Is Quantum Computing Impossible? (ieee.org)

"Quantum computing is complex and it's not all it's cracked up to be," writes Slashdot reader nickwinlund77, pointing to this new article from IEEE Spectrum arguing it's "not in our foreseeable future": Having spent decades conducting research in quantum and condensed-matter physics, I've developed my very pessimistic view. It's based on an understanding of the gargantuan technical challenges that would have to be overcome to ever make quantum computing work.... Experts estimate that the number of qubits needed for a useful quantum computer, one that could compete with your laptop in solving certain kinds of interesting problems, is between 1,000 and 100,000. So the number of continuous parameters describing the state of such a useful quantum computer at any given moment must be at least 2**1,000, which is to say about 10**300. That's a very big number indeed. How big? It is much, much greater than the number of subatomic particles in the observable universe. To repeat: A useful quantum computer needs to process a set of continuous parameters that is larger than the number of subatomic particles in the observable universe. At this point in a description of a possible future technology, a hardheaded engineer loses interest....

[I]t's absolutely unimaginable how to keep errors under control for the 10300 continuous parameters that must be processed by a useful quantum computer. Yet quantum-computing theorists have succeeded in convincing the general public that this is feasible.... Even without considering these impossibly large numbers, it's sobering that no one has yet figured out how to combine many physical qubits into a smaller number of logical qubits that can compute something useful. And it's not like this hasn't long been a key goal.... On the hardware front, advanced research is under way, with a 49-qubit chip (Intel), a 50-qubit chip (IBM), and a 72-qubit chip (Google) having recently been fabricated and studied. The eventual outcome of this activity is not entirely clear, especially because these companies have not revealed the details of their work...

I believe that, appearances to the contrary, the quantum computing fervor is nearing its end. That's because a few decades is the maximum lifetime of any big bubble in technology or science. After a certain period, too many unfulfilled promises have been made, and anyone who has been following the topic starts to get annoyed by further announcements of impending breakthroughs. What's more, by that time all the tenured faculty positions in the field are already occupied. The proponents have grown older and less zealous, while the younger generation seeks something completely new and more likely to succeed.
He advises quantum computing researchers to follow the advice of IBM physicist Rolf Landauer. Decades ago Landauer warned quantum computing's proponents that they needed a disclaimer in all of their publications.

"This scheme, like all other schemes for quantum computation, relies on speculative technology, does not in its current form take into account all possible sources of noise, unreliability and manufacturing error, and probably will not work."

Comments

Simple answer

[Ukab the Great]

Quantum computing is simultaneously both possible and impossible.

Re: Simple answer

[Presence Eternal]

It is good to see people thinking outside of the box.

The cat keeps distracting them.

Re:Simple answer

[msauve]

Wave if you're a particle!

Re:Simple answer

[arglebargle_xiv]

You can use it crack certain cryptography problems faster;

One in particular: That maths wonks are running out of excuses to design new algorithms. There's only so many zero-knowlege group key management IND-CCA blind signcryption schemes you can publish before people fall asleep. By coming up with this unicorn-magic break-all-existing-algorithms space-alien wish-fulfilment technology, said maths wonks get another ten to twenty years of publishing papers on algorithms resistant to unicorns, magic, sharks with lasers, and so on. That's why there's so much concern about post-unicorn cryptography... uhh, sorry, quantum, it's addressing an academic publication problem, not an actual real-world threat.

Re:Simple answer

[glenebob]

Yep. You can use it crack certain cryptography problems faster; problem though, the algorithm scales differently and doubling the key size makes it much harder to crack. Whereas, using traditional brute force on regular computers, doubling the key size only helps a little bit.

Is it opposite day? I must have missed the tweet.

I see Quantum Computing every day

[SuperKendall]

I can't be the only one here that goes to look for a bug that vanishes when I am doing any kind of problem.

Now THAT is Quantum Computing.

Re:I see Quantum Computing every day

[complete loony]

I can't be the only one here that goes to look for a bug that vanishes when I am doing any kind of problem.

Now THAT is Quantum Computing.

Nah, that's a heisenbug.

Huh?

[JaredOfEuropa]

A useful quantum computer needs to process a set of continuous parameters that is larger than the number of subatomic particles in the observable universe

I thought that the whole point of quantum computers was that there's no need to describe or process all possible states. And that the difficulty of practical quantum computers is that the qubits need to "work together": you can't just make 1 cubit, then make 1023 more and build yourself a 1024 cubit computer.

The guy obviously knows way more about quantum computers than I do. But I've never seen the difficulties of quantum computing described in this manner.

Re:Huh?

[Ramze]

I think the key problem is theory vs physical reality. In theory, if you have a set of qubits entangled with zero noise at near absolute zero, you can send a quantum program to the qubits & have them process your data without you worrying about what their individual states are & then capture their completed output.

In reality, how do you entangle enough qubits to be useful? How do you prevent noise or correct for the errors of noise? How do you ensure your qubits are properly entangled? How do you accurately send your quantum program to the qubits for processing? How do you aide in processing the qubits accurately without generating more noise? How do you extract the output without generating more noise? And ultimately, how are you going to ensure that you are entangling 10^300 attributes of your qubits perfectly in the first place, much less correcting for errors in processing them?

I think the TL,DR is that this quantum physicist sees all the places errors can creep in and how difficult it can be to correct for them. The answers he sees coming from the community seems to be to just add more qubits for error correction - or even process the same data with multiple quantum computers or with multiple paths through the same qubits.

I understand his/her frustration. It seems a difficult task to precisely manipulate qubits using modern technology, and an impossible task to know and/or set the states of everything to ultimately know for certain whether an error has been generated.

Re:Huh?

[Comrade Ogilvy]

I think his point roughly boils down to: We are counting this 1000 qubit computer being a juxtaposition of all 2^1000 states, in the first place. How do you know you have achieved that status? Are you going to measure those 2^1000 states? No, of course not.

The initial state is something that is easy to describe in the abstract on a chalkboard, but what it really means in the physical world is extremely problematic.

There is a lot of handwaving around this point from the quantum computing enthusiasts. "Oh, if we can get things 0.1% correct, we can just run the experiment many times." That would be true. If only.

My counter argument is that getting a random 2^500 states rights in a 1000 qubit machine may already be too close to impossible. You do not have time before the heat death of the universe to run your calculation 2^500 times, to compensate for this shortcoming.

Re:Huh?

[epine]

In reality, how do you entangle enough qubits to be useful? How do you prevent noise or correct for the errors of noise? How do you ensure your qubits are properly entangled? How do you accurately send your quantum program to the qubits for processing? How do you aide in processing the qubits accurately without generating more noise?

I've never believed in quantum computing, because I've never seen a lay publication that does half of these questions justice.

Under you've seen the ceiling properly described, a technology simply doesn't exist.

No one in this field ever bothers to describe the ceiling.

In CMOS, you always had "when does the transistor become too small?" Some of the early answers were wrong (100 nm was once mooted as a frightening bogie man), but at least you would read sensible speculation.

At what point, in a practical sense, does the quantum entanglistor become inseparable from local environmental noise?

Silence. Crickets. Crickets on top of crickets. Crickets inside of crickets. Crickets alive and dead at the same time. All kinds of crickets. But never any sensible speculation.

Re:Huh?

[Ramze]

I tend to agree, and apparently so do IBM, Google, et al. Still, the larger the system, the more error prone it becomes. Obviously, we have quantum computers (or at least functioning parts of ones) working today and can entangle up to 50 qubits or more with relative stability... but, the question is whether we can do it at the scale needed to be "useful" (according to this individual) without losing the signal for all the noise.

This person's perspective is that what we naively see as an engineering problem to be resolved with future refinements is actually an issue that can't be resolved because nature at a fundamental particle physics level can't be controlled or tuned to the degree necessary to get one working, nor reasonably checked for accuracy because the states to be checked are beyond astronomical.

Re:Huh?

[sjames]

Hot fusion is also "just an engineering problem".

Re:Huh?

[Anonymous Coward]

Hot fusion is also "just an engineering problem".

Using the word "also" makes it look like you are grouping fusion and quantum computing into the same level of possible, which is both not true and possibly showing a deep misunderstanding of the phrase "just an engineering problem"

It comes down to how different people use the word "impossible"

To some, impossible means the laws of physics explicitly do not allow it.
To others, impossible means the laws of physics may not yet exclude it but there are no examples to demonstrate it could happen.

"Just an engineering problem" was coined specifically for people who use the word "impossible" to describe something that is currently happening on a massive scale and clearly possible, but human beings can't do it.

Yes, fusion is "just an engineering problem" because it is happening, with every star including our sun, and has been occurring for billions of years.
That fact alone demonstrates that it is in fact possible. It is happening. Claims it is impossible to occur are just outright false.

Quantum computing however isn't the same thing. There are no examples of it happening to point to and prove its possibility. All we have is that the laws of physics don't seem to exclude it as an option.
That is NOT what "just an engineering problem" applies to.

"Engineering" is taking a process and making it happen.
That process needs to be defined first, and if it isn't, means there is *nothing* an engineer can do about it.

From a physics perspective, fusion is pretty simple. Apply enough pressure and force to atoms so their electron shells overlap.
From an engineering perspective, that may be straight forward but is FAR from simple to accomplish.

Quantum computing on the other hand is not simple. We have nothing in nature doing that to follow. We have guesses and assumptions many of which haven't been shown as fact. There is damn little to engineer about it all.

The phrase pretty much only applies when people claim as impossible things that are already occurring and clearly possible.
Anything not demonstrated as possible isn't in the realm of engineering because it very well may not be possible until that proof is shown.

Not impossible... just even harder to exploit

[igor.sfiligoi]

The author makes a great point about the near impossibility of perfect, error-free quantum computation.
But this has been realized a few years back by most quantum algorithm developers, too.

Many recent algorithms assume that the quantum computation will be partially faulty.
And they work around it.

Yes, that makes these algorithms harder to design and they are less efficient compared to the ones assuming no errors, but they still seem to provide a way forward.
I would definitely not write off quantum computing yet.

Makes no sense

[cryptizard]

So the number of continuous parameters describing the state of such a useful quantum computer at any given moment must be at least 2**1,000, which is to say about 10**300. That's a very big number indeed. How big? It is much, much greater than the number of subatomic particles in the observable universe.

I am struggling to come up with some way that this part makes any sense at all. It sounds like the kind of thing someone who is definitely not an expert the area would say. He is expressing the number of possible configurations of 1,000 qubits but that is only something you care about if you are simulating a quantum computer with a classical one. The whole point of quantum computers is that you don't have to do that.

Also a simple counterexample to this sentiment is given later on, when mentioning that Google already has a 72-qubit computer. Just storing the states of a 72-qubit machine would be substantially more than the entire capacity of the internet, implying that since we somehow did it then enumerating all the states is not necessary.

If an elderly but distinguished scientist says...

[g.random]

If an elderly but distinguished scientist says that something is possible, he is almost certainly right; but if he says that it is impossible, he is very probably wrong. -- Arthus C. Clarke

Reductio and absurdum

[sgunhouse]

By his logic... my very first computer was an RCA VIP, it came with a whopping 2K of RAM. That's a measly 16384 bits - not counting internal registers, flags, etc.So to actually model all the possible internal states of just the RAM is 2^16384 which is roughly 10^500. I'm sure you know how the rest of the argument goes.

A thousand qubits is simply 1000 mutually interacting particles. You're not trying to represent every possible state (and as the possible states are infinite, you couldn't). His argument is complete nonsense and tells you nothing at all.

Re:Pretty much my take also

[ceoyoyo]

That's not quite what quantum supremacy means. It's not their quantum computer doing a computation faster than a conventional computer. That would be a very slippery benchmark. First question... what conventional computer?

Quantum supremacy means demonstrating that your quantum computer can complete certain computations with less computational complexity than a classical computer. In the typical examples, the classical complexity is exponential and the quantum complexity is theorized to be subexponential.

The quantum computer may very well take much longer than the classical computer in clock time, and any problem Google solves with a 72 qubit chip is going to be a toy example.

Let the past be your guide

[Tablizer]

Quantum physics is always teasing us with almosts: almost instantaneous communication, almost energy out of nowhere, almost backward time travel, etc.

After all these teases, I'd bet on quantum computing having an inherent flaw nobody has discovered yet.

Schrodinger Lucy is holding the football again...

Unrealistic expectations

[rkordmaa]

First of all, a quantum computer is not a regular computer with added magical pixie dust, it's not a "better" computer, it's a very different type of computer. Generally much more limited computer at that, but it can solve a certain subset of problems that a conventional computer practically cant. All a quantum computer needs to do in order to be a roaring success is to solve one such impossible problem. I suspect we are pretty close to that.

Quantum computer is more like a test tube than a computer. In the sense that the best way to find out how a chemical reaction will run is to do it in a test tube, instead of trying to simulate in on a classical computer. Quantum computer is just more generic than that and you can reduce wider range of problems down to quantum algorithms.

Core arguments of the article

[gotan]

To my understanding these are the core arguments of the article:

1) The feasibility of quantum computing is based on the assumption, that the effort (e.g. for error correction) scales with the number of qbits (in the example 1000), not the dimension of the superimposable state vector (2^1000). According to the author it is not yet proven that that is the case.

2) For a useful quantum computer it must be possible to manipulate qbits (with quantum gates) at will, i.e. move them around and "process" them like we do with classical bits in a classical computer nowadays.

3) In theoretical concepts of quantum computers perfect quantum gates are assumed, but quantum gates are physical devices. Rotating a spin by 90 deg might be achieved by applying a magnetic field of a given strength for a precise length of time. But in the physical world the precision of such manipulations is always finite, so maybe the result is somewhere between an 89 and 91 deg rotation and the axis might be slightly off too. Such imprecision might even occur when storing or transferring qbits (the information) in/between their physical storage. In lengthier calculations such errors add up, a bit like in analog computers. That would (severely?) limit the usefulness of quantum computers.

This is very unlike classical logical gates where anything above a certain voltage is interpreted as "1", anything below as "0" and logical gates consist of voltage controlled switches (transistors) in either "on" or "off" state that is clearly defined and leaves a wide error margin in terms of voltage.

To summarize: The physical world is far messier than the theoretical concepts of quantum computing and it has yet to be shown, that error correction mechanisms to control that "messiness" are feasible.

These problems are not new, and AFAIK there are theoretical as well as experimental efforts made to counter them. The article presents a very disillusioned view of the advances in that respect and suggests that it might be even impossible to overcome the problems. Sadly, instead of making the points by giving examples of the efforts and the advances or non-advances that were made, a lot of space in the article is simply wasted by pointless comparisons of the number of superimposable quantum states to the number of particles in the universe and the like. The question is not how big that number is but if it really represents the size of the obstacle/necessary effort on the way to quantum computing.

OTOH it should be noted, that even the theoretical concepts of quantum computing, i.e. quantum information theory, broadened our understanding of quantum mechanics. E.g. experiments on entangled states like EPR, delayed quantum eraser or "quantum teleportation" (which should really be named "quantum state teleportation") can be viewed from a new perspective.