FCO-IM alignment review of the Unified Modelling Schema (YAML). Targeted to 'future' — the format does not yet preserve core FCO-IM semantics, because readings and names are flat strings with no role model or metadata behind them.

Key gaps:

1. N-ary fact types: only supported by reifying into a relationship type (composite PK + one binary relationship per role + free-text TypeReadings). This loses (a) role-to-placeholder binding in readings, (b) the distinction between a true n-ary fact type and an objectified one, and (c) subset/external uniqueness constraints.
2. Names with spaces: FCO-IM object/fact type names routinely contain spaces; embedding them inline in free-text readings is ambiguous to parse and needs quoting/escaping or placeholder-based readings.
3. Tuples/populations: the format is schema-only, so example populations (which FCO-IM is built from) have nowhere to live; a tuple acting as a role player only survives as FK chains between reified types.
4. Multi-locale: names and readings are untagged strings with no locale, so a model cannot carry multiple languages and translations are indistinguishable from alternative readings.

Resolution path requires two upstream additions to UMS: an optional structured reading form (ordered, named role placeholders bound to roles) and a population/instance section. Full feedback drafted in Doc/UMS-FCOIM-feedback.md and sent to the format author.