Hacker News Reader

Comment by hannob | Hacker News Reader

I feel this starts from a very wrong assumption.

There is the idea floating around that in the future we'll just replace classical computers with quantum computers. But even if we skip the challenges of actually doing that, what would be the point?

From what we know today quantum copmuters aren't magical computers that make everything faster. Instead there are a couple of very special problems that QCs can calculate much faster.

If you keep that in mind it's clear where QCs job is: As specialized equipment doing calculations for very particular problems, e.g. for scientific use.

That also means noone will likely ever need a quantum operating system. We'll just continue using "normal" operating systems that interact with specialized hardware for specialized calculations.

View on HN

This is a bit tongue-in-cheek, but predictions are notoriously hard to do.

"I think there is a world market for maybe five computers." -- Thomas Watson, chairman of IBM, 1943.

"The Americans have need of the telephone, but we do not. We have plenty of messenger boys." -- Sir William Preece, chief engineer of the British Post Office, 1876.

"Who the hell wants to hear actors talk?" -- H. M. Warner, Warner Brothers, 1927

I don't see how pointing out incorrect predictions from the past is constructive or helps address the question. In the same line of thought you could have collected a list of correct skeptical predictions and listed a bunch of technologies that haven't worked out.

And what is the question here exactly? OC points out a couple of assumptions that are made (I agree for instance that I don’t see the QM replace classical computers anytime soon)

But then also makes a few assumptions that might prove to be true or not. That is to say : we don’t know.

The people who I quote in the past also had a delineated idea of what the future might entail, but it’s impossible to look very far ahead.

You are argumenting with a fallacy.

The OP argumentation might be incomplete, but is at least solid argument for why we won't need a quantum operating system.

If you believe there is or could be a need for a quantum operating system, _explain why_. If you don't know, that's ok.

But arguing that "The OP argument cannot be correct because other people made incorrect predictions about other stuff in the past" is pointless and useless.

Maybe some application of quantum computing appears in the future that requires a quantum operating system, or maybe there is one such application that already exists today and we don't know about. I don't know of any, and like the OP I also think of Quantum Computers as "accelerators" (like GPUs, FPGAs, Xeon Phis, ...) for certain class of problems.

I guess I am looking at it from a different angle, in which case, and following your reasoning, you are indeed correct and I misconstrued OCs intention/message.

Fundamentally, I disagree. The person you are conversing with is applying a sort of Bayesian prior to the reasoning about quantum computers. I think this is actually a very wise thing to do.

- they're specifically picking a prior made of only the skeptical predictions of famous people about some technology that turned out to be wrong.

- they aren't updating the prior with new information, they're just pointing it out as sufficient on its own

Using the two points above anyone can make a bad argument about anything - just pick the data points that fit your narrative, state that narrative as a conclusion and don't update it when presented with other facts.

> I think there is a world market for maybe five computers." -- Thomas Watson, chairman of IBM, 1943.

This quote always rubs me wrong, because the problem is that we’ve changed definition of what a computer is.

If you come to think about it, what he was referring to is what we would now call cloud infrastructure, and there’s not really that much more than 5 of those...

Yep, and that works for cloud infrastructure. But we also have mobile/IoT ...

While I'm not discounting your assumption about limited quantum use cases for now, I'm going to suggest that when we think hard about a problem - we as humans usually don't see the true potentials of a technology. We usually assume we understand, but depth of understanding doesn't come until hands on experience, depth of knowledge increase.

I'm guessing we will find significant uses cases to replace classical computers.

While not impossible, I would say that it would require a significant change in our understanding of quantum computation. That is, everything we know so far about quantum computing suggests it is only useful in certain complex use cases, and even then, with much help from a classical computer - we don't know any quantum algorithm that could be computed entirely on a quantum computer, they all have classical parts.

I imagine that a quantum computing device will eventually be similar to a GPU now - plugs in to a CPU, does what it needs to wherever its strength lay, and over the course of time we will find more and more uses for it (or entirely new fields like bitcoin).

I am fairly certain that there are already OS's designed to run entirely on a GPU rig.

> I am fairly certain that there are already OS's designed to run entirely on a GPU rig.

A "GPU OS" would still run on a CPU, it's mostly about feeding tasks to the GPU efficiently. The design goals between CPUs and GPUs are strikingly different; CPUs are much more designed for conditionals and branching while GPUs do massively parallelized number crunching. The CPU is the robot arm that picks and shovels things into the GPU pipeline.

My understanding is that quantum computers are very similar to GPUs in this regard, and I don't think a classical "OS" makes sense to directly run on either.

(P.S.: even more so, I'd expect QCs to model similar to FPGA accelerator cards)

I second this. Something like a quantum computers will, at least for most computing applications and consumer market, be something that is a peripheral to your "classical" computer just like FPGAs or GPUs today.

I would like to think that in 10 years I'll be buying a used 64-qubit card on Ebay, like a 2GB video card. What will be the equivalent to gaming which drives this tech to ever-higher performance? DIY protein folding life hacking?

Plus QPU is a nice little name we could use for it

Imo deep learning would be a better example, Bitcoin is no better because of GPUs (nor is it mined on then anymore)

This isn't an OS in the Windows sense of the term; this is LabVIEW for Quantum, and is sorely needed.

It doesn't seem to me like the point of this is to "replace" classical computers, but to consolidate the tools people need to interact with a variety of quantum computing technologies:

> We have a grand vision: an operating system that makes quantum software portable across qubit technologies; that is scalable to millions of qubits; and, that teases the highest possible performance out of every qubit – even for applications like error correction that need fast feedback loops. Deltaflow-on-ARTIQ is our first step towards achieving this vision.

OK, but let's paint a picture and see where it goes:

QC's become several orders of magnitude better at a few very specific calculations.

We work out to frame some set of algorithms that are currently slow on conventional computers so that they're fast on QC's.

Some genius works out how to frame some currently-obscure method of (for example) AI research into the new subset of QC-friendly algorithms.

Some other geniuses work out how to train this new breed of fast AIs how to interact with users in a meaningful way.

The next mobile revolution happens, everyone carries an AI in their pocket, the Culture looms nearer :)

My point being that if we make the hardware faster, we'll find an application for it. No-one could have predicted that the humble graphics card (solely intended to make the specific vector calculations of 3d gaming faster) would power a financial speculation industry worth trillions of dollars. Make it fast and some genius will work out how to make it useful.

Or a genius could discover some killer application for one of 1000 different competing technologies that are not quantum computing but replace current technology instead. This giant ramp up of something that is still long-term high-risk is kind of strange to watch. Surely all that funding would be better spent doing fundamental research on hardware, rather than developing the IT layers. I guess at least the stupid software patents will expire sooner once it finally becomes real.

I lived in University accomodation with a researcher who was working on transistors for optical computers, and predicted them taking over from silicon in "10 years, tops". That was in 1989.

The relationship between funding and hype is always weird. But it must be really hard having to decide which of several projects to fund, and "what people are talking about" is obviously one of the metrics.

Question for those with active imagination: how would quantum computing change Human Computer Interaction? If it doesn't then a kernel driver module for the quantum computing hardware chip would suffice.

Ooo fun. Okay.. really stretching here..

Lets say decoherence is no longer an issue. You've got your qbits suspended in some sort of fancy diamond lattice, and for some reason you really want your scheduler and all your hardware IO to be running via quantum algorithms. So you've got your fusion core crystal loaded into your compute crystal, and you really just want to eliminate any need for silicon at all.

So now we're talking about a self contained smart-diamond, probabally embedded into a human forehead. It pretty much sits around doing its own thing, observing the world around it and computing out all possible fates of everything it sees, and every once in a while your tiny little monkey brain can reach out and ask it dumb questions like "Hey, how much of my life savings should I dump into some random new cryptocurrency??" and it will be obliged to look into it momentarily and respond accordingly.

Isn't this the same argument people were making about those huge mainframes 50 or so years ago? If there's one thing we can all agree on is that we don't really know the implications of a new technology in the long run.

something like a Quantum Co-processor for accelerating quantum algorithms would likely be the next step, instead of standalone 'quantum-centric' processors with their specific code and protocols.

> From what we know today quantum copmuters aren't magical computers that make everything faster. Instead there are a couple of very special problems that QCs can calculate much faster.

More details: https://www.smbc-comics.com/comic/the-talk-3

Reply on news.ycombinator.com