The law of requisite variety has fascinated me since I first learned about it several years ago in a course on cybernetics and design. The law is rigorously defined in W. Ross Ashby‘s Introduction to Cybernetics, but for our purposes we may loosely say that the more variety a system has, the more options it has in a given situation — and the more options it has, the more capable it is of obtaining a desired state of affairs. As Ashby points out, this offers a way of understanding the survival value of certain apparently useless characteristics of complex organisms:
This point of view enables us to resolve what might at first seem a paradox—that the higher organisms have sensitive skins, responsive nervous systems, and often an instinct that impels them, in play or curiosity, to bring more variety to the system than is immediately necessary. (212)
Organisms “bring in” variety by experiencing variety and internalizing it (aka learning from it). The more variety they acquire, the better able they are to maintain the parameters that allow them to survive. Thus, in the shifting conditions of life, it is important to acquire more variety than is immediately necessary by experiencing a diversity of stimuli and situations.
The same goes for organizations. The more variety a company has, the better able it is to cope with changing conditions. Companies already understand this this principle in a simple way with respect to investments and products, where it’s known as “diversifying”. So why is it that when I glance at the job boards, even for forward-thinking groups like the Rails community, that non-technical qualifications are almost never mentioned?
The ability to communicate well, for instance, is something that makes almost any person more effective. Effective communication implies verbal sensitivity. With regard to variety-acquisition, verbal sensitivity functions much the same way as skin sensitivity does for all higher organisms. The more carefully a person listens, the more variety he will acquire through conversation.
Likewise, excellence in any area is evidence that a person has the capability to do great things. If those great things are in an unrelated domain and in addition to programming expertise, that should be a very good sign for a hiring manager. It indicates broad intelligence and exposure to experiences which will have produced a greater variety in the individual, and thus a greater flexibility of thought and action.
Several years ago, I applied for a job at a startup that had a unanimous hiring policy: the whole team had to agree or the person wasn’t hired. I got turned down, despite the fact that the founder liked me, because one of the developers thought I had insufficient “passion for web development”. My response is that web development is exactly the wrong thing to be passionate about. It’s the wrong level of detail, especially for a startup. Development is what you practice so that you can create awesome products, or, for the more ambitious, so that you can change the world. Technical excellence is a prerequisite, not the end itself.
That’s why I always cringe when I read ads looking for “ruby rockstars” and the like. What they’re really saying is “we want people for whom the most important thing is the how, not the what”. A corollary is “we want people whom we can use to achieve our fixed ends, not people with whom to discuss and discover ends”. For startups, business models can change from week to week. Thus the ability to think about ends and not merely implementations is of the utmost importance. Hiring a “ruby rockstar” is a bit like hiring an editor who is a “grammar rockstar” instead of one who is actually thinking about the topic and critiquing the ideas.
To return to the story, why is it such a poor idea to hire by consensus? Because it guarantees monoculture. It limits the variety of employees to the most recognizable and common sort. It decrease the variety of the organization and hence reduces its ability to produce desired outcomes.
“But what if I just want my employees to do what they are told?” Go ahead, hire those people. You just removed all of the reflective intelligence from your organization. Next time you ask for something stupid, you can be sure you’re going to get it.