Peak Architecture

I believe peak architect occurred somewhere around 2013, that was the year with the most people employed with architect titles ever. Since 2013, we are now entering the post-big-architect era, where architect roles are quickly shrinking in the tech world. This is a good thing.

The number of roles with architect titles in industry was getting ridiculous. I can riff 20 titles in less than a minute if I try: Data, Technical, Web, Digital, Application, Cloud, Security, Infrastructure, Network, Business, User Experience, Information, Middleware, Enterprise, Solution, Sales, Product, Project, Principal and my personal favorite, Sharepoint. The question is, what did all those architects do?

My contention is that they completely messed up most companies by proclaiming they knew how to select and manage technology at large companies, and forced vast numbers of engineers to conform to a highly limited set of technologies in the name of fiscal conservatism. Then retreated to their own worlds to rest and consider the next great set of technologies they would enshrine five years from now.

What is the optimum number of architects for a company?

The answer to this, like all math problems, is 0, 1 or some hard to derive number. Zero and one are actually quite good answers. Zero is complete engineering anarchy, you’ve decided to let the engineers do whatever they want. If they’re good engineers, they might actually make an amazing system by developing external APIs which hide all the technology choices from the outside world.

One is also a great answer, you’re basically saying someone needs to think about the company as a whole, and one insanely smart person can get a whole company pointed in the right direction, or they can spend a lot of time thinking and watching engineers do whatever they want. One architect can spawn ideas, if they’re in touch with the engineers, that just may be enough to get people pointed in the right direction.

A better number is some ratio to the number of engineers. But what ratio to pick? Too low and you have architects running over each other to tell engineering teams what to do. Something like the commissioned sales model, “that team was mine, I talked to them first.” Too few and you’re not much better than 0 or 1. My personal favorite ratio right now is 1:100, one architect for every 100 engineers. That seems to be enough to get some rational standardization (if you don’t like standardization, deal with it, every company beyond startup mode has some set of standards), while allowing engineers to fight the power and pick new technologies when necessary. Since technology is progressing faster and faster, it’s essential to keep new technologies flowing into an organization, and old ones leaving (sometimes 6 months is old!). Relying on a few individuals to follow these trends and pick the right tools for everyone is impossible. But sourcing newness from those engineers willing to push a new idea over a few intentional roadblocks will keep your organization current enough to remain relevant.

Peak architecture has passed, time to jump on the bandwagon and become engineers again!