While the screams and roars of the "plz no more #gnupg" crowd can be heard, do consider viable ways for the Linux community to replace it.

It's very easy to point at minisign/ssh and similar projects and go "use that". But it misses the convenience of binding an identity to a key, along with having a key distribution mechanism.

I think standardizing `.well-known/public-key` for lookups, and explore a path where we can jam a tlog into it would be interesting.

https://github.com/C2SP/C2SP/issues/192

Filippo wrote about age-keyserver recently; https://words.filippo.io/keyserver-tlog/

And Saotok is exploring end-to-end encryption for the fediverse; https://soatok.blog/2025/12/15/announcing-key-transparency-fediverse/

Which are approaches I think is extremely relevant to solving this problem.

@Foxboron the easiest way is to switch to an alternative #openpgp implementation, of which there are many. The closest in general functionality is probably sequoia, but most people don’t need a full-fat codebase and can use anything that supports the SOP interface

@andrewg

This post is about moving away from openpgp entirely.

@Foxboron Also, an established web of trust is non-trivial to replace. It's a similar problem to "everyone is still on facebook because that's where everyone is."

@monsieuricon
Yes, especially when the tools you might want to replace it with simply do not concern itself with the problem.

Depending on the specific use case for WOT, I think that a centrally signed bundle of keys could be a viable replacement in some cases.

@Foxboron we do have something in the works to replace our web of trust already -- hopefully coming in the next couple of months. I was already planning to start allowing ssh keys at that point, so hopefully we'll have paved ourselves a way out of our OpenPGP dependency by mid-year.

@Foxboron GnuPG is IMHO great and basis of security for everything I do :-)

@jarkko @Foxboron I'm pretty sure this discussion is in response to https://gpg.fail/ :)

@Foxboron Hardware supports it broadly. Also, I completely rely on 'pass'. Any alternative should be at least fully backwards compatible and not breaking any tools that depend on GnuPG (or OpenPGP).

@monsieuricon @Foxboron @jarkko Then use another PGP implementation. @sequoiapgp may not have the same issues.

The problem I have with GnuPG right now is that they are the most inflexible PGP implementation in all the wrong ways. My time working on RPM and DNF has made me seriously unhappy with GnuPG.

@monsieuricon @Foxboron

dude, not against switching gnupg :-) just worried about hardware and software that rely on it.

Now that I think about it there's no really huge issue in the end of day as presumably new system can also use smartcard keys (yubikey) :-) Sorry, I did not think this through as there is no really any interference with pass.

@neal @monsieuricon @Foxboron @sequoiapgp see my comment to Konstantin. I was too hasty with my conclusions.

@Foxboron @monsieuricon Arch Linux has their own public key authentication/identification processes anyway, just publish a signed list which maintainer is which public key, we don't actually need a baked in web3-like cryptographic-identity system.

If anything, tying "this MPI is my identity" too closely is the reason why many maintainers haven't rotated their signing key in 10+ years. #39c3