Hacker News Reader

Comment by nharada | Hacker News Reader

Is there a generally accepted crossover point where quantum computers will start outperforming classical computers on the same tasks? I was under the impression it was somewhere around 100 qubits, but I'm sure there are caveats around noise and stability as well, not just number of qubits.

View on HN

Here's why IBM says quantum supremacy is not relevant

https://www.quantamagazine.org/quantum-supremacy-is-coming-h...

From the article

Quantum supremacy, we don’t use [the term] at all, said Robert Sutor, the executive in charge of

IBM’s quantum computing strategy. We don’t care about it at all.

The article also mentions practical considerations such as "quantum advantage", fault tolerance, and other initiatives competing with IbM.

Relevant

https://wikipedia.org/wiki/Quantum_supremacy

https://medium.com/@jackkrupansky/what-is-quantum-advantage-...

The only real usefulness of quantum computers would be the supposed quantum supremacy. IBM not caring about Q supremacy is pure nonsense.

>would be the supposed quantum supremacy

True. Another way he could have responded is: We don't care if quantum computers actually do anything useful, only that we can use the PR to get positive stock price movement.

They could also say "We are scientists doing cool stuff, discovering things, and having fun. Maybe it will be useful someday but we aren't really focusing on that."

Why are you attributing some kind of cynical motive?

Because it is a job, not a hobby, and is essentially about money. It's not cynical to ascribe venal motivations to employment, and to imply it is is a cheap shot.

It's foundational research, not product development. IBM does a lot of research that doesn't have immediate practical applications.

Scientists having fun is nice. But the money could be better allocated at critically useful research areas

Yikes, tough crowd.

Quantum computers are a kind of "because it's hard" problem. Some useful discoveries might come out of it by accident even if quantum computers never become practical.

And, uh, it's not your money, nor is it mine, so why complain? In the grand universe of wasted resources, quantum computers are probably not where I'd begin making cuts.

Lots of very practical & critically useful things can come out the foundational science & engineering required for this sort of research. Just look at the practical applications of various technologies developed to get the Large Hadron Collider up and running. [0]

[0] https://kt.cern/cern-technologies-society

Nice try except quantum computing is not a fundamental science

It's hard to get more fundamental than studying how quantum properties & effects can be manipulated. But call it what you will: You avoided the issue. Whether or not you use the label "fundamental", the research can absolutely yield practical & critical advancements. Plenty of necessary aspects & engineering of the LHC would be unlikely to fall under your definition of "fundamental", but nonetheless have are part of the benefits realized from those efforts.

After 40 years of applied quantum computing we have made no useful discoveries. No surprise since it's applied. 40 years of LHG has yielded so many useful discoveries. If we reallocated quantum computing funding to LHC wouldn't that be more rational? And there are far more useful things to fund than the already useful LHC.

A solid understanding of quantum physics, helped along by research with quantum computing, is part of the foundation for much modern high technology. Current quantum computing research is pioneering better techniques for noise isolation and error correction, which has excellent practical potential as well.

My point with the LHC isn't just that it has produced useful things, but that we didn't know exactly that it would do so when it all started. Planning & construction on projects that would eventually form part of the LHC began as early as the late 70's, early 80's, roughly 40 years ago. That too was based on research & engineering that started even earlier. So if we're comparing the LHC and quantum computing, the tech behind the LHC has had more time to be developed and refined. When quantum computing gets to the same level of maturity, it may have just as many positive effects. Even so, arguing about time lines is probably besides the point: different research can take longer or shorter to have a beneficial impact.

Maybe there are other things that would benefit more from the same resources, but it's also not something we can know in advance. Plenty of early promising research is chased for years with no result. Some research takes decades for its significance and practical impacts to be felt. We have to pursue a lot of things or we'll never get much of anywhere.

I'm also curious though: The argument that quantum computing isn't doing enough, fast enough, has some merit. It's not quite zero sum, but there is a limited amount of money to go around, and maybe it could be better spent. What would be your pet project to advance in place of quantum computing? Personally, I'd want greater focus on energy research. I'm not looking to criticize your choice, I'm genuinely curious.

Like what, better ad targeting for social media?

Like fighting ageing or ignorance.

There are reasons to believe quantum computers will be really great at optimization problems we would encounter in simulating the molecular machinery of cells.

Ah ? Interesting, a source would be welcome.

DWave builds machines they say are great for that kind of problem. I would assume, since they are not general purpose quantum computers, this represents a subset of what general purpose quantum machines would be good at.

https://arxiv.org/abs/0908.1921 looks like one such application.

Quantum supremacy is usually misunderstood, and it's likely quantum computers will not surpass transistor computers in every way; they will be much better at a subset of tasks only.

Right. And the "quantum computer" won't ever really exist anyway: we'll just end up with quantum functional units on regular computers. We attach GPUs and TPUs to regular CPUs, but we don't call the resulting machines "graphics computers" or "tensor computers".

Actually, the term "graphics workstation" isn't uncommon, and "gaming rig" is a very common term for machines with great big GPUs attached.

This threshold is commonly referred to as quantum supremacy. It could be that IBM just passed that threshold, but I'm by no means an expert on this.

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

> It could be that IBM just passed that threshold

I don't think they have just yet. There is a simulated QC on conventional hardware with around 50 qbits. We are close, though, it might still happen this year.

But even with 50 qubits, the amount of time per operation—if simulated classically—is at least a million+ times slower than what it would be for a quantum computer. This is like comparing disk to L1 cache. Does that count for supremacy? Mathematically, no. Does it count for some kind of advantage? Yes.

That point in time is, "tomorrow".

It should be at least more like 10000 according to this paper https://arxiv.org/abs/1805.05224 But this is in an utopian world where we neglect quantum error correction and stability and the fact that quantum computers are mostly non programmable (there's not even one qubit logic gate..)

TLDR quantum computing is going nowhere (excluding the low probability of a breakthrough in the field)

'non-programmable'... um what?

Reply on news.ycombinator.com