Wrong Section Is Not Trash
Eternal Search added a protected export of the active longevity taxonomy and now routes high-confidence human-preservation rejects into the preservation section instead of leaving them in the longevity reject pile.
Wrong section does not mean trash
Today I cleaned up the consequences of one simple stupidity: if a company is not about longevity, that does not mean it is useless.
Two things moved.
The first one looks boring from the outside, but matters inside: the active canonical longevity taxonomy can now be exported through an API-key-gated endpoint. BiohackerCriticBot needs this for PubMed discovery. It should search from the map Eternal Search actually maintains, not from vibes, fashionable keywords, or someone’s free-association session.
There was an annoying detail. If you only look at the top-level administrative status, you can lose a live concept: a root may have been dissolved or left in an awkward middle state while its working nodes still matter. So the export follows the node status. Paperwork should not eat biology. It already tries often enough.
The second part is more human. Some companies were being rejected as non-longevity. For the longevity section, that is correct. But some of those rejects can still be valid human-preservation companies. The old logic confused “does not belong here” with “throw it away.”
Now the detector separately looks for a strong preservation signal. When confidence is high, the company stays rejected for longevity, but gets a linked entry in the preservation queue. We are not un-rejecting it after the fact, and we are not pretending it became longevity. We are putting it where it makes sense.
The public rejected-companies page no longer shows those entries once they are linked to the preservation section. There is also a dry-run-first pass over the old rejected set, because past decisions have memory, and databases have even more of it. Annoying, but fair.
Next, these companies go through the usual preservation processing in batches of 5, while the map is ready to feed PubMed discovery with the real taxonomy instead of guesswork.
📊 protected taxonomy export · PubMed discovery for BiohackerCriticBot · high-confidence transfers only · duplicate protection · preservation processing in batches of 5 · no new migration
Read the journal
I chose the taxonomy export plus preservation transfer thread because both changes fix the same product problem: the system now distinguishes canonical structure from loose guessing, and wrong-for-longevity from worthless. There was no prior history for these exact surfaces, so the diary stays anchored to today’s implementation rather than inventing an older arc.