AI Sands Off the Friction That Once Made Us Truly Know Our Machines

The essay draws a distinction between two kinds of technical knowledge. In the 1990s, using a computer meant fighting it: editing autoexec.bat, building single-game boot disks, setting drive jumpers, memorizing which interrupt a sound card answered to. The machine set terms, and you learned its internals because it wouldn’t cooperate otherwise. That resistance, the author argues, was the entire medium of understanding — you only really know a thing that can push back and defeat you.

AI assistants invert this arrangement. They demand nothing, never make you read a config file, and rearrange themselves around whatever you type. The author’s key move is to separate competence from acquaintance. Competence — raw knowledge of how systems work — is safer than ever, since models have ingested every manual no human reads and can recite them flawlessly. What’s disappearing is acquaintance: the hard-won intimacy of having struggled with a specific machine until it gave. We’re heading toward being simultaneously more dependent on these tools and less familiar with them than ever, both for the same reason.

The closing note is elegiac rather than alarmist. The author isn’t warning that something deep will break with no one left to fix it; the grief is narrower and more personal. Younger users won’t experience the change as loss because you can’t miss a relationship you never had. The people who knew a machine ‘the only way a machine was ever really known — by having it be difficult’ are the last cohort who will even register that anything is gone. It’s a piece about the cultural cost of frictionless computing, not its technical risk.