The fourward method
Four layers , four dimensions , sixteen facets . A lightweight framework for mapping your architecture and moving it forward , four at a time .
In the first article , I introduced a mental model for backend systems . Four lenses , four codebase layers . That model works for a single service . But zoom out and the question changes : how do you assess an entire technology landscape without drowning in spreadsheets ?
You use constraint . Specifically , the number four . Why four ?
- Working memory holds about four items comfortably . Enough to reason about , not enough to get lost in .
- A year has four quarters , which gives you a natural re-assessment rhythm .
- A 4x4 matrix fits on a single screen . You can see the whole landscape at a glance .
- The limit forces you to group things meaningfully . No catch-all “other” bucket when you only have four slots .
A product manager once told me “I can see the matrix” , referring to how I navigate architecture decisions . So I made an effort to describe what that matrix actually looks like .
Fourward is a play on forward . Four layers , four dimensions , sixteen facets . Each facet rated now and where it needs to be . The gap is the work . And because it’s structured data , agents can work with it too .
AI doesn’t replace good engineering practices . It amplifies them . Give it a clear model and it moves faster . Give it nothing and it drifts … and you with it .
The full method is on the Fourward Method page .