Conversation

Notices

1. Embed this notice
   lj·rk (ljrk@todon.eu)'s status on Tuesday, 14-Jan-2025 19:21:46 JST lj·rk

   • Rich Felker
   • sammi
   • Joe Lanman
   • 🌱@ambiguous_yelp:ahimsa.chat

   @khm @ambiguous_yelp @dalias @sammi @joelanman While I agree that it's nowhere "close", the "Harvest Now, Decrypt Later" is a tangible threat -- even if they're 10 years out.

   However, I wouldn't bring that up for Signal since that'd require a *lot* of Harvesting, a *lot* of cracking and we'd only get some of the data to decrypt a whole Signal interaction since only the KEX would be affected and things like PFS etc. help against a lot of simpler attacks.

   This *is* a threat for, say, secret govt. documents where deciphering has a big impact even 20 yrs later and the computing requirements (given you have access to quantum computers) are comparatively acceptable.

   • Embed this notice
     Rich Felker (dalias@hachyderm.io)'s status on Tuesday, 14-Jan-2025 19:21:44 JST Rich Felker

     in reply to

     • sammi
     • Joe Lanman
     • 🌱@ambiguous_yelp:ahimsa.chat

     @ljrk @khm @ambiguous_yelp @sammi @joelanman It's more like 10000 not 10.

   • Embed this notice
     Rich Felker (dalias@hachyderm.io)'s status on Tuesday, 14-Jan-2025 19:36:43 JST Rich Felker

     in reply to

     • sammi
     • Joe Lanman
     • 🌱@ambiguous_yelp:ahimsa.chat

     @ljrk @khm @ambiguous_yelp @sammi @joelanman There's a difference between "quite some development" and "quite some marketing". Same deal with the AGI.

     The only real advancements in the QC field have been repurposing the QC machinery to do simulations of physical processes well suited to the QC model. Not anything resembling solving classical computing problems efficiently.

   • Embed this notice
     lj·rk (ljrk@todon.eu)'s status on Tuesday, 14-Jan-2025 19:36:44 JST lj·rk

     in reply to

     • Rich Felker
     • sammi
     • Joe Lanman
     • 🌱@ambiguous_yelp:ahimsa.chat

     @dalias @khm @ambiguous_yelp @sammi @joelanman I disagree, we had quite some development in the recent years and while this is no guarantee, a lot more progress has been made than earlier. In other terms, it's not the same velocity of development but has accelerated.

     But: There are still real problems to be solved and at any given point we may run into a wall. And one real problem could also be that we have WW3 before any QC get's successfully constructed, whelp.

     But regardless, that's really no helpful debate and I'd rather not bet on that for some stuff and rather be safe (hybrid schemes) than sorry.

   • Embed this notice
     Rich Felker (dalias@hachyderm.io)'s status on Tuesday, 14-Jan-2025 22:30:29 JST Rich Felker

     in reply to

     • sammi
     • Joe Lanman
     • 🌱@ambiguous_yelp:ahimsa.chat

     @ljrk @khm @ambiguous_yelp @sammi @joelanman I'm not denying real development on the hw level. I'm saying that development is interesting but unrelated to any path to breaking classical cryptography.

     Same way I'm not denying there's been development in LLM field, but am absolutely denying it has anything to do with a path to AGI.

   • Embed this notice
     lj·rk (ljrk@todon.eu)'s status on Tuesday, 14-Jan-2025 22:30:31 JST lj·rk

     in reply to

     • Rich Felker
     • sammi
     • Joe Lanman
     • 🌱@ambiguous_yelp:ahimsa.chat

     @dalias @khm @ambiguous_yelp @sammi @joelanman I'm not saying that there's a lot of marketing bs, you don't need to repeat that in front of me.

     But there /has/ been quite some development in QC on a hardware level, denying that is seriously denying reality. Even more, there /has/ been increased velocity in the development.

     The only question is whether you consider this "relevant" or "much" or not. And while I don't think it's anywhere near to be done, I find the development worrying enough to think about this scenario in *some very select* contexts. Not as a threat now, but as something to consider and better protect against rather than not doing anything

     Like we should've done something against the climate crisis in the 1920 already, even though the impact was quite far off, but signs where there. It's quantum resilience not "the big scary quantum is coming".

   • Embed this notice
     Rich Felker (dalias@hachyderm.io)'s status on Wednesday, 15-Jan-2025 01:25:25 JST Rich Felker

     in reply to

     • sammi
     • Joe Lanman
     • 🌱@ambiguous_yelp:ahimsa.chat

     @ambiguous_yelp @ljrk @khm @sammi @joelanman No, that's a misconception based on the mathematical fantasy of infinite precision measurement. There are claimed ways you can bypass that using large numbers of physical qubits to implement logical qubits, but there's a lot of handwaving involved, and the scale you need to do anything useful, even if it does work, becomes astronomical.

   • Embed this notice
     🌱@ambiguous_yelp:ahimsa.chat (ambiguous_yelp@social.coop)'s status on Wednesday, 15-Jan-2025 01:25:26 JST 🌱@ambiguous_yelp:ahimsa.chat

     in reply to

     • Rich Felker
     • sammi
     • Joe Lanman

     @dalias @ljrk @khm @sammi @joelanman But we already know mathematically how to use QCs to break classical encryption through factorisation, its just a matter of scale, you just do the same algorithm but with more qubits to factor larger primes, and since qubits are increasing exponentially its not a stretch to say we will pass that threshold evenutally unless some major obstacle is reached in QC development