Other Methodologies¶
Beyond Scrum, XP, Kanban and DSDM, several methods are worth knowing. Some are still used directly, and others contributed one idea that everything since has absorbed.
Crystal¶
A family of methods (Clear, Yellow, Orange, Red) sized by team size and criticality, on the argument that a 6-person website and a 200-person medical device cannot share a process. Its contribution is that methodology weight should scale with the cost of failure.
Core practices: frequent delivery, reflective improvement, osmotic communication, which is the learning that happens simply by sitting within earshot of the team.
Feature-Driven Development (FDD)¶
Five processes around small, client-valued features of two weeks or less, with an explicit up-front domain model and a class owner per area. It scales to larger teams more comfortably than XP and appeals to organizations that want documented design.
Covered in full on its own page: Feature-Driven Development.
Rapid Application Development (RAD)¶
Prototype-driven, timeboxed, and heavily dependent on user involvement and tooling. Historically important as DSDM's direct ancestor and as the origin of "build a prototype and evolve it". Vulnerable to prototypes being shipped as products.
Scrumban¶
Scrum's cadence and roles, plus Kanban's WIP limits and flow metrics. Typically arrived at rather than chosen, when a Scrum team keeps having its Sprint broken by urgent work and decides to make the interruptions visible instead of pretending they do not happen.
Spiral model¶
Risk-driven, iterative, with each loop covering objectives, risk analysis, development and planning. Not agile, but it is where "address the highest risk first, in iterations" enters the mainstream. See Other Process Models.
Shape Up¶
Basecamp's method: six-week cycles, then a two-week cool-down. Work is shaped at a middle level of abstraction before it is handed to a team, and the team gets full autonomy over how to build it. Distinctively, the appetite is fixed and the scope is cut to fit, which the method calls hammering scope, and projects that will not fit are stopped rather than extended.
Roughly when each appeared¶
graph LR
RAD[RAD, 1991] --> DSDM[DSDM, 1994]
SPIRAL[Spiral, 1986] --> UP[Unified Process, 1999]
SCRUM[Scrum, 1995] --> AM[Agile Manifesto, 2001]
XP[XP, 1999] --> AM
DSDM --> AM
CRYSTAL[Crystal, 1990s] --> AM
FDD[FDD, 1997] --> AM
AM --> KB[Kanban for software, 2007]
AM --> SU[Shape Up, 2019]
SCRUM --> SB[Scrumban, 2008]
KB --> SB
What each one contributed¶
| Method | Idea that outlived it |
|---|---|
| Crystal | Process weight should scale with team size and criticality |
| FDD | Client-valued features as the planning unit, with an explicit domain model |
| RAD | Evolutionary prototyping with heavy user involvement |
| Scrumban | Cadence and flow control are independent choices |
| Spiral | Attack the highest risk first, iteration by iteration |
| Shape Up | Fixed appetite, variable scope, and the courage to stop a project |
Check Your Understanding¶
What is Crystal's central argument?
In Shape Up, what happens when a project will not fit inside its six-week cycle?