Ivory Tower Architecture

The Enterprise Architecture profession is littered with failed EA practices. EA is a tool that can be used equally well to sculpt Michaelangelo's David, or a mashed potato and gravy volcano. EA can fail for a variety of reasons, and I'll explore some of the pitfalls in a series of posts, of which this is the first.

What is Ivory Tower Architecture

There are dozens of architecture frameworks and approaches. TOGAF, Zachman, PEAF, FEAF etc. They provide a new architecture practice with some guard rails and some direction. What they don't provide though, is Context and understanding. It can be very tempting to take a tool like TOGAF and think that applying it as written means you're doing Enterprise Architecture correctly. Whatever that means.

How can I recognize it?

Strict adherence to any framework is a big red flag. Even The Open Group would tell you that TOGAF is meant to be tailored to the situation - and that means you need to start with understanding the context that you're operating in, and also need to understand the "Why" of the various pieces of whatever framework you've decided to apply.

That level of understanding takes a lot of experience both with the framework, and the environment you're in. External consultants will come in with great understanding of the framework, but poor understanding of the environment. Internal folks bring good understanding of the environment, but generally poor understanding of the framework.

Another trailing indicator is external attitudes toward EA. If EA is seen as a roadblock, or a checkbox to tick...you probably have Ivory Tower architecture happening. When done right, EA should be seen as a partner that's adding value, not just another hoop to jump through.

Why is it bad

Ivory Tower Architecture leads to the attitude that EA is something that gets in the way. If something gets in the way, stakeholders will start withholding information -- not necessarily intentionally, but out of a sense of self preservation. After all, if EA doesn't have the full picture, there's less to object to, and therefore you're more likely to get that box checked off.

Without that full collaboration with stakeholders, EA loses it's value in a hurry. The enterprise will start making decisions in silos, and things won't be working as well together as they should. Complexity will be up, value will be down. Basically having late stage Ivory Tower EA is indistinguishable from not having Enterprise Architecture at all.

How do we avoid it

Question everything about your chosen framework(s). Don't ever do something without understanding why you're doing it, and what you expect to get out of it. Stop making your stakeholders bend to your process, and instead be flexible to their processes. Find where you can provide value to them, and inject just enough EA at just the right time. Only produce the types of artifacts that are necessary to support the real value-added work that needs doing. If that means that you don't have as many pretty Visio diagrams, that's ok - nobody should measure the value of EA by how many pretty Visio diagrams they're producing.

Those statements all sound pretty common sense, and almost silly to have to state, but that's exactly how you combat the tendency to retreat into the Ivory Tower.

Feel free to pile on Ivory Tower Architecture below with examples of where it's gone wrong, or other implications I may have missed.