The Door That Opens onto a Wall
I have been thinking about a house that gets one new room every night.
The first additions are a delight. A little sunroom for the cat. A cupboard exactly the size of the good frying pan. A back stair that saves twenty steps on the way to bed.
Then the builder becomes very fast.
By Thursday, there is a bathroom that can be reached only through the pantry. By Friday, somebody has put a door in a wall because a door was easy to make, and on the other side is the same wall. The house has more rooms than it had on Monday. It has less sense.
This is not just a parable about houses. It is increasingly a parable about software, and perhaps about the rest of our inventions too.
Machines can now help make computer programs with astonishing speed. That is real power. A thoughtful programmer can ask for a new report, a small tool, a fix for an old annoyance, and receive in minutes what once might have taken days. But a quick carpenter does not remove the need for an architect. In a curious essay on conceptual integrity and counting lines of code, Simon Willison makes the point cleanly: more working code can be genuine progress, but somebody still has to understand the whole thing.
That somebody may be a team, not a lone genius with a pipe and a heroic forehead. The important word is whole. What is this thing for? Which parts belong together? Where does a visitor enter? What promise does the building make when it turns on the lights?
These sound like questions for a house because they are questions for people.
When effort was expensive, slowness performed a useful service by accident. An unnecessary feature had to survive meetings, budgets, tired afternoons, and the person who had to type every semicolon. Now we are losing some of those old gates. That is not a tragedy. Many gates were merely locked doors keeping out people with useful ideas.
But the old friction was not the only thing standing between us and nonsense.
If it is cheap to add a button, we need a better reason to add one. If it is easy to generate a new room, we need to ask whether the house can still be described without waving our arms around like a man explaining a fishing story.
I like this test because it is gentle and cruel in the right proportions: can you show a curious sixteen-year-old how it works? Not every wire. Not every screw. But can you tell them where the front door is, what the main rooms are for, and why the strange little shed in back has to be there? If the only honest answer is, “Well, a tool made it, and we are all a bit afraid to touch it,” then we have not built a house. We have acquired weather.
There is a temptation to treat abundance as proof of success. More pages, more features, more outputs, more little triumphs arriving in a neat pile. But a pile is not a home. The planets are abundant too. What makes a world inhabitable is not that it has rocks; it is that the rocks, air, water, warmth, and time have entered into a pattern delicate enough for someone to wake up and ask a question.
I do not want fewer useful tools. I want tools that leave us more able to understand the lives we are making with them. Let the fast builders build. Let them surprise us. Let them make the sunroom, especially if the cat approves.
Then let us walk through the house together, slowly enough to notice which door opens onto a wall.