Link: http://weblog.tetradian.com/2012/09/30/at-unicom-ea-conf/
A big shift in enterprise-architecture indicated at a relatively-small but increasingly-important conference: that was my impression of Unicom EA 2012 in London late last week.
At first glance, there might seem nothing much that’s particularly new or different in the conference-programme: an unusual emphasis on panels and questions from ‘the floor’, perhaps, but that’s about it. Yet the big difference is in what was not in the programme: there’s none of the obsessive IT-is-the-centre-of-everything assertion that we’ve seen almost everywhere else in ‘the trade’ for most of the past decade. We’re at last reaching towards a true ‘architecture of the enterprise’.
And that shift in perspective was echoed from all sides, all the way through the conference. To give just two examples, Robbie Forder’s wide-ranging “Why do architectures fail?” and Philip Hellyer’s delightful “Thinkers and Doers: A framework for EA value” both focussed strongly on the human aspects of EA rather than the technical ones. Hellyer included a quote from Gerry Weinberg’s Secrets of Consulting, that “it’s always a people-problem”: that whatever it looks like, and no matter how technical it seems, in reality it’s always down to people to somehow (re)solve every problem – and our architectures need to be aware of that fact at all times.
There was only one presenter who briefly pushed the old IT-is-the-solution-to-everything line – and it looked a bit lame and myopic even to him by about halfway through his presentation. There was only one participant who insisted for a while that “ultimately EA is always and only about computers” – and in a coffee-break discussion he acknowledged that that assertion didn’t make sense even in his own strongly IT-oriented industry. That’s a very big shift from even a year ago – let alone the TOGAF conference some five years back where I was publicly dismissed as some kind of nutcase for asserting that ‘enterprise-architecture’ must be about more than just the enterprise-IT, that it literally means ‘the architecture of the enterprise’ as a whole.
In short, the conference indicates that we’re at last starting to arrive at a place where EA can actually do something useful at a true enterprise-wide scope. Hooray! 
We’re not there yet, though. The next trap after IT-centrism is business-centrism, where we break out of IT but the ‘inside-out‘ mindset still prevails: merely a shift from one ‘the centre’ to another, rather than reaching the necessary understanding that in a viable architecture, everywhere and nowhere is the centre, all at the same time. For example, one participant expressed his exasperation that, right near the end of the conference, I was the first person there to mention the words ‘business-case’. The huge danger of that trap is that it seems even more credible and defensible than IT-centrism – after all, we’re supposed to be working for our organisation, and it’s the business that’s paying the bills. Yet the reason why it’s a trap is best expressed by Chris Potts’ dictum that “customers do not appear in our processes, we appear in their experiences”: our architectures must start from that broader enterprise-context, not solely from the perspective of the business alone – otherwise we risk creating a business-oriented architecture that might seem perfect in itself but has no real-world context, and therefore has no real-world use or value.
The other challenge we face is that most of the tools we need are still stranded back in the old IT-centric space. At the conference, Harman van der Berg gave a great demonstration of the BizzDesign toolset, and its power and versatility in describing structure and change in architectures. The catch is that its underlying metamodel is Archimate, which is still rigidly centred around the old IT-centric pseudo-layers of (IT-only) ‘Infrastructure’, ‘Application’ and ‘Business’ – where ‘business’ is ‘anything not-IT that might affect IT’. So whilst there’s no doubt that BizzDesign is a great toolset, and arguably one of the best available at present, it won’t allow us to describe anything more than a small subset of the actual architecture of the enterprise – which can make it more of a hindrance than a help in real-world practice. The need for toolsets and metamodels that do support how we now need to work in EA is an ongoing challenge that toolset-vendors urgently need to address.
Overall, though, signs of real progress along the path towards a real enterprise-architecture. Feels good.
