upvote
Never heard of DNTLS, thanks for sharing. I skimmed the Website and am a bit uncertain what the ai angle is. Shouldnt naming be, well, just naming?
reply
Yeah, the website is a holdover from when we were pitching AI VCs a few months ago. We quickly determined that the whole system is now "Cancer Capital" (see the other front page HN thread) and have pivoted to just bootstrapping from a close syndicate of like minded individuals. We plan to update the website this month, sorry it's a bit behind.

In short, AI identities were just a happy accident that comes with the system/architecture. It's not tied to AI at all.

But if anyone is interested in talking about what we're doing more, happy to connect at hn@sepositus.com.

reply
Do you have a brief explanation of what DNTLS is that you could post here?
reply
Sure, happy to put something here as a comment.

Alice registers `alice.dntls` and Bob registers `bob.dntls` on the DNTLS network. During the registration process, they generate PQ key pairs that are registered along with the name. Alice's and Bob's name are hashed before being stored on the network. Bob knows Alice's name, so he can perform the necessary hash computation to look up Alice's public key material on the network. Likewise, Alice can do the same for Bob.

Bob wants to send a file to Alice. Bob takes his name key and signs the document with it and sends it to Alice. Alice looks up Bob's public key material on the network and verifies the signature.

Bob now stands up a new website, but he only wants Alice to access it. He sets up a standard HTTP server but slightly modifies it to be "DNTLS native." He does this by requiring mTLS on incoming TLS connections. The connecting party must identify themselves with a signed certificate. Each name has what we call a "name record" that allows publishing arbitrary metadata signed by the name key. Bob publishes a standard "HTTP" record in his own name record that points to the IP address. Alice now goes to connect to Bob's website. She opens her "special" browser and types in bob's name. The special browser looks up Bob's name record, finds the published IP address, and attempts an mTLS connection. Bob's server is configured to _only_ allow connections from Alice. Since Alice signed her TLS connection with her own name, the connection is allowed, while every other is rejected.

Alice now wants to communicate with Bob's agent. Bob publishes a subname called `agent.bob.dntls`. In that subname's record he publishes an A2A packet that contains the information for connecting to his agent. But, like the website, the agent is listening on a TLS connection that rejects anyone except Alice. She uses an A2A tool to initiate a connection using her name key and is allowed to make a mutually secured connection to Bob's agent.

Bob wants to connect to a VM he purchased that runs the website. He configures SSH with his name key as one of the recognized users. His SSH connection simply leverages the name key to authenticate him to the machine. But he shares the machine with another person and wants to share a secret with them. So he creates a SOPS encrypted file with his name and this other person's names as the only recipients. They both securely access the secret using their respective name keys.

I'll leave it there, but hopefully that's descriptive enough.

reply
The technical implementation aside, "what" prevents this from going down the exact same path as ICANN? E.g. how do we know next year the registration fees won't be 5x or certain registrations denied?
reply
A good question. Unfortunately, all I can say is we have a solution that involves specific European countries, laws, and regulations. As I said in a previous comment, happy to share more outside of this venue, but not all details are 100% public at this point.
reply
If your solution can't survive public scrutiny, it doesn't seem like a trustworthy solution?
reply
We're simply choosing to build in a small group of closely aligned individuals and organizations right now. Many will become the first participants in helping run the public network across the globe. The network will launch completely in the open, including open sourcing the code. People can choose to wait until then to make their decisions or if they are particularly interested they can reach out to me about joining earlier.
reply
A solution built in isolation by a group who just publicly tried to ride the AI wave with it is not a promising set of signs, but I have no problem leaving it at that until all is public and hoping it regains trust then!
reply
deleted
reply
Does alice "own" alice.dntls after she registers it? If so, how do you prevent name squating?
reply
Names are not able to be traded, so it's first come first serve. It's impossible to completely defeat name squatting, but you can make it less enticing. In the above case, Alice can "sell" her name to Bob on some aftermarket but the only way to do so is by transferring the cryptographic material. This is fraught with issues, especially considering once Bob has the keys he has no obligation to pay.

Names are valid for one year and range from $10/yr up like current domain names. Letting a name expire opens it back up to being registered again.

We have quite a few other "tools" in play behind the scenes that make name trading/squatting extremely impractical, but I won't go into those details here :)

reply
Seems like you could use an escrow to defeat this and sell domain names safely and easily.
reply
Like I said, it's impossible to "defeat" it outright because the concept doesn't really lend itself to this sort of structure. If I buy a plaque with my name on it , therefore owning it, the only thing that could really keep me from selling it would be some centralized authority. Even then, black markets have existed for drugs for centuries.
reply
the only real way to have a decentralized domain system is something like https://github.com/pubky/pkarr

but you lose the ability to have short domains.

also until such a time that pkarr is widely adopted, you are better off using .onion domains anyway. it becomes a question of requiring custom DNS client vs. Tor browser.

both approaches use a DHT.

reply
There is already ENS.
reply
The adoption has basically been non-existent. I have a lot of theories on where they've gone wrong. Being so heavily tied to a blockchain (Ethereum being the worst due to its state bloat) is probably the biggest mistake I think they made.
reply
isn't yours tied to crypto and AI?
reply
No, see the above comment.
reply