Lost in translation

An interesting problem I've noticed in many software businesses is that you find this understanding barrier between software people and non-software people. A non-software person just can't understand what a software person is doing – at least, not without becoming a software person themselves. So they have to rely on indirect understanding: either measuring that person's eventual output, or being willing to trust that they are doing the things they say they are.

Of course, you can just hire a software person to tell you whether the other software person is doing what you expect. But then you've just moved the barrier. Now you need to trust the new person. I think it's for this reason, among a few others, that software businesses are generally considered to be best run by software people. That way the management doesn't have that kind of understanding barrier.

But software hasn't necessarily earned special unique snowflake status on this. I realised later on that, in fact, I have the same understanding barrier with sales. I know the general idea of sales: you convince people to buy things. I can tell if it's going well (people are buying things), or badly (people aren't buying things), but beyond that I don't really have the skills to evaluate. Beyond looking at those obvious measurements, I wouldn't be able to evaluate whether a head of sales was doing a good job.

I believe this is a general problem with work in any kind of expertise-driven field; either you need to have the expertise to evaluate it, or you have to trust someone else who does. It's concerning how many instances of this problem the average company must have and, of course, the obvious resulting failures like embezzling. But I also think it's worth considering how many companies you might never hear of because they try to swing for the fences in two expertise-driven fields, say, software and architecture, and run into the understanding barriers of both.

So is the answer just to find trustworthy people to act as your understanding barrier sherpas for every field you might need? Maybe, but that strikes me as an inherently fragile answer. Even if you can reliably and scalably find trustworthy sherpas, they still have to translate things into concepts that are meaningful to you, which creates limits in what you can understand and what you can contribute back.

If that's important, then I think it's better to just learn the things. Become a software person, become a sales person, become an architecture person; at least enough that you can understand and evaluate the work of others. If you aren't able to do that, you'll just end up interacting with them superficially, like a tourist who doesn't speak the language.