Antipatterns
How pattern misapplication produces the failures the framework names
The recognisable failure modes of pattern misapplication: Partnership done as Delegation, Delegation as the unexamined default, Agency without supervision, Utility without graceful degradation, and the named failure typologies from training, integration and piloting.
Overview
What the antipatterns catalogue covers
- The antipatterns catalogue names the recognisable failure modes that emerge when one of the four cognitive patterns is applied to work that belonged to a different pattern. No pattern is intrinsically superior to another; each antipattern is the wrong shape for the work it has been assigned, not a flaw in the pattern itself. The catalogue is the framework's diagnostic companion to selection: selection sets the allocation, antipatterns describe what goes wrong when the allocation is mistaken or quietly drifts.
- The catalogue is organised in three layers, each addressing a different scale of failure. Per-pattern antipatterns sit closest to the canon: Performative Partnership, Drift to Delegation, the Unexamined Default for Delegation, Mandate Drift for Agency, Cognitive Misfit for Utility and the rest. Cross-pattern failure modes name shapes that recur across more than one pattern, including Theatre Across Patterns, Implicit Pattern Drift and Undesigned Fallback. A third layer collects named failure-mode typologies inherited from The Cognitive Partnership: training failures, integration mistakes and pilot failures.
- Each typology is treated as a reusable diagnostic. Every entry in the catalogue follows a consistent shape (name, description, consequence and diagnostic signal) so a practitioner who suspects an antipattern can confirm the match without re-reading the canon. The catalogue is most useful paired with a workflow map and the recorded pattern selection: the antipattern names what is going wrong, the map shows where, and the selection record reveals which assumption has stopped holding.
Contact
Want to diagnose an AI deployment that is not going well?
Reach out for a conversation about the signals your team is seeing and which antipattern they might map to.
hello@nutropic.ai