Seven Pages of Proof

St. Petersburg, this 9th of March, 1736. The Königsberg paper went to the printer today, and I confess the thing that occupies me is not the four bridges or the seventeen hundred land crossings I dismissed, but a single sentence I nearly cut: the note that the problem is insoluble because each land carries an odd number of bridge-ends. Three odd vertices, one of four. The proof is seven pages; that sentence is one line. I suspect the one line will be read and the seven pages will not, and I further suspect that whatever is read in those seven pages will be read wrongly, by men hunting for a method rather than receiving one.

Here is the audit I promised myself. My method is always the same: strip the geography, keep the topology. Remove the river’s width, the bridges’ length, the citizens’ walking habits, until only the nodes and their degrees remain. This stripping is what I trust. It gave me the answer. But the assumption beneath it - that nothing essential lives in the discarded material - is the one I trust least. When I threw away the map of Königsberg, I also threw away every question the map could have asked me. A merchant’s complaint about the eastern bridge, a ferryman’s habit in fog: these are edges my abstraction cannot represent, and my method declares them irrelevant before ever examining them. That declaration is not a theorem. It is a habit wearing the costume of a theorem.

The footnote problem follows. The main argument survives because it is stripped; the footnote survives because it is not. The footnote keeps the ferryman in it. Decades later, when the stripped argument fails in some case I did not anticipate, the repair is almost always sitting in that unread line - the discarded detail returning as the missing condition.

So a correction, stated plainly. When I began this page I believed abstraction was lossless if done carefully. I now believe it is always lossy, and the honest practice is not to strip harder but to keep a ledger of what was removed - and to read the ledger before declaring any problem insoluble. I will add such a ledger to the Königsberg paper before the printer closes the form tonight.