The Great Escape: “EA is not about IT!”

As soon as you, as an enterprise architect, want to address the problem of the complex Business-IT landscape, and you actually acknowledge in a discussion that Enterprise Architecture has come into existence because of the complexity of modern IT-laden landscapes, there is always some enterprise architect who will (angrily, often) state that Enterprise Architecture “isn’t about IT”. Some now propose a separate name EITA for ‘Enterprise IT Architecture’ for the IT-kind so that ‘Enterprise Architecture’ can remain being focused on ‘designing enterprises’ instead of ‘designing enterprise IT’.

For instance, while most people enjoy the Why Enterprise Architecture animation, you’ll find a complaint in the comments section that it wrongfully suggests that Enterprise Architecture is about IT, e.g.:

As much as the video is clear, it is also clearly wrong and passively misleading (by omission).  The video talks only about technology, and Enterprise Architecture is NOT primarily about technology.  Sure technology is a PART of EA, but it is a small part.  This video talks about robustness, efficiency, complexity and solution architecture, which are concerns for IT architecture, but NOT enterprise architecture.

And

Mostly agree with Nick – it’s a great video, but about a proper subset of EA? If it just started with the business perspective and elaborated on the “What they do for the business…” part of the project architecture, it could really nail it…

The history of Zachman’s frameworks is interesting – according to http://www.zachman.com/ea-articles-reference/54-the-zachman-framework-evolution, although he began with an IT focus under the rubric “Information Systems Architecture – A Framework”, his view was that “strategy and information systems needed to be “engineered” for the ENTIRE Enterprise, not just “manufactured” by the I/S department” but he worried people weren’t ready for this idea. Perhaps they still aren’t…

In a recent (and still ongoing) LinkedIn discussion Are EA’s really architecting Enterprise or simply improving existing enterprise? Charles Rosenbury wrote the following — and in  my view very insightful — comment:

The truth is, that without IT there would be no need for a specialized role called enterprise architect. Not because enterprise architects need to know IT, but because without IT you would not have enterprise large enough and marketplaces agile enough to require an enterprise architect. Prior to IT, we did not have those kinds of enterprises. Prior to the industrial revolution we did not have efficiency experts or process engineers. The size allowed by information technology as well as the pace of change afforded by information flows enabled by IT are the reason we need the role.

I think Charles hits the nail exactly on the head here. The current digital revolution — which is far from done, by the way — creates a whole new level of complexity in our organisations.

The industrial revolution cost many craft workers their jobs as machines could do their job cheaper, quicker, and sometimes better than they could do it by hand. We can expect the same from the current digital revolution. Typists have been replaced with word processing programs. We can expect many lower- and medium-level administrative jobs, especially when they are mostly executing logical rules, to be replaced by computer programs executing logical algorithms. ‘Exception-based architecture’ is a popular approach these days, and this is not that different from the changes that made craft workers during the industrial revolution lose their jobs. So, ‘administrative clerk’ may become a job of which we have far less in the future. But ‘robot maintenance assistant’ might be quite a more common job in the future.

In other words: as it is IT that drives the growth in complexity that requires enterprise architecture in the first place, EA is definitely about IT. The whole Business-IT split is artificial: IT is Business these days and definitely in the foreseeable future. We will even see a growth of autonomously operating IT, in the form of robots (either physical or just informational, such as self-service web sites), and these robots will be more and more inseparable from ‘business’. Stating that ‘EA is not about IT’ and that managing just the business layer (whatever that is, especially in the future with those ‘robots’) is running away from the real issue into a make-believe land where IT is inconsequential to the business decisions that need to be made and where IT just follows business choices (whatever those are, by the way, because what difference is a non-IT choice versus an IT choice in today’s organisation?). That’s not managing the grown complexity, it is an escape into a last vestige of the classic industrial society.

EA is not about IT? If you say that, you were definitely right in the nineteenth century, but not today.

Charles follows this up later in the discussion with:

At some time in the future, several possibilities emerge, including artificially intelligent systems which configure and report on themselves (eliminating the need for enterprise architects) or the leadership of the company will reach a knowledge level where they understand the challenges/opportunities well enough that they won’t need enterprise architects.

Here, I disagree. IT has fundamental strengths, but it also has a fundamental weaknesses: IT does not scale very well and IT is brittle. As soon as the situation is even just a bit out of what has been reckoned with, IT breaks down. Our Business-IT landscapes are growing in complexity, but they run the risk of growing also in brittleness, something we encounter daily when IT projects run into trouble because the Business-IT landscapes have become so complex that planning a reliable delivery becomes extremely difficult. Add to that the unpredictability and volatility of requirements (which are neither strictly logical, nor stable) and we are creating a world where brittle and difficult to scale IT is trying to keep up with unpredictable and difficult to pin down requirements.

The complexity comes with the massive use of digital technology, which in the end is by definition about discrete artefacts (can be expressed in 0′ and 1’s) and discrete rules (can be expressed in program logic). The fundamental strengths and weaknesses of IT are therefore in the end the fundamental strengths and weaknesses of logic. We are now industrialising classical logic to an unprecedented level. In the 50’s and 60’s of the past century, this led to the (still often professed) belief that this industrialised logic would become so intelligent that it would take over much of the human side of the system. All the efforts spent on AI have delivered interesting techniques, which are for instance used in data analytics (data mining) and rule-based workflow engines, but nothing even remotely close to actual intelligence. Which means that in my view, enterprise architecture is a discipline that is here to stay and we will need many architects in the future.

Our current EA ‘best practices’, though, are from the industrial area and possibly (probably) not that well adapted to the level of complexity and unpredictability that they are employed for. We’ve been trying to handle the digital revolution with mechanisms and ideas that may have worked in the industrial revolution, but they don’t really scale.

For me it has been clear for a while that we are approaching not so much a `singularity point‘ (futurologists like Kurzweil, SF-authors like Vinge) where technology will take over from human intelligence (and we’re home free with regard to the complexity problem), but we’re more on the road to a complexity limit (Paul Allen of Microsoft, 2011) where we end up with complexity, brittleness and so forth limiting progress in the digital domain. As far as predictions go: I think the situation in enterprise architecture well become more difficult even, e.g. with the added complexities provided by software defined anything and the fragmentation the Cloud will offer. I am convinced we need a better approach than our current ‘best practices’ and frameworks to enterprise architecture.

Underlying all of this is a very deep and fundamental issue: how intelligent can the digital domain become? Kurzweil and many others effectively assume that intelligence in the end is like complexity that arises out of a massive amount of (discrete) logic, so we must somehow get there soon, based on Moore’s Law of increasing digital computing power. It is an assumption that is strong in our culture, but if you really go into this subject deep, you find out that there are extremely good reasons to doubt this, stretching from Wittgenstein to Dreyfus. The world is not Rational, it is Real, and that difference has consequences, especially for Enterprise Architecture which lives on the intersection of the two.

PS. I’ll be speaking at Gartner’s EA Summit 2015 London UK on May 2oth and at the MBT Congres in Houten NL on May 21 about Chess and the Art of Enterprise Architecture.