- From: Adrian Gropper <agropper@healthurl.com>
- Date: Thu, 11 Dec 2025 20:16:12 -0500
- To: Alan Karp <alanhkarp@gmail.com>
- Cc: Melvin Carvalho <melvincarvalho@gmail.com>, Christopher Allen <ChristopherA@lifewithalacrity.com>, Credentials Community Group <public-credentials@w3.org>
- Message-ID: <CANYRo8jc7hUyuHUDfX+MjVbR+M8=D6SJ1Nofffe5iF=QDo8y7w@mail.gmail.com>
I agree with both Melvin and Alan. The one human that controls melvincarvalho@gmail.com is defined by their biometrics. I use World(coin) as one perspective on human identity. It's not perfect but it exists. Our "definition" of SSI seems to exclude biometrics and that's fine since we get to make the rules. It follows that everything SSI is at least once-removed from biometrics by some (digital) tech. The tech can be self-sovereign if it's compiled from source and self-hosted - with various sub-classes such as open source code that is independently verified and standards-based commodity hosting that does not lock you in. The fundamental aspect of SSI as independent of regulation (in the sense that Bitcoin is independent of regulation) can be satisfied by strict open source self-hosted tech and to a large extent by various sub-classes. Signal might be an example of this. The tech that implements SSI can look like a wallet or an agent and I don't know how to draw the distinction. Can someone help? Adrian On Thu, Dec 11, 2025 at 7:51 PM Alan Karp <alanhkarp@gmail.com> wrote: > On Thu, Dec 11, 2025 at 4:31 PM Melvin Carvalho <melvincarvalho@gmail.com> > wrote: > > >> I've been working on a parallel track, I call it *Self-Sovereign Agents >> (SSA)*. Agents need the same things humans need: >> >> - Identity they control >> - Memory that persists and is portable >> - Reputation that's verifiable >> - Ability to interact without platform lock-in >> > Be careful here. There is one human that is melvincarvalho@gmail.com. > (At least that's the presumption, but you could be sharing that account > with many others.) Things are quite different for AI agents, because you > can have many instances that start from the same model. What identity do > you mean? The model's or the instance's or the party supplying the initial > prompt? Maybe what you need is an identifier for the agent that is tied to > identities, e.g., Melvin's Perplexity Agent #10. > > -------------- > Alan Karp >
Received on Friday, 12 December 2025 01:16:28 UTC