Not All Friction Is Waste

We spend a lot of time trying to make work easier.

And usually, for good reason. A process has too many steps, we simplify it. Something repetitive can be automated, we automate it. Information is hard to find, we make it easier to access.

Friction looks like waste.

But imagine someone learning their way around a complicated system. They run into a problem and spend an afternoon tracing what happened: following one dependency into another, getting a few things wrong, asking questions, and eventually figuring it out.

An experienced colleague could probably have given them the answer in ten minutes. That would certainly have been faster. But the afternoon produced something the ten-minute answer might not have: a mental model. The next time something breaks, they know a little more about where to look and what questions to ask.

The learning sciences have studied versions of this for a long time. Research on desirable difficulties has repeatedly found that making learning easier in the moment doesn't always produce better learning later. Some effort that feels inefficient while we're doing it helps with retention and our ability to use what we've learned somewhere else.

I wonder if there is a broader lesson there for how we work.

Consider a team with no room in its schedule. Every hour is accounted for. Every person is fully utilized. It looks wonderfully efficient until something unexpected happens, or someone needs time to explore an idea that may lead nowhere.

Some organizational research finds that a degree of unused capacity (what researchers call slack) can support experimentation and learning. Too much can become wasteful, but too little leaves little room to explore.

The same thing can happen in conversation. A disagreement slows a meeting down. Someone asks a question that reopens something we thought we'd settled. Another person doesn't quite understand and needs us to explain it again.

It would be easy to see these as obstacles to getting through the agenda. Sometimes they are. But sometimes that awkward question exposes an assumption nobody had noticed. Explaining the idea again reveals that we didn't understand it as well as we thought. A disagreement forces two incomplete views of the problem to meet.

The difficulty isn't automatically valuable. Friction can just as easily be bureaucracy, confusion, unnecessary conflict, or badly designed work.

The useful question may be what the friction is producing.

Is it simply consuming time and attention? Remove it. Or is it building understanding, judgment, capability, or a better shared sense of the problem?

That distinction feels increasingly important as AI becomes part of everyday work.

We now have an extraordinary ability to remove cognitive friction. We can summarize instead of read, generate instead of draft, ask instead of search, and get an explanation before we've had much time to wrestle with the question ourselves.

Much of that is genuinely useful. But occasionally it may be worth asking what else disappeared with the effort. The search may have taught us the territory. Writing may have clarified what we actually thought. Debugging may have built an understanding of the system. A conversation that could have been replaced by a summary may have been maintaining a shared understanding between people.

The point isn't to make work deliberately difficult. Difficulty by itself has no virtue. It's simply to recognize that some friction slows the work down, and some friction is part of how the work gets better. As it becomes easier to remove friction, perhaps we need to get a little better at telling the difference.

Further reading