21 architecturalist papers: always be right, right now.

(From my personal blog: http://wrongtool.kostadis.com/21-architecturalist-papers-always-be-right-right-now/)

One of the particular challenges of being a software architect is that the average manager and vice-president of engineering assume you are incapable of adding value, right now.

One of the particular challenges of being a software architect is that the average manager and vice-president of engineering assume you are incapable of adding value, right now.

Engineering managers have releases they need to deliver to now. They have engineers who have problems now. The managers have budget squabbles now. The teams have debates with product management now.

An architect has visions of the future, and those visions are often years away from delivery, and worse, even more years away from solving any of the immediate problems engineering management has.

And so the VP of engineering begins to see the architect as a distraction. The architect distracts in two ways. The first is that they are always complaining that the team is not investing in the future. The second is that the things they want are unimplementable. And worse, the architect is typically unwilling or unable to go figure anything out. Powerpoint gets produced, confluence pages get edited, wiki pages get updated, and in some cases, small prototypes get written, but no useful code gets changed.

And thus, the VP tries to manage that distraction. There are two fundamental approaches. The first is to keep them away from anything that matters and fire them at some opportune moment. The second is to focus them on a small problem. A small problem keeps the architect from distracting other people and makes it easy to measure their delivery of value. In other words, force the powerpoint jockey to write code, and either they figure out how to write code, or they get fired.

And you know what, the VP of engineering is right.

The thing about being a software architect is that you always have to be right. What I mean is that you are guiding a team, and the direction has to be correct. If it’s not correct, then the team is heading towards a catastrophe. You can change course over time, but it always has to be the right course.

In addition to always being right, you also have to be sufficiently high-level, so everyone is doing the right thing. Pulling this off is another tricky thing to master. If the correct answer is not inclusive of the entire organization, then somebody is doing something wrong. Furthermore, if it’s too low-level, then there are no white-spaces for engineers to innovate.

Another customer for the long term architecture is the CEO/GM and Product Management team. They have to see enough value that they can talk about it to their customers.

In short, you need to be right, and high-level enough that you can’t be wrong, but if that’s where you end, you fail.

The software architect must also add value right now. What does that mean? It means if the head of product management chooses to fund X features, all X are stepping stones to your architectural vision. If the engineers have to design something, it should be evident from the high-level architecture what they should do. If the managers have to figure out how to trade off long term vs. short-term execution, they should make that decision in the context of the global vision.

But it’s more than that. Managers operate in terms of things that need to get done with some set of resources. And so it’s vital that the architecture, be broken down into discrete tasks.

And so my mantra,

• I must always be right.
• I must be sufficiently high-level to be always right.
• I must be right, right now.