Hacker News Reader

Comment by ckbishop | Hacker News Reader

RSA in trouble when?

When a company that makes quantum computers is on this page: https://en.wikipedia.org/wiki/RSA_Factoring_Challenge

1 qubit prototype can crack RSA? 1million scaled out qubits is still talk

Not even 1 qubit, just "substantial progress towards the realization of a topological qubit" (from the accompanying Nature paper).

Please someone give input on this. It's extremely important and worrying.

If this is genuinely worrying to you, take some solace in that post-quantum alternatives are undergoing standardization and implementation right now (Signal and iMessage, for example, have already deployed some PQC, as have others).

However, this announcement is a nothing-burger. As I mentioned down-thread, you should view any QC announcement/press-release with extreme skepticism unless it includes replicable (read: open-source targeting hardware other researchers can test on) benchmarks for progress on real-world use-cases (e.g., Shor, Grover, or a newly-identified actually-interesting use-case). OP does not. Nothing to see here.

Worth saying, I am not a cryptographer—I do cryptography-adjacent research engineering. However, given the level of hype going around this industry, I think it's fair to at least expect to see the spec-sheet as it were.

All the best,

Would it be smartest for one to sell crypto right now while normies are still oblivious of what's about to happen?

No. Crypto will be safe against quantum computers.

Can you expand on this? I find this topic difficult to find solid information on, for some reason.

I'm not sure about other cryptocoins, but Bitcoin does not use encryption, it only uses authentication, which requires a DSA (Digital Signature Algorithms). Bitcoin's current DSA would in fact be broken by a cryptographically-relevant quantum computer (CRQC). However, there are DSAs - like Lamport signatures and Merkle signatures - known since the 1970s, whose security depends only on the existence of ANY secure hash function. There is no known way to break any widely used hash function using quantum computers. So I reckon that the only change to Bitcoin would be to swap out the current DSA for a different one.

I'm not sure about the downsides of quantum-resistant DSAs.

Chat, is this true?

If by “crypto,” the grandparent meant “cryptography,” this is not true. Most widely-deployed asymmetric/public-key primitives (e.g., RSA, elliptic curve cryptography (ECC), etc.) are quite fragile against an adversary with a cryptographically-relevant quantum computer (CRQC). To clarify how fragile, the general consensus/state-of-the-art as far as I am aware, is that Shor's algorithm (which breaks asymmetric primitives) requires about 2x the number of perfect, logical qubits as the RSA key-size (e.g., ~4000 qubits for factoring RSA 2048); however, because none of our qubit designs have a low enough error rate, you need about 1000 qubits to simulate/error-correct for a single logical qubit—so, currently, it's expected you would need around 4_000_000 physical qubits to factor RSA-2048. Post-quantum cryptography (PQC) is specifically the subset of cryptography that is designed to withstand attacks from quantum-enabled adversaries; it is still being actively designed, studied, standardized, implemented, and deployed.

If instead, the reference was to “cryptocurrencies,” most cryptocurrencies I am aware of depend on non-PQ constructions, and fall into the same buckets as RSA and ECC. Some systems, like Bitcoin, are in significant danger without large overhauls—if a practical CRQC is actually realized. There are efforts underway throughout the cryptocurrency communities to try to prepare for such an eventuality, but to my knowledge, none of them have major adoption yet.

As a final note on investment advice: I don't give out investment advice. :)

All the best,

As I've mentioned to another commenter, Bitcoin relies only on the existence of an arbitrary DSA. Quantum computing-resistant DSAs have been known since the 1970s. I reckon that swapping out Bitcoin's current DSA with a quantum-resistant one would not count as a major overhaul. https://news.ycombinator.com/item?id=43113682

It would probably require a “hard fork,” which is generally considered to be a major change in the Bitcoin world.

All the best,

I am curious as to whether this update would need a hard fork or soft fork. Soft = backwards compatible, meaning nodes on the old code still talk to the network as before, and new nodes have extra features (I think?). This update is adding a layer of complexity to the DSA but not adding a new feature (soft fork) but requiring that the updated DSA be used (hard fork). Maybe allow either to be used in the meantime and then if there is an exploit in the future, we're already half way to consensus on the new fork?

Thank you for taking the time to respond. I personally lend at least some degree of credence to their claim, given that this is Microsoft we're talking about and not some startup.

If their claim is true, then would that present an issue to RSA encryption? I find it difficult to find information on this topic that is digestible to a layman.

My understanding is that the benefit of quantum computing is parallelism, and I'm not sure how today's encryption standards would be safe from brute force attacks.

No. If their claim is true, they have a new prototype of a single qubit that they say could enable faster scaling up of qubit arrays (which means asymmetric/public-key cryptosystems like RSA will be in trouble sooner than we thought they might be). However, this work does not demonstrate that scaling potential at all. In the spirit of Betteridge's Law of headlines, if such a thing were easy for them to demonstrate, why would they announce this now, with a single logical qubit, rather than when they've demonstrated at least some scaling potential?

This understanding of QC is common, but isn't quite right. Quantum computation is actually really hard to parallelize (which is why Grover, though a bit frightening since it halves the security of symmetric primitives, is actually kind of damning for QC—because you can't parallelize that search really at all, so halving is the best a quantum adversary can get against things like AES-256).

I stand by my assertion that, until a QC announcement includes replicable benchmarks on actual use-cases, such things can be safely dismissed.

If you continue to be concerned (not necessarily unhealthy), engage cryptographers and security engineers to help your projects build know-how on hybrid (in this case, classical/PQ) cryptosystems, and get them deployed sooner rather than later.

All the best,

Reply on news.ycombinator.com