3 ms·

I don’t know anything about IPFS and would like to know more. When I visit an HTTPS URL I see content and some authenticity (of the server, at least) tied with

I don’t know anything about IPFS and would like to know more. When I visit an HTTPS URL I see content and some authenticity (of the server, at least) tied with the transport mechanism. IPFS provides the content and a distributed transport. Does the protocol include authentication of the author? Is it up to the content author to include their own signature protocol outside of IPFS? [If you post an ipfs:// link claiming it is a New York Times article, how do I know it’s real?]

hecturchi 6y ago
IPFS has a mechanism called IPNS where any /ipns/author-hash can resolve to /ipfs/hash and the ipns record is signed and can only be provided by that signer/author. But this is just another way of authenticating the "author". You can also use dns (if you can trust it), or you can use signed content, or you can get the ipfs hash through a channel you trust. The main idea though, is that IPFS content is authenticated by default because it is referenced by its own hash. The problem on obtaining a hash you can trust is just a layer above and solvable in multiple ways, as needed.
- gorgoiler 6y ago
  Thanks. That’s informative. I would be happy with the DNS route only if I had faith in DNSSEC receiving wide adoption. As far as I’m aware every resolver on every system I use will silently fail open to non-DNSSEC DNS. It feels like that’s going to be the case for a long time yet, so maybe the rely on DNS you can trust route isn’t the best option. To be clear: I’m not worried about malicious nodes on the IPFS network corrupting data which I’ve asked for. It’s more about how to avoid being phished with ipfs:// links I find posted online whose contents claim to be something they are not. I hadn’t realised how much I relied on the safety of the padlocked domain name at the top of my browser, and on the domain name in links over which I hover. With plain IPFS every click feels like a new website?
issamehh 6y ago
My understanding is that the integrity of the contents is assured due to the identifier being a hash of the content. From there I don't think it's really needed for a whole protocol to verify the author. Couldn't something as simple as a gpg signature be sufficient? If they signed the hash and gave it to you then it would be good
- gorgoiler 6y ago
  I suppose I could copy and paste an inline signature from the ipfs content into some other software. Standard support in the protocol for this would be very helpful though. (‘hecturchi’s response hints at how IPFS achieves this, I think.)
rpdillon 6y ago
Not at all an expert on this, but my impression is the trust model is different for IPFS than for the web. IPFS is a distributed CAS, so the address is the hash of the content (this can be verified by the client). Whereas the web focuses on "I'm talking to who I think I'm talking to", IPFS focuses on "I'm getting the content I asked for". So, to answer your question, I think authorship is an orthogonal concern to IPFS's focus, which is verifiably delivering the requested bytes. All that said, IPNS likely bridges the content/authorship gap you're asking about. > A name in IPNS is the hash of a public key. It is associated with a record containing information about the hash it links to that is signed by the corresponding private key. New records can be signed and published at any time. https://docs.ipfs.io/concepts/ipns/ https://docs.ipfs.io/concepts/ipns/ In this context, that key would be held by the author, allowing the author to publish different content by signing it with the key.
SulfurHexaFluri 6y ago
Its basically the same as a torrent or a public key. You have to first trust the source of the IPFS hash but if you trust the hash you can trust the content downloaded with it. If for example NYT tweeted an IPFS hash, you could trust the content was from them.

hecturchi 6y ago

IPFS has a mechanism called IPNS where any /ipns/author-hash can resolve to /ipfs/hash and the ipns record is signed and can only be provided by that signer/author. But this is just another way of authenticating the "author". You can also use dns (if you can trust it), or you can use signed content, or you can get the ipfs hash through a channel you trust. The main idea though, is that IPFS content is authenticated by default because it is referenced by its own hash. The problem on obtaining a hash you can trust is just a layer above and solvable in multiple ways, as needed.

gorgoiler 6y ago
Thanks. That’s informative. I would be happy with the DNS route only if I had faith in DNSSEC receiving wide adoption. As far as I’m aware every resolver on every system I use will silently fail open to non-DNSSEC DNS. It feels like that’s going to be the case for a long time yet, so maybe the rely on DNS you can trust route isn’t the best option. To be clear: I’m not worried about malicious nodes on the IPFS network corrupting data which I’ve asked for. It’s more about how to avoid being phished with ipfs:// links I find posted online whose contents claim to be something they are not. I hadn’t realised how much I relied on the safety of the padlocked domain name at the top of my browser, and on the domain name in links over which I hover. With plain IPFS every click feels like a new website?

issamehh 6y ago

My understanding is that the integrity of the contents is assured due to the identifier being a hash of the content. From there I don't think it's really needed for a whole protocol to verify the author. Couldn't something as simple as a gpg signature be sufficient? If they signed the hash and gave it to you then it would be good

gorgoiler 6y ago
I suppose I could copy and paste an inline signature from the ipfs content into some other software. Standard support in the protocol for this would be very helpful though. (‘hecturchi’s response hints at how IPFS achieves this, I think.)

rpdillon 6y ago

Not at all an expert on this, but my impression is the trust model is different for IPFS than for the web. IPFS is a distributed CAS, so the address is the hash of the content (this can be verified by the client). Whereas the web focuses on "I'm talking to who I think I'm talking to", IPFS focuses on "I'm getting the content I asked for". So, to answer your question, I think authorship is an orthogonal concern to IPFS's focus, which is verifiably delivering the requested bytes. All that said, IPNS likely bridges the content/authorship gap you're asking about. > A name in IPNS is the hash of a public key. It is associated with a record containing information about the hash it links to that is signed by the corresponding private key. New records can be signed and published at any time. https://docs.ipfs.io/concepts/ipns/ https://docs.ipfs.io/concepts/ipns/ In this context, that key would be held by the author, allowing the author to publish different content by signing it with the key.

SulfurHexaFluri 6y ago

Its basically the same as a torrent or a public key. You have to first trust the source of the IPFS hash but if you trust the hash you can trust the content downloaded with it. If for example NYT tweeted an IPFS hash, you could trust the content was from them.