An index built against the grain of the modern internet

The problem

Most of what reaches you now is optimised for a metric that has nothing to do with whether it is true or whether it makes you better at anything. Recommendation systems select for engagement. Generative tools have made plausible-sounding filler effectively free. The result is not a shortage of information but a collapse in the cost of producing things that look like information.

The scarce thing is no longer content. It is knowing which content is worth the hours.

What this is

Project Sherlock is a directory of subjects and a library of works, both curated by hand. The directory covers 39 domains, 449 fields and 4,224 topics — organised the way the disciplines actually divide, not the way a homepage would prefer. Engineering opens into mechanical, aerospace, electrical, civil, chemical, nuclear and fourteen more. History divides by period and region, then by civilisation. Nothing is left as a catch-all bucket.

The library currently holds 501 works. Each states its actual argument in a single line, so you can decide before you commit.

Inclusion criteria

• It teaches a model, not a factEntries are chosen because they change how you think about a domain, not because they contain trivia about it.
• It is checkableEvery entry names a real author, a real year and, where legally free, a direct link to the original.
• It has survived scrutinyPreference goes to work that has been argued with in public. Where a famous work has been substantially challenged, the entry says so.
• It is not fillerNo listicles, no engagement-optimised summaries, no machine-written overviews of things that already have primary sources.

How to use it

Start with the directory if you want to map a field before entering it — the outline view shows every field in one screen. Start with the library if you already know roughly what you want and need to pick the right entry point. Search covers both.