If you are a long-serving computing practitioner who has been through mainframes in data centres to mini computers in departmental computing and then to PC networks and IT, you might just recall hearing about Conway’s Law. Well its coming back again as we move into clouds! Melvin Conway’s thesis, the piece of work that gave birth to the concept of Conway’s Law, first surfaced in 1968 as part of the shift into departmental computers. Essentially Conway’s point was that in designing enterprise business models, computer solutions, even products to take an organisation to market, will always mimic the enterprise’s own communication structure.
Conway’s Law = …organizations which design systems … are constrained to produce designs which are copies of the communication structures of these organizations.
Some good examples of what this might look like, based on the original thesis, can be found on Wikipedia but you can get a more up to date view from 2008 work at Harvard Business School and at Microsoft Research. To understand the interest and why it comes up at times when technology innovation leads to business change, let me provide my own experience relating first to what it meant at the time of PCs and networks, then what it means now in the context of clouds.

Each of our technology eras has resulted in a new business model, organisational structure, set of working methods, and perhaps most importantly of all, a new competitive value proposition. Okay that’s not a new point, but at each shift there has been a key dependency on a core piece of technology which at the time seemed impossible to justify within the existing communication and organisational model. Can you imagine working without email? Well in the early 90s many, even most, enterprises couldn’t figure out the business case for email.
At this time the organisational model was both hierarchal and rigidly separated by departments, each of which had a departmental computer and set of applications that enabled them to automate and keep track of their own processes and resulting data. Though some office automation products existed such as IBM Personal Services and Digital Equipment Corporation All-in-One, paper and the interoffice memo ruled. Networked PCs and client server technology capabilities led to new business models based around business process re-engineering (BPR) concentrating on optimising the horizontal flow across the departments. On the people organisational side, this introduced matrix working, a person’s ability to perform their unique role in multiple different processes, and that’s when the fun started!
Who was responsible to whom, and for what? If the people still worked in departmental organisational structures and the critical issue lay in a flow process in which their department performed a minor role, how did the issue get communicated? Up the hierarchy within the department until a departmental head spoke to another departmental head? Sounds stupid now, but that’s how it was at first. The whole point about email was it changed ‘who could communicate with whom and about what’ into a new communications structure that enabled the flexibility of matrix working within business processes rather than departments.
So how do we shift towards a ‘services’ model based on cloud technology with its inherent agility towards frequent change, and a focus on optimisation of events by deploying people’s expertise, if we are still working with the communications capabilities and organisational structure of matrix working? Who pays for this collaborative stuff? It’s the email issue of 1990 all over again! Back in the ‘90s email arrived in the enterprise as increasing numbers of groups of workers started using different email products to be able to do the work that was now expected of them. So the fundamental driving force of optimisation of the enterprise ended up being held back through the fragmentation of the communications structure.
It’s the same today, everywhere across the enterprise the people with the most ‘active’ roles that can really make a difference to optimising events and opportunities are organising collaboration tools for their own group. At the top, denial of the need, or the fact that this is happening at all, is all too often the order of the day. Hence, why Conway’s law is back again. It’s there to help us all to understand the link to communications and organisational structures, when discussing how to change our business models.
It even helps to explain some things in the last year or two! Try this December 2009 blog post on SAP Sapience.