About document signing

On Mon, May 18, 2020 at 11:50:30AM +0000, defdefred wrote:
> 
> Why not simply using GPG?
> 
> Signing all document at publication time (oneshoot computation) and 
serving gemini page with server/path/document.gmi + 
server/path/document.gpg could assure you that the original document is 
free of hacking and don't break minimalist gemini browser.

Unless I've badly misunderstood something, this solution:

1) Provides literally nothing in the way of confidentiality (the first
request happens totally as plaintext, right?).  That means your ISP
can monitor everything you read via Gemini and compile a behavioural
profile to sell to marketers and/or to have stolen by state security
serivices.  States with censorship regimes can requires ISPs to
terminate connections when forbidden keywords are detected in plaintext
content.  In this respect, Gemini would be no improvement over Gopher.

2) Provides authentication and integrity if and only if the client has
some totally unspecified way of securely acquiring the public key
required to validate the signature downloaded in the second request
- i.e. if we assume that the really hard part of public key cryptography
is already solved.

TLS certainly has problems, but it's better than this!

Of course, there is nothing wrong with signing files to be distributed
via Gemini and if this makes sense for people they can and should do
exactly this.  But it's not in any way a sensible blanket solution.

Cheers,
Solderpunk

---

Previous in thread (2 of 9): 🗣️ jan6 (a) tilde.ninja (jan6 (a) tilde.ninja)

Next in thread (4 of 9): 🗣️ kaoD (elkaod (a) gmail.com)

View entire thread.