The situation that there exist modules which are finitely generated but not projective; finite generation alone does not guarantee projectivity, so finitely generated modules may lack lifting or splitting properties characteristic of projective modules.

Principle

Projectivity is a lifting/splitting property tied to exactness of Hom functors; finite generation is a size constraint. The principle is that size constraints do not imply homological lifting properties unless stronger ring-theoretic hypotheses hold.

Demonstration

Concrete example: over the ring Z, the module Z/nZ is finitely generated but not projective (projective Z-modules are free), so existence of nonprojective finitely generated modules is immediate in many non-semisimple or non-regular rings.

Misapplication

Assuming all finitely generated modules are projective, or using splitting arguments that require projectivity without verifying it; this leads to incorrect decompositions or false lifting of maps.

Consequence

When such modules exist, one must employ homological tools (Ext, projective resolutions, Tor) to study obstructions, construct projective covers when available, and be cautious about conclusions that require projectivity.

Reversal

In contrast, when every finitely generated module is projective (e.g., over semisimple Artinian rings), module theory simplifies markedly: short exact sequences split and classification reduces to direct-sum decompositions of projectives.

Boundary

The assertion concerns finitely generated modules over a specified ring; it excludes infinitely generated modules and focuses on projectivity as a homological property rather than freeness, which is a stronger notion in many contexts.

Semantic Tension

Tension lies between 'projective' and 'free' and between 'finitely generated' and 'finitely presented'—each pair creates subtleties: finitely generated does not imply projective, nor does projective always imply free without extra hypotheses.

Synthesis

Existence of nonprojective finitely generated modules is a common boundary phenomenon showing that finite size does not ensure favorable homological behavior; it motivates the use of resolutions, Ext-theory, and additional ring conditions to recover splitting and classification results.