Thread by @GeePawHill

Yesterday, we talked about Alice's City On The Hill and her approach to getting there.

geepawhill.org/pro-tip-the-al…

I offered, instead of the Alice approach, an approach that was Human, Taken, Local, and Iterative.

Today, let's consider this business of Procedural -> Human.

Every system for making software is mixed system, with three kinds of thing in it: the human kind, the artifact kind, and the procedural kind.

Humans are, you know, *persons*. Artifacts are things like the code, the tools, the documents, the tests. Procedures are the descriptions -- usually framed as rules -- by which the humans interact with each other and the artifacts: meetings, forms, rules, that kind of thing.

Procedures and artifacts get intermingled a lot, because often the goal of the procedure is to transition from one artifact to the next. Since this is so common, let's go with it, we'll lump them together and call them the non-human parts of the system.

So. To go from a procedural approach to change to a human approach begins when we realize how much more powerful the human parts of our mixed system are, compared to the non-human parts.

Now, the non-human stuff in our mixed-system is powerful, there's no question of that. But the human stuff in our mixed systems is *super*-powerful.

Humans always dominate in mixed systems, and they can do it for good or ill. That's important, "for good *or* ill", so let's take a couple of cases.

In Dekker's _Field Guide To Undestanding Human Error_, he describes (and advocates) an ongoing revolution in the trade of "incident investigation". Dekker's an expert in airplane incidents, but has spent his career looking at all sorts of similar situations.

In old-school investigation, you'll notice an extraordinary number of accident conclusions that are called "operator error". This is formal-talk for "that person made a mistake".

Now, this is usually technically accurate, in the sense that there usually *is* a human there, and the human usually *did* break a procedure, and that usually *was* the most proximal cause of the incident.

Case closed, yeah? Dumb humans and their inability to follow rules.

Not so fast. See, the humans in these systems are bright, successful, hardworking, very well trained. Further, they're very *good* at following procedure. When one doesn't follow procedure, we have to ask *why* she didn't. When we do that, our story very often changes. A lot.

Every case is different, go read his work, but the cases I find most fascinating, one level back from their details, have very similar patterns. They amount to this: the human broke the rule because the human is smarter than the rule most of the time. This time, it didn't work.

See, humans, for *good*, routinely break procedure, even in as procedure-laden a business as flying an airplane. And they do this because they are more successful when they do.

There are *lots* of variations on this theme, and lots of implications about system design (not to mention the irresponsibility of tagging "operator error"). Read Dekker's work, as I've over-simplified it here.

But the thing I want to get across: in mixed systems, the new school says, far from being the source of most system failures, humans are the source of most *success*. (In incident investigation, of course, success = safety.)

In mixed systems, the humans selectively and frequently break procedure in order to make the system successful.

What about the human superpower over the non-human used for ill? This one's easier to see. There's a kind of labor strike called a "work to rule" strike.

In a work-to-rule strike, we all show up for the job, and we all follow the procedures *exactly*. We treat them as perfect unbreakable law, inviolate.

Here's what happens: actual work grinds to a halt almost immediately.

This kind of strike is very effective, because there's nothing owners can do about it. Anyone who's ever had a sullen teenager knows how infuriating doing exactly what was said can be, and how utterly non-productive.

So I hope you're getting a feel for this: in mixed systems, with humans, artifacts, and procedures, the most super-powerful elements are the humans. They can make the mixed system successful, or they can easily make it implode.

The implications of this for our approach to change are enormous, in obvious ways and more subtle ones. We'll talk soon about some specific ideas for taking advantage of that insight.

For the moment, I needed to get the insight itself across. Again, for the homework, just let that soak in. Maybe recount a few of your own stories about this good *or* ill.

If we're talking about an approach to the City On The Hill, we want to focus that approach not on the procedural, but on the human, because the humans in our mixed system will be the deciding factor in our success.

Have an unusual and pleasantly surprising Sunday! I'll talkatcha soon.