I personally think these effect systems are trying to solve the wrong problem. What we really want is capabilities[1], and the means of enforcing them, which I

I personally think these effect systems are trying to solve the wrong problem. What we really want is capabilities[1], and the means of enforcing them, which I suspect is best done with linear types and uniqueness types. Granule[2] is certainly of interest as it combines these two typing disciplines. Austral[3] is another worth looking into, though uniqueness and referential transparency is not one of its goals, it does provide capabilities which are encapsulated in linear types, which serve a similar purpose as uniqueness types for guaranteeing unforgeability. [1]:https://en.wikipedia.org/wiki/Capability-based_security https://en.wikipedia.org/wiki/Capability-based_security [2]:https://granule-project.github.io/ https://granule-project.github.io/ [3]:https://austral-lang.org/ https://austral-lang.org/