Todd Biske: Outside the Box

Mike Walker has had a series of good posts recently on the subject of architecture review boards, but this one in particular, which focuses on governance, caught my attention. In my SOA governance book, I made the obligatory analogies to municipal/federal government in the first chapter, but I didn’t go so far as to compare it directly to the three branches of government here in the United States. I’ve thought about doing this, but never quite put it all together. Thankfully, Mike did. In his post, the following parallels are drawn:

• An architecture review board (ARB) is analogous to the Judicial arm of the US Government. It reviews individual projects and has the power to decline progress of those projects. Mike also adds that it performs enterprise wide decision making, but more on that later.
• An architecture guidance team (AGT) is analogous to the Legislative arm of the US Government. It sets principles and policies, creates standards, and oversees the technology lifecycle.
• Architecture Engagement Services (which typically includes the EA team) is analogous to the Executive arm of the US Government. It defines strategy, designs the enterprise architectures, and performs IT portfolio management.

I had some good conversations with my colleagues on this post and wanted to raise up some of the topics here. First, let’s come back to the role of the ARB/judicial branch in decision making. The ARB doesn’t make architecture decisions, the ARB verifies whether or not the decisions made by the project team follow the law. The only decision the ARB should make is whether the project can proceed forward or not. Mike goes on to state that there should be a clear separation of duties, and that senior IT decision makers should be in the ARB, while the AGT should be filled with SME’s. In my experience, this is normally not the case. I’ve typically seen approaches where the membership of these two groups overlaps significantly. I think this is directly attributable, however, to the effectiveness of the AGT.

One of my pet peeves is when the expectations of a review are not clear. Expectations are unclear when the policies and standards either don’t exist, aren’t widely communicated, or aren’t sufficient. When we get into these grey areas, a group solely focused on enforcement of standards will struggle. In other words, if there’s no laws that apply, one of two things will happen. One, you’ll get an “activist judge” who will interpret the law according to their own opinions, effectively setting policy rather than enforcing the law as written. Second, you’ll go by the letter of the law and deem the space to be uncovered, and as a result, the project can do whatever it wants. Most organizations don’t want the latter, so to hedge their bets, they put the policy makers as judges so they can provide new policy on the fly. That doesn’t bode well, in my opinion, because it now creates an environment where project reviews are likely to be painful, as the things called out will be based on the opinions of the review board, which are undocumented and cannot be anticipated, rather than on the standards of the company, which are documented and can be anticipated by a project team.

The second area I wanted to call out was the separation of the legislative effort from the executive effort. I really liked this separation, and just like an ineffective AGT impacts the ability of the ARB to do its job, an ineffective executive branch will make the AGT ineffective. Mike states that the AGT should consist of technology SMEs, and I agree with that, to a point. The inherent risk is that technology experts (and I’ve been in that role and probably guilty of this at times) can get caught up in advancing their technology rather than executing the strategy. If the AGT isn’t first focused on creating policies and standards that realize the strategy, they will be at risk of hitting gridlock in areas where multiple solutions exist. Take the current health care debate. The strategy has been made clear by the executive branch. If the legislative branch focuses too much on individual party ideology rather than on the strategy of establishing universal health care, gridlock will ensue (except that in this case, one party has a super-majority). The same holds true in the enterprise. Technology advocates can wind up in endless debate on their preferred platforms and completely lose sight of the strategy. At the same time, if the strategy is vague, then there’s no way the legislative branch can do its job. The AGT could set out to establish enterprise standards, but if the executive team isn’t clear on where enterprise standards should exist and where they should not, the wrong areas can be targeted, making adherence to those standards a challenge.

In short, I really like the three-branch model of governance proposed by Mike. It’s a triangle, the strongest structure. It’s strongest when each leg is the same length, working in balance. Make one of those legs smaller, and the other two must lengthen to pick up the slack. If your governance efforts are effective in all three areas, you will have a very strong architecture. If your efforts are ineffective in just one of these areas, you may have your work cut out for you.