It occurred to me that there are 2 distinct points at which the SP 'learns' something about the user in basic SSO - the first is whatever bit of information is provided to the SP to enable IDP discovery, and the second is whatever the IDP, once discovered, asserts to.
Our various identity systems differ in both steps, as shown below.
Time is the horizontal axis, the vertical is a representation of how 'much' the SP knows at any one time
Normal caveats about over simplification, not drawn to scale, etc.
I was hoping this would lead to some sort of "Madsen's Privacy Law of Minimal Area" - expressing the premise that, all else being equal, the area under the curve should be minimized for optimal privacy but can't quite reconcile that with Credentica's tech having greater area than others. Maybe I should just ignore it - I wouldn't be the first researcher to do so for data that didn't fit my theories.