Hacker News Reader

Comment by sgt101 | Hacker News Reader

20 cycles... I think this kind of result demonstrates the gap between the aspirations of the QC community and the expectations of technology spectators (like investment banks) - how many cycles would be required to do "How to factor 2048 bit RSA integers in 8 hours using 20 million noisy qubits."?

View on HN

While breathless press announces are indeed often too hyped, I think we should keep in mind the history of the transistor. We went from fragile expensive single transistors to VLSI in half a century. And we consider that amazing.

That doesn't mean the same thing will happen with QC, it just means that maybe we should QC some slack, and keep cheering for the incremental improvements that arrive every couple of years?

https://en.wikipedia.org/wiki/History_of_the_transistor

Yup - agree, and also "genetics tech" 70 years from Crick and Watson to mRNA vaccines and then actual genetic engineering in adults with CRISPR [1].

This kind of progress is amazing, I think we should start the clock on QC from either Shor's paper or maybe from the first 2 bit QC in 1998 - so that takes us to say 2060?

[1] https://www.nature.com/articles/d41586-021-01776-4

There's one big issue with QC currently. There is no mathematical demonstration that QC is scalable. All the speed-ups need some form of error correction to scale the way digital computation has. All hardware advances are just increases in computational fidelity so far to allow more useful qubits. Shor has an error correction algorithm for a single qubit, and there have been improvements to that, but nothing general enough to show that QC can even scale like digital computers can.

From my understanding, the quantum threshold theorem states that if you have a low base error rate (1-3%) you can apply multiple levels of a single quantum error correcting code to get arbitrarily low levels of errors on a logical qubit ("remove errors faster than you introduce them"). Right now that means that you need approximately 1000x physical qubits for 1 logical qubit, but to me that's a solid mathematical/theoretical demonstration that you can scale QC.

You're right. I've looked at quantum codes before, but apparently my understanding was lacking - thanks for the info.

I dont think anyone other than snake-oil salesmen have seriously suggested that quantum computers will break RSA as used in real life in the near future. We're simply quite far away from that milestone.

I think that there is a bigger gap that many think between what the community sees as realistic and what is read into that by investors and other technologists. I think that many people believe that a current 50-60qubit device "runs" in the way that a valve based computer did in the 1950s - for hours before a bug got into a valve (for example), and I also don't think that people appreciate that it takes 36 hrs to turn them on and calibrate them. I believe that most technologists think that getting to 20m noisy qubits is the challenge - but the reality is that there are a plethora of challenges that must be overcome to actually move an 20m noisy qubit device to a usable 20m noisy qubit device.

Reply on news.ycombinator.com