A herdbook, not a farm app

An animal arrives with a national number, usually an ear tag, sometimes an RFID transponder, and — if it was imported — a number issued somewhere else entirely. The herdbook’s first job is to hold all of them at once and to notice when they disagree.

The international form follows ICAR: country, sex, twelve digits, and the three-letter breed code. For an imported animal it is built from its country of origin, never from ours — a Dutch cow does not become Russian by being written into a Russian book.

Thirty-two countries are recognised for identity construction. A transponder number that fails the ISO check digit is stored but never exported as valid.

Ancestry rarely comes from one place. Part of it is on the certificate the farm sent, part in a bull catalogue, part in an earlier import. The herdbook assembles the tree from all of them and remembers which fact came from where, so a later contradiction can be traced instead of argued about.

Test-day results are stored as they were measured, and lactation figures are derived from them rather than typed in. That ordering matters: a lactation total someone entered by hand cannot be re-checked, while one computed from test days can be recomputed the day a rule changes.

Classification is recorded trait by trait on the linear scale, and the composite scores are computed from them. Who did the classifying is stored with the record — a score without a classifier is an opinion, not a measurement.

An index is an opinion about what matters, expressed as weights. Rather than hide one opinion inside a single number, the system carries several published profiles side by side and shows the weights outright, so a breeder can see why two rankings disagree.

Bulls, heifers under a year, heifers over a year, cows. Rank is computed over the entire category first and the herd filter is applied afterwards — so a farm sees where its animal stands in the country, not where it stands among its own.

A herdbook that accepts whatever it is sent launders bad data into official-looking records. So the system runs its own checks — on identity, pedigree consistency, event ordering, access rights and every export — and publishes what they found.

Findings are not hidden while they are being fixed. An association that shows only green is either very lucky or not looking.

Data a farm cannot take with it is data held hostage. The system serves the ICAR Animal Data Exchange interface (version 1.5, JSON Schema 2020-12 and OpenAPI 3.1) over 14 collections, location-centric as the standard prescribes.

Where a national registry is mandatory, its exports are generated too — 20 templates in the Russian case. That obligation belongs to a country, not to breeding: a herdbook elsewhere simply does not carry it.

The system is mapped against 10 ICAR guideline sections and 23 compliance items, each with its state and the evidence behind it — a check that proves it, a page that shows it, or the code that does it. 6 items are partial and 10 are not started, and they are listed as such.

Two things are named plainly on those pages: the association is not an ICAR member, and ICAR marks are status-based, so no ICAR logo appears anywhere here. Following the guidelines and being certified against them are different claims, and only the first one is being made.

One association runs on this system today. Its book is open to look at — the interface is Russian, but the structure is the one described above, and numbers read the same in any language.

For a walkthrough in English, or to talk about keeping your own book on it, write to mail@plem.online.