Modularity is a key aspect of software and architectural design, setting explicit boundaries between different parts of the system. But we have been banging on about this since the 70s, and still we are creating big balls of mud -- now even the distributed kind. Either modularity as a concept is insufficient or maybe there are aspects here we seem to get wrong. We know that good fences make good neighbours, but only when the boundaries are placed correctly. How can we create robust and sustainable modular designs when identifying those boundaries are so challenging?

In this talk, we are going to take a closer look at why modularity is needed, what it actually can do for us, and how we can increase our chances of getting it right by taking a systems thinking approach. The claim made is that a holistic view of the problem space is critical; one that consider all its parts, including the business and all the people affected. Software development today is a inherently a sociotechnical endeavour and any modularisation effort, be it information hiding, SOA, microservices, DDD, Team Topologies and more, must take this into account in order to be able to create solutions that are sustainable and have the necessary conceptual integrity. Maybe this will guard you against the dreaded distributed big ball of mud, the killer of agility and productive collaboration.

New talk that builds on the previous talk and blog post of mine, "Microservices Without DDD is Risky Business!", which are available at https://www.linkedin.com/in/trondhjort, extended with basic systems theory,