[Sigia-l] How the design process fits into the agile methodology

[Adrian Howard]

On 18 Feb 2007, at 09:20, Ziya Oz wrote:

> Livia Labate:
>
>> the agile nature of the development really only speaks to the  
>> development
>> cycle
>
> I wonder if there's general agreement on that. (I myself don't have a
> position.) If the development cycle has, say, 30 steps, it's  
> relatively easy
> to conjecture how steps 3-30 may evolve through iteration. But who
> determines what the *first* 2-3 steps will be?

A common myth that seems to have emerged about agile methods is that  
there are is only one "level" of step in agile methods, or that  
people only look one step into the future. This is not the case.

For example if you look at XP you have typically have feedback loops  
operating at the level of:

* hours (test-driven design)
* days (daily stand up meetings)
* weeks (iterations)
* months (release plans)
* quarters (quarterly cycle)

Obviously the short-loops are more implementation focussed while the  
long-loops are more strategic. The key difference with agile methods  
is that they attempt to do the absolute minimum amount of up-front  
work to make progress, and rely on feedback much more than prediction.

It turns out that, at the lower levels of hours/days/weeks, the  
minimum amount of up-front work is a _lot_ smaller than many folk,  
myself included, originally thought. Getting there has involved  
inventing a few new methods, and pushing some existing ones in some  
new directions. A whole bunch of the IA/UX/IxD/whatever world's work  
lives quite happily at these levels.

Figuring out how to do the minimum possible amount at the higher  
levels is, as far as I am concerned anyway, still very much a work in  
progress. Obviously a whole bunch of the IA/UX/IxD/whatever world  
lives at the higher levels too.

Cheers,

Adrian