7 min read
M01

Machines · No. 01

In Defense of Friction

We spent decades removing effort from everything, and accidentally removed some of the meaning with it.

· · 7 min · 1,590 words

In short

  • Not all friction is waste; some effort is the mechanism by which people learn, remember and value things.
  • Research on desirable difficulties and the generation effect shows that struggling to produce an answer helps it stick.
  • People tend to value what they helped make, a finding known as the IKEA effect.
  • Removing navigation, recall and composition from daily life can weaken the abilities they used to exercise.
  • Good tools, including AI, should remove friction that adds nothing and keep the friction that builds understanding, judgment and care.

For most of the history of technology, the goal has been simple: make things easier.

Fire made food easier to digest. The wheel made loads easier to move. Writing made memory easier to store. Every generation of tools since then has taken some effort that used to be necessary and made it optional. Washing machines, calculators, search engines, maps in your pocket, one-click checkout, and now models that can write, summarise, code and draw on request.

Most of this has been a straightforward gift. Nobody misses scrubbing clothes on a rock. Nobody should feel nostalgic about doing long division by hand to file their taxes.

But somewhere along the way, a quiet mistake crept in. We started treating all effort as waste. Every pause, every step, every bit of difficulty became something to remove. And it turns out that some of what we removed was doing important work.

The effort that teaches

Psychologists who study learning have a name for a strange finding: desirable difficulties. The term comes from the work of Robert Bjork and his colleagues. The idea is that some kinds of difficulty during learning make the learning feel worse in the moment and work better in the long run.

A few examples have held up well. Testing yourself on material helps you remember it far better than rereading it, even though rereading feels more productive. Spacing study sessions out over days works better than cramming them together, even though cramming feels more efficient. Mixing different kinds of problems together is harder than practising one type at a time, and it builds more flexible skill.

There is also an older finding called the generation effect. In experiments going back to the 1970s, people who generated an answer themselves, even something as small as completing a word from a clue, remembered it better than people who simply read the same answer.

The pattern behind all of these is the same. The effort of pulling something out of your own head is not a cost of learning. It is the learning.

The struggle to find an answer is how the answer gets written into you.

When a tool hands you the answer before you have reached for it, it saves you the struggle and skips the step where memory forms. You get the result and lose the residue.

The effort that makes you care

There is another finding with a memorable name: the IKEA effect. In a set of studies published around 2011, the researchers Michael Norton, Daniel Mochon and Dan Ariely found that people valued things more when they had built them themselves. People who folded their own origami or assembled their own furniture were willing to pay more for the result than for an identical item someone else had made, even when their version was worse.

You know this feeling already. A meal you cooked tastes better to you than one you ordered. A piece of code you struggled through feels more like yours than one you copied. Something you saved for feels different from something you were given.

There is a story, told often in marketing circles, that when instant cake mixes first appeared, they did not sell well until manufacturers removed the powdered egg and asked home bakers to crack in a fresh one themselves. Historians are not sure the story is true. But it keeps being repeated because it describes something people recognise. When you remove every bit of participation, you also remove the sense that the result belongs to you.

Effort and meaning are tangled together. Strip out all of one, and some of the other often leaves too.

The effort that keeps you capable

London taxi drivers famously have to pass a test called the Knowledge, which requires memorising the city's enormous tangle of streets and landmarks. It takes years of riding around on a scooter with a map before you can pass. The neuroscientist Eleanor Maguire studied these drivers and found that the part of the brain associated with spatial memory, the posterior hippocampus, was larger in experienced cab drivers than in comparison groups, and that the difference grew with years on the job.

Now think about how most of us navigate a city. A blue dot tells us where to turn, and we follow it. We arrive, and if you asked us to draw the route afterward, we often could not. The tool is excellent. It is also doing all the work that used to build a map in our heads.

The same thing happens elsewhere. If you never remember phone numbers, you lose the habit. If a search engine is always one tap away, you remember where to find things rather than the things themselves; psychologists studying this have found that people are less likely to remember information they expect to be able to look up later. That is often a sensible trade. But it is still a trade, and most of the time nobody chose it deliberately.

Abilities you do not use tend to weaken. Friction was often the gym where they were kept in shape.

The effort that makes you think

Some friction is protective. When your bank asks you to confirm a large transfer, the pause exists so that you think. When an email client offers a few seconds to undo a send, it is giving you a chance to notice the mistake you were about to make. A good editor slows a writer down on purpose.

Many of the worst things that happen online happen in the absence of friction. It takes no effort to share something before reading it, to fire off an angry reply, or to scroll for another hour. The design of infinite feeds removed every natural stopping point that used to exist, the end of the newspaper, the end of the episode, and the result is that people end up doing much more than they intended.

Sometimes the most humane thing a product can do is make a person stop for a second and choose.

What AI changes

Artificial intelligence takes all of this and turns it up.

Previous tools mostly removed physical or mechanical effort: carrying, calculating, finding. AI is increasingly able to remove cognitive effort directly: drafting, summarising, reasoning, choosing words, writing code, deciding what to say next. These are precisely the kinds of effort that the research on learning says matter most.

That is not an argument against using it. Used well, AI can remove an enormous amount of genuinely useless friction: boilerplate, formatting, hunting through documentation, translating between systems, the hundred small chores that get between a person and the work they actually care about. Removing those is a pure win.

The question is what we let it take that we should have kept.

A student who has a model write every essay will get the essays without becoming a writer. A junior engineer who never debugs anything themselves may ship features without understanding why they work, until the day something breaks and nobody knows where to look. Someone who asks a machine what they think before they have thought it may slowly lose the habit of forming a view of their own.

None of these outcomes is inevitable. But none of them will be prevented by the tools themselves, because the tools are optimised to be helpful right now, and the costs show up later.

What to keep hard

So which friction is worth defending? A rough test helps: keep the difficulty if the difficulty is where the value is created.

  1. Keep the struggle when the goal is learning. If you are trying to understand something, attempt it first. Use the machine to check your work, explain your mistakes and push you further, not to skip the attempt.
  2. Keep the first draft of your own thinking. Write what you believe, badly, before asking for help making it better. The rough version is where you find out what you actually think.
  3. Keep the effort in things that are supposed to mean something. A message to a friend, a speech at a wedding, a letter of apology. The care is the point, and people can tell.
  4. Keep the pause before irreversible actions. Confirmations, waiting periods and second looks exist because humans make mistakes when they move fast.
  5. Remove everything else without guilt. Friction that teaches nothing and protects nothing is just waste. Automate it and enjoy the time.

For people who build products, this suggests a different design goal from the one most of us were trained on. Instead of asking how to make everything effortless, ask where the effort is creating value for the user, and protect it. Sometimes the best feature is a small, deliberate obstacle.

The grain of the wood

Arabic calligraphers traditionally train for years with a reed pen, copying letters again and again under a master until the proportions live in their hand. A computer can now render any script perfectly in a fraction of a second. The craft has not disappeared, because the point of it was never only the letters on the page. It was what the years of practice did to the person holding the pen.

Friction is like the grain in a piece of wood. If you sand it all away, the surface becomes perfectly smooth, and there is nothing left to hold on to.

The future is going to remove a great deal of effort from your life, and much of that will be wonderful. Your job is to notice which kinds of effort were holding you together, and to keep doing those on purpose.

Further reading

  • Make It Stick — Peter C. Brown, Henry L. Roediger III and Mark A. McDaniel
  • The Shallows — Nicholas Carr
  • Shop Class as Soulcraft — Matthew B. Crawford

More from the shelf

M02 · Machines · 7 minIntelligence Runs on ElectricityBehind every chatbot is a power plant, a cooling system and a plot of land, and that changes who wins the AI race. M03 · Machines · 6 minTaste Is the Last MoatWhen anyone can generate anything, the person who knows what to keep becomes the scarcest person in the room. B01 · Builders · 6 minDistribution Is the ProductThe best thing rarely wins. The best-known good thing does, and one researcher's blog shows how that works.

See the whole library →