behind Conway’s Law, therefore, is less about how changing software organizations can lead to better software, but rather how companies can best leverage changing technology in order to transform their organizations.
Hints at how to answer this question surprisingly come from the world of devops – surprising because the focus of devops is ostensibly on building and deploying better software more quickly. Be that as it may, there’s no question that technology change is a primary facilitator and driving force for the devops cultural and organizational shifts.
If we didn’t have the cloud computing example of fully automated deployment and operational environments, and if we didn’t have today’s dramatic innovations in continuous development, continuous integration, and continuous delivery tooling, then devops would never have left the whiteboard stage. There’s no question that the devops story is a tale of technology-driven organizational change.
The devops technology landscape, however, doesn’t have an end-to-end, seamless technology story. On the contrary, this landscape is cluttered with dozens of various tools, many open source, all of which are in various stages of maturity. Therefore, it’s not readily apparent how to apply Conway’s Law, since we’re trying to leverage diverse toolchains of tools and technologies in order to help evolve our organizational structures.
In fact, Conway’s Law describes how such a diverse tooling marketplace came to be in the first place, as open source teams are generally quite modular, and thus will produce modular software.
Once we’ve solved the organizational challenges of digital and devops, breaking down silos in order to deliver customer value at velocity, then we can expect Conway’s Law to kick in again, and predict the rise of end-to-end software solutions as the result of horizontally self-organized teams.
It’s the middle piece, however, we’re struggling with now: how do we leverage a diverse set of disparate technologies to facilitate the cross-cutting reorganization of our businesses, contrary to the observation of Conway’s Law? And do we really think going against Conway’s Law will actually work?
Not to fear. In fact, the technologies that underpin devops aren’t as diverse and disparate as the application of Conway’s Law might suggest. True, the creators of all the tools in our devops or digital tool belts are working mostly independently of each other, a pattern of behavior which in the past has led to incompatible software mishmashes.
Today, however, we’re seeing the rise of what we might call crowdsourced architecture, as all of these teams work within the context of mature communication protocols, RESTful interfaces, and other emerging architectural trends like containerization and microservices.
As a result, mostly independent doesn’t mean completely independent. Instead, we have the loosely connected communication structure that assures us that Conway’s Law is still alive and well. What’s changed is the context for this notion of an organizational communication structure.
Conway was referring to the communication structures of companies or software development organizations or perhaps individual software development teams. Today, however, we’re talking about the broader technology community itself.
There was no way Melvin Conway could have observed such crowdsourced architectural maturity back in 1968, because the telephone and snail mail-based communication structures of the day weren’t conducive to crowdsourcing anything. In contrast, today we have a plethora of tools and processes for facilitating communication and collaboration across traditional projects, teams, and open source efforts.
The end result is essentially a two-level application of Conway’s Law: a collaborative extended community of technologists that creates not simply a collection of disparate tools but rather chainable tools that leverage crowdsourced architectural principles to facilitate a level of coordination and interactivity we’ve never seen before.
This coordinated technology environment, in turn, facilitates the reorganization within companies, as they now have the tools they need to break down organizational silos, and people within those companies self-organize along horizontal lines, connecting customer experience to back-office software development and operations.
Conway’s Law, therefore, does work both ways. Organizational structures impact system design, and system architectures impact organizational structures as well. In the final analysis, however, Conway’s Law remains stubbornly observational. The underlying causal story – why such observations are remarkably universal – remains to be told. Stay tuned!
Intellyx advises companies on their digital transformation initiatives and helps vendors communicate their agility stories. As of the time of writing, none of the organizations mentioned in this article are Intellyx customers. Image credit: Melvin Conway.