Iterative Architecture During Feature Shaping with Feedback Loops
I often discuss how and when to address software architecture while building a new feature. I answer this here with the terminology of Shape Up, but the answer is not specific to Shape Up but applies generally (however you call this stage your process).
TLDR: Architecture needs to be addressed during the shaping of a feature. Only then can the implementing team work autonomously during the delivery cycle. It would be unfair to throw things at the team, without even remotely clarifying how to implement them.
In this group, an idea is collaboratively “shaped,” meaning Mitigating risks and defusing ticking time bombs (which could blow up later during delivery). However, there’s a constant battle against diving into details too early.
Furthermore, in Shape Up, there are no estimates (fixed scope that ends up with variable time), only an “appetite (fixed time, variable scope).” It’s about how many resources one wants to invest. For example, two programmers and one designer should spend six weeks on this.
