[cap-talk] Authority vs. Information Flow

[David Wagner]

Toby Murray writes:
>Ahh. I don't understand this at all. I think this must be the root of
>where we diverge.
>
>Consider
>
>def carol(){
>  bob()
>}
>
>def bob(){
>  alice()
>}
>
>(When Carol is invoked, she invokes Bob who invokes Alice.)
>
>We've analysed the code for all three entities. We know this is what
>will happen. Yet I argue that Carol can cause Alice to be invoked.
>Apparently you would disagree, no? 
>
>Could you explain why that is the case. To me, reasoning that Carol can
>cause Alice to be invoked is the most natural thing in the world.
>Notions of what is fixed etc. do not come into it.

I too think this is where we diverge.  Congratulations on an excellent
example.  You've painted me into a corner that I do not know how to
recover from.  I don't have a coherent account of this very simple
example.  I agree that I would like to say that Carol will cause Alice
to be invoked, but I don't have a theory of causation I like that will
allow us to conclude this simple fact.  Obviously I'm more confused than
I realized.  I'm stuck and don't know how to get unstuck.  Sorry.

>I don't see how it makes sense to say "trusted (i.e. code about which we
>know how it will behave) code has no authority". I don't understand at
>all.

A tiny nitpick: Just because we know how an object O will behave does not
necessarily mean that we're prepared to consider O trusted or trustworthy.