Greg Price

@kevinrns @Di4na @linuxiac Unfortunately that's not what the original source says (which they helpfully link to): https://www.numerique.gouv.fr/sinformer/espace-presse/souverainete-numerique-reduction-dependances-extra-europeennes/ The move off Windows to Linux is specifically for the DINUM itself. The mandate for all ministries is much looser: a plan for "reduction of extra-European dependencies". That can mean a lot of things. The DINUM Linux migration is one of three examples they give; the other two are about various SaaS services.

@djb I don't understand why these folks say that the motivation for hybrid schemes is about "harvest now, decrypt later". Seems a lot more straightforward than that: if you deploy a new PQ scheme on its own, and then someone has a break in that scheme, that's bad! That's bad if it means they can forge signatures, just like it's bad if it means they can decrypt messages. Whereas if instead you deployed a hybrid with a non-broken scheme, then you're still secure. Are they… assuming that if there is such a break, everyone will just switch right back to a proven scheme? That seems confused in all sorts of ways (not least because the break might not be public). But that's the closest I'm coming up with to a rationale for that story.

@crazyeddie @kevinrns @Di4na @linuxiac One of the things I think everyone involved in that policy has in mind is last year when the current US administration ordered Microsoft to cut off the email account of a European official working in Europe for an international institution, because it didn't like that institution doing its job: https://apnews.com/article/icc-trump-sanctions-karim-khan-court-a4b4c02751ab84c09718b1b95cbd5db3 That was all over French news when it happened, and for good reason. I think it radicalized a lot of Europeans on this subject, showing that the use of perfectly ordinary tools could be turned into a weapon if they're run by a US company.

@mattblaze@federate.social A real friend would have said "dude, that's scummy and also stupid, what are you thinking, drop it". So if he does have real friends, he probably knew they'd say that and didn't ask them. He knew what he was doing was scummy, so asked his most slimeball buddy instead.

@Elliptickiwi @djb Yeah, I think that's the most plausible argument for dismissing hybrid crypto. If practical QC for cryptanalysis comes very soon, then layering on ECC or any other non-PQ crypto won't be of any use. So if you're real sure of that timeline, then you may as well skip it. (I agree with you and Dan that that level of certainty seems unwise. Even if one thinks that timeline is 80% likely, then cheap measures that maintain security in the 20% case are worthwhile.) Even then, though, I don't see what the HNDL story has to do with it. For one thing, in that world, the QC means the non-PQ layer is easy for the adversary to "decrypt later".

@djb @multisn8 Moreover, to further spell it out for any readers not familiar with the term: a literal safety blanket is something used by very small children, aged about three or four. So saying that hybrid schemes are a "safety blanket", as the first post linked above does, is saying that people who think hybrid crypto is needed are acting on childish fears of the kind that an average five-year-old has grown out of.

@dreid Good question! I am also interested in the answer.