Are we making a Category Mistake?

[Really a new thread, so resending in a self contained way with a change of
subject]


In response to my question about preflight, Anne said:

We are concerned that a per-origin model would not be implemented correctly.
In addition it would be somewhat of a pain in case of different services
maintained by different parties hosted on a single origin which we expect to
be reasonably common.


Exactly. I agree that we should assume that an origin is a largely
uncoordinated set of "different services maintained by different parties".
We are not serving the interests of an origin by enabling an origin as a
whole to opt-in by preflight, since an origin's interests lies in
recognizing the diversity of interests, behaviors, and assumptions hosted at
that origin. For the same reason, we are not serving the interests of an
origin by adding that origin's name to the ACL for a resource. The origin is
not a unitary intentional entity that should either be or not be allowed
access as a whole. Likewise, the intentional entities that should be allowed
access are defined by internal coordination within the entity, not by origin
boundaries.

This group had the wisdom to avoid the "origin is a unitary intentional
entity" category mistake in designing the preflight -- even though the cost
of avoiding this mistake was high. We should exercise this same wisdom in
deciding how to allow access to resources.