Tuesday, July 13, 2010

It's easy to spot a purist...

...they're the ones with no skin in the game. I came across this whilst reading about Hugh MacLeod's new book on his blog. It set me thinking about how true this can sometimes be from an architecture perspective. Steve Jobs famously said “real artists ship”. Whilst there is a point to building good architectural models, defining principles that should be followed and road maps that describe the directions architects will be travelling in there comes a point that we must say “enough is enough, it's time to ship”. The fact of the matter is that software development is, and always will be, a complicated activity and there is often no clearly defined answer to the wicked problems we need our systems to address. In fact, two of the attributes of a wicked problem namely, wicked problems have no stopping rule (instead the problem solving process ends when you run out of resources rather than when some 'correct' solution emerges) and solutions to wicked problems are not right or wrong (they are simply better, worse, good enough or not good enough) would seem to encourage the kind of purist behaviour where nothing ever ships. The danger with architecture (especially Enterprise Architecture) is that the artefacts being created are intangible things like models, road maps and principles and these by themselves do not make working systems. The people that produce such artefacts could often be accused of having “no skin in the game”. So how should you ensure that the architects in your organisation do indeed have skin in the game and knows that shipping is actually all that matters? Here is my SKIN (Specify Kill Integrate Negate) rule of thumb for ensuring delivery takes protracted (and unnecessary) analysis.
  • Specify to the right level of abstraction and detail, no more no less. The models, road maps, architectural principles that are defined by architects should be understood and usable by as many stakeholders as possible. This may mean having different views (levels of abstraction) of the same model but ensure if this is the case, those views are consistent. Think who your stakeholder is meant to be before building the artefacts and don't build it if you can't explain it clearly to that stakeholder.
  • Kill unwanted artefacts. Sometimes with the best will in the world you will build an artefact that does not have a purpose, is not read by anyone or has expired. Kill it, don't let it linger on causing confusion and giving you an unnecessary maintenance overhead. Look at something like TOGAF for a good set of useful artefacts.
  • Integrate the work of the architect(s) with real downstream activities. Don't let them work in ivory towers which have no grounding in reality. When planning a project and considering what artefacts need to be planned it is useful to think of what the inputs are to the task that is to create the artefact. If you cannot see any of the artefacts that are output from architecture tasks being used than maybe it's time to consider killing off those artefacts (see above).
  • Negate unnecessary activities and tasks. Every task should have an output. This is usually in the form of a work product (a model, a piece of code, a plan or whatever). If you find you have some architecture tasks or activities (groups of relates tasks) that are not producing useful artefacts then its probably time stop performing those tasks.

No comments:

Post a Comment