Answers are grounded in this paper and cite their sources.
Only one kind of machine can test whether something helps against a chronic illness: a body that has the illness. Very many of these machines are running, since people with chronic illness try things on themselves all the time, and almost none of the results are kept. This essay describes that problem, shows why it is built into the situation, and says what would change it. The search is reduced to four facts: a shelf of things one could try, a few of which work; no way to tell them apart except by trying on a body with the illness; weeks per trial; and a noisy answer. Every trial yields two things, a private answer (did it help me?) and a public one (does it help people like me?), and the public one is thrown away. Going through everyone who could collect it (companies, public funders, doctors, researchers, patients) shows that each makes a reasonable move and that the shelf nevertheless stays unsorted; patients are the only ones who lack just one thing, a memory for results. The essay then describes what a shared record of patients' own outcomes, here called an emergence basin, would have to hold: the failures as well as the successes, a person's own outcome behind every entry, and coverage of more than what is popular. Because reporting helps everyone except the reporter, such a record has to give something back to the person who reports. A video platform is treated as the generator of candidates and the patients' record as the place where they are checked. A practical section gives a method that one person can use today: ask who has done what I want to do and succeeded, find those cases (and the failures), and compare them to see what the successes have and the failures lack. The limits are stated: the method sorts and does not prove, it sees benefit before harm, and it is not medical advice. The essay builds on The Denkstunde: A Unit of Account for a Finite Mind (doi:10.5281/zenodo.23413033). The numbers of the worked example come from a small simulation; script and output are included. Languages: the English original, with translations into Chinese, Hindi, Spanish, French, Arabic and German. The English text is the reference version.
No takes yet. Share an insight, caveat, or question.
Bieg et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: