🗄️As We May Think

The record of what people know is growing faster than anyone's ability to find things in it, and the reason is how we file. An index puts each item in one place under one heading. The mind does not work by headings but by association, one thought calling up the next. So build a device that stores everything a person reads and writes and lets them tie any two items together into a trail that can be followed, kept and handed to someone else.

• Read on Friday 24 January, the evening I finished Meadows, and finished on Saturday morning. Long for an essay: most of an hour.
• Published in July 1945 by the man who had run American wartime research. He is asking what scientists should do with themselves now.
• Every linked-notes tool I have used and abandoned quotes this essay on its website. I had never read it. Four rebuilt systems, and I had not read the founding document.
• The first half is a tour of gadgets that did not happen, in loving detail: dry photography, a camera worn on the forehead, a machine that types what you say. The second half is the one that mattered.
• He imagined the links. He did not imagine that I would spend three years arranging them and not writing.
• Read beside 🏛️ Ada Lovelace, who was a century ahead of him on what a machine can and cannot do for thought: 💬 The Analytical Engine has no pretensions. The memex originates nothing either. It keeps the trail, and the person at the desk still has to think.