[FRIAM] *-sovereignty

[uǝlƃ ☤>$]

Ah, thanks. So you were talking about robustness in both cases. Sorry for my confusion.

I suppose there's also some ambiguity in "global". Sometimes I use "non-local" to indicate information bound to the context, but where there are still encapsulated/opaque regions. And then "global" means *everything* is accessable, even if it's encapsulated inside some sub-region.

Your comment about colorings sounds like you're expressing doubt about univalence axiom. But, in my ignorance, it seems to rely on a singular definition of equivalence. For well-defined things like graph coloring, rules like "no adjacent color" seem to establish an unambiguous equivalence. But such things rarely obtain over dirty objects like humans (or even computing in the wild e.g. intrusion detection or forensics). As long as our authentication methods assume such tight equivalences, we'll be susceptible to adversaries. 

A great example are these silly authentication apps. Sure, it's harder to steal one's authenticator creds than it is to capture one's SMS traffic. But the flaws are in the same category. The biggest joke is the big 3 credit agencies tendency to have you verify against your previous addresses ... or vehicles you've owned. Pffft. For those sites that make you select a series of personal questions, I pity the poor fool who answers them with actual historical facts from their life. If you're targeting me and you *don't* know my mother's maiden name, then you're a *terrible* hacker. 8^D


On 10/21/21 8:57 AM, Jon Zingale wrote:
> """
> Could the verifier be allowed a global understanding using something
> akin to homomorphic encryption, though?
> """
> 
> In some sense I would suppose yes for FHE, but the method of verification
> in ZKP seems not to be. Again, you mentioned playing fast and loose with
> the bindings. It would be great to really understand FHE systems better,
> and there is always plowing through the Gentry paper or checking in on
> how far Google has publicly gotten with it. From what I understand about
> FHE, one encrypts some data (whole databases, perhaps) and then one can
> operate on that data in its encrypted form via homomorphisms. Now one
> can operate meaningfully on the data without having access to the data.
> I would suspect it is necessary to present a limited DSL of homomorphic
> actions to make this privacy truly work. One wouldn't want one of those
> accessible homomorphic actions to be to simply decode the database.
> 
> 
> """
> So, I would have said: Just like propositions participate in many
> proofs, identities can employ many agents. But we're playing fast and
> loose with our bindings. "Agent" often means the object/thing, whereas
> "identity" means the attributes of that object/thing. So, maybe you
> accidentally flipped that as you went along in the post?
> """
> 
> I am considering the bipartite (hyper?) graph at the top of Stephen's
> earlier Wikipedia reference[ω]. There they use the word entity instead
> of the word agent. I do mean many proofs for a proposition, for instance,
> the proposition could be that there are an infinite number of primes.
> A piece that I could easily be missing is in the "colorings" formal
> analogy. There, do different formal proofs of a statement give different
> colorings? Is there ultimately an isomorphism between possible proofs
> and possible colorings? This part doesn't seem right to me, I would be
> surprised. So, I know I am missing something.
> 
> [ω] https://en.wikipedia.org/wiki/File:Identity-concept.svg <https://en.wikipedia.org/wiki/File:Identity-concept.svg>

-- 
"Better to be slapped with the truth than kissed with a lie."
☤>$ uǝlƃ