My Quantum Computing Swan Song

Over the past couple of years my thoughts about quantum computing have occasionally drifted towards a key observation about my early experiences with classical computing, specifically that I knew very little about the details of transistors, their electrical engineering and their physics, but I did know that they enabled the computers that I used — without any knowledge of the details of those transistors. A very simple, but very powerful observation.

I did know a little about logic gates, AND, OR, NOT, NOR, NAND, and flip flops, but I didn’t even need to know that to use classical computers — to their fullest.

Logic, yes, but in a mathematical sense, not an electrical engineering sense. Sure, later I did learn a lot more about the electrical engineering, digital electronics, and computer engineering, and even a little of the physics, but none of that was needed to use and exploit classical computers, for software development, or applications of classical computers.

All that I or any software developer needed was the programming model, or for even low-level assembly language work, the computer architecture, registers, addressing, data formats, and the instruction set architecture, even down to bits (and bytes and words), but as abstract, logical information, not the electrical engineering or physics of gates and transistors.

The programming model. That’s all that matters, really. A halfway decent programming model.

That vision, that experience, has stood the test of time, all these decades, unshakable. A true Rock of Gibraltar.

We need to get to a similar state of affairs for quantum computing.

Without that, quantum computing isn’t particularly interesting to me.

Now, my personal vision for quantum computing, as a goal, is to get to that same exact moment, just for quantum computing and “qubits”. I put “qubits” in quotes, because we desperately need a metaphor for quantum state, in terms closer to the basis states and bit strings for entangled basis states. That is all any software developer should have to care about, not the current metaphor of qubits which are really devices, closer to transistors or flip flops than to the logical information bits of classical computing.

Yes, we do need a clear notion of quantum information, but at an abstracted, logical level, and envisioned as an extension to classical bits, not the level that physicists prefer to work at.

Programmers of quantum computers and designers of quantum algorithms should be able to work with high-level, abstract quantum information, not raw qubits, which are more analogous to the transistors (or vacuum tubes!) and flip flops of classical computers.

To be clear, I am not suggesting that software developers should be able to transfer their existing code and knowledge directly over to a quantum computer as is. We’re not seeking to cater to people who say things like “Okay, here’s the way I do things on a classical computer, show me how to do the same things on a quantum computer.”

Bringing a classical computing mindset over to quantum computing is a non-starter. Just don’t do it!

We do indeed need a quantum-native programming model for quantum computers. And it’s okay if it looks radically different than classical programming models. Just as long as it is easy to learn, easy for classical software developers to adapt to, and quickly results in software developers feeling both confident and comfortable, and productive, and reliably delivering dramatic quantum advantage with the new, quantum-native programming model.

Quantum-inspired algorithms? That’s fine, well and good, sometimes, but that’s about solutions which still run on classical computers, not about solutions that exploit native quantum computer hardware, particularly quantum parallelism.

What the programming model for quantum computers should look like is anybody’s guess. Maybe the only thing I know for sure is that it should be quantum-native, tailored to the capabilities of quantum computers, particularly quantum parallelism, and not merely trying to mimic or simulate the programming model of classical digital computers lock, stock, and barrel. What should replace raw qubits and the current gate model is unclear, but it needs to be something, anything that is an abstraction that people can relate to, not physicists. Not something analogous to the transistor and electrical engineering gate-level of classical digital computers.

All I ask for is a halfway decent programming model for quantum computing that offers that same level of satisfaction that I and everyone else experienced with even the simplest and primitive of early classical digital computers.

I’m not as interested in simulation as I am in analytical computation. In classical digital computing there’s not real distinction, but quantum computers whose qubits have probability amplitudes can be used for simulation as one would use an operational amplifier in a classical analog computer, where each qubit is analogous to an operational amplifier, rather than a single binary bit of information. Qubits can be used to model individual physical systems, such as atoms or electrons, with evolution such as Hamiltonian evolution. But, this quantum-style of simulation has more in common with classical analog computers than with analytical computation, where registers of qubits represent numbers (the bits of numbers), coupled with quantum parallelism to evaluate an entire solution space at one time, rather than the looping and iterating of a classical digital computer.

This distinction between simulation and analytical computation is interesting, but not as much to me. I’m focused more intensely on analytical computation. The question is how to arrive at a decent programming model for quantum computers which is optimal for analytical computation rather than focused so heavily on the needs of analog-style simulation, as current quantum computers are.

It may well be that simulation and computation on quantum computers needs two radically-distinct programming models. It may indeed, but I don’t want to prejudge that at this juncture. In fact, we may not know the answer until we fully grasp and fully elaborate and fully distinguish the divergent needs of simulation and analytical computation.

Classically, we can simulate just about anything on a classical digital computer — including quantum computers themselves (up to about 50 qubits), that’s how powerful and easy to use the programming model is for classical digital computers, functionally, although it does have its limits of capacity and performance. But I wouldn’t want to preordain that the programming model for analytical calculation on a quantum computer must be saddled with that same requirement. It wasn’t a requirement on classical computers, it just happened to work out well given the fantastic flexibility of the programming model.

Sure, we do want to replicate a lot of the flexibility of the classical programming model, but that is not an absolute slam-dunk by any means. We’ll just have to see how it plays out.

Eventually, it would be great to have a truly high-level programming language for analytical quantum computing, much as FORTRAN was for classical computing in the 1950’s, but even before then, we need a great programming model at the machine language level, comparable in power and ease of use to the assembly language(s) of classical computing, which packed a lot of punch and were relatively easy to use, in contrast to current quantum programming models.

All I ask for is a halfway decent programming model for analytical computing on a quantum computer that offers that same level of satisfaction that I and everyone experienced with even the simplest and primitive of early classical digital computers. That plus quantum parallelism to gain a dramatic quantum advantage over our beloved classical computers.

By all means, let the physicists and chemists have their simulation capabilities, just don’t treat the users and their needs for analytical computing as second class citizens, as they are being treated at present.

That’s it, that’s my personal vision of what quantum computing should be. Anything less (or more, loaded with the jargon of physics) is a sham, a horrific abomination.

Besides the inadequacy of the quantum programming model I have deep concerns about the upper limit for entanglement of qubits, the upper limit for simultaneous quantum states for deeply-entangled qubits (2^n quantum states for n qubits), and the lower limit for granularity or discrimination of probability amplitude for those 2^n quantum states and fine-grained calculations based on probability amplitudes for those 2^n quantum states, such as for quantum Fourier transforms (Shor’s factoring algorithm being the worst case so far, and farthest from fruition.)

I wrote something about this looming issue almost five years ago:

• Beware of Quantum Algorithms Dependent on Fine Granularity of Phase
• https://jackkrupansky.medium.com/beware-of-quantum-algorithms-dependent-on-fine-granularity-of-phase-525bde2642d8

That said, I do still have hope that quantum algorithms and circuits using 48 to 64 qubits might still be technically and theoretically feasible, with 2⁴⁸ to 2⁶⁴ quantum states, even if there is no credible hope for hundreds, thousands, millions, billions, or trillions of deeply entangled qubits with extremely fine granularity of probability amplitude. It is unclear, and there are no experimental or theoretical results that ascertain either the upper limit for deep entanglement or the lower limit for reliable granularity of probability amplitude. I mean, at some level, basic quantum uncertainty will swamp and dwarf any finer granularity. Everybody else seems oblivious to this prospect. Oh well!!

If I sound too negative, here are a pair of proposals that I have written with a more positive attitude, that I think have at least a remote possibility of being practical, and in the relatively near term:

• 48 Fully-connected Near-perfect Qubits As the Sweet Spot Goal for Near-term Quantum Computing
• https://jackkrupansky.medium.com/48-fully-connected-near-perfect-qubits-as-the-sweet-spot-goal-for-near-term-quantum-computing-7d29e330f625

And

• Modest Proposal for a Quantum Minicomputer Based on a 4x4x4 Cube of 64 Photonically-connected Near-perfect Discrete Qubits with Full Qubit Connectivity
• https://jackkrupansky.medium.com/modest-proposal-for-a-quantum-minicomputer-based-on-a-4x4x4-cube-of-64-photonically-connected-4bda9a382734

Alas, the quantum computing community is too far gone to pay attention to or express any interest in any such modest and practical proposals. Sigh! Oh well!!

With that, by all rights, I should just draw a line in the sand, and end my personal efforts to get deeper into the current vision — and reality — of quantum computing, and endeavor to steadfastly refrain from getting anywhere near or even thinking about quantum computing until the purveyors of quantum computing come to their senses and come up with a more sensible programming model for quantum computing, or maybe simply step aside and let experienced, and wiser, professional computer engineers take the reins and re-envision a more appropriate programming model for quantum computing.

I need to see compelling evidence that people are seriously moving towards such a better programming model, even marching, and with a compelling degree of alacrity. Much research is desperately needed on this front. It appears to be too much to ask for to expect physicists to take any leadership on this front.

Indeed, let the physicists focus on fixing the damn qubits.

And then have them work with professional computer engineers to design and engineer full qubit connectivity.

We need a simple and powerful programming model, high fidelity qubits, and full qubit connectivity, and then we wouldn’t need such complex overtooling and complex compilers/transpilers and complex error correction schemes as we see today that are soaking up so much of our scarce talent. Get the basics right at the lowest levels and we won’t need such horrific complexity at the higher levels, or at any level.

You might ask why I even bother paying any attention to quantum computing if it’s in such dire straits. It’s a fair and great question. Well, it looks like an impending trainwreck to me, and maybe I figure that something, anything, I might say might avert or mitigate that wreck. And, besides, it’s enlightening to accurately chronicle the chronology of trainwrecks. You know, help people learn from their mistakes.

On a lot of days I feel as if I’m watching a trainwreck play out in slow motion, or at least the train moving towards a situation which can’t end any other way than badly, very badly.

Although on some days it just feels as if the entire quantum computing sector is simply bogged down in a swamp (e.g., quantum error correction), spinning its wheels, going nowhere.

And other days it feels like everyone in quantum computing is living in one of those cities floating in the clouds, oblivious to the real world below.

It’s too easy to fantasize that some day, maybe not soon, people will come to their senses and get real about building a usable quantum computer.

The bottom line is that people, researchers, vendors, are simply not moving fast enough in the direction of a computing environment which is as intellectually satisfying and productive as what we’ve had with classical computing since the advent of programmable electronic digital computers back in the 1940’s. Fast enough? I should say that they’re not moving in that direction at all, at present. A thousand times faster would still be zero, no movement, no progress.

To get back to reality…

I am retired, but still retain a significant interest in quantum computing.

But being retired, I don’t have the near-endless extra time of the youth to wait lots of years for vendors and researchers to come to their senses.

This Swan Song is not a final abandonment of any interest in quantum computing, but just clearly circumscribing why I really do think that the sector has reached a stage where there is alternatively a barren desert, dense jungle, bottomless abyss, or intimidating ocean before it, with no credible plan, vision, or even hints about how to cross it.

Now, let’s see what comes next!

For now and the near-term and foreseeable future, I have to ask…

Is quantum computing dead?

True, analytical computation that is, not analog-style simulation.

Besides a Monty Python quip of “It’s not dead yet!”, the answer is basically a resounding… yes, quantum computing is dead, effectively. Sure, there’s plenty of going through the motions, lots of activity — twitching, but without any truly inspiring progress bringing a useful, production-scale quantum computer palpably close to fruition.

Brain-dead. Or maybe in a death-like zombie mode, on autopilot, moving mindlessly without going anywhere or getting anywhere.

Existing at an animal-like subsistence level without any human, intelligent, intellectual drive.

Kind of a Frankenstein-like “monster”, with all the parts there (well… sort of), but lacking animation in any useful sense.

Have any truly new (and exciting) ideas about quantum computing surfaced in the last few to five years?

IBM introduced the 27-qubit Falcon in 2019. Sure, we have many more qubits now, but we NEVER achieved full utility for even 27 qubits — Quantum Volume of 2²⁷ (134,217,728.) Granted, Quantinuum has achieved QV 2²³ (8,388,608.)

IBM itself has abandoned the Quantum Volume metric, never getting past 512, replacing it with metrics like EPLG and CLOPS, which… NOBODY understands or knows how to relate to practical real-world applications! Tower of Babel, anyone?!!

IBM went from 27 qubits to 65, 127, 433, and finally to Condor with 1,121 qubits. With Osprey delivered five months late and then withdrawn after only four months, and Condor only briefly running in the lab and its chip shown on stage, but never available to ANYONE for actual use.

Since then, IBM fell back to 133 qubits for Heron. A revised Heron has 156 qubits, although the upcoming Nighthawk will have only 120 qubits. Not a very inspiring progression.

Try finding ANYBODY who can recite what the EPLG or CLOPS are for ANY of these systems!

Won’t quantum error correction (QEC) and fault-tolerance save us? Nope. As I’ve written, I see that QEC will ultimately fail. Sure, it may indeed work in a fashion, but it is too horribly complex and complicated to be practical. It will be like a big and bloated airplane that might look fine and taxi around fine, but is too heavy to get off the ground. And it won’t be able to handle the errors that will matter most, fine-angle rotations.

Diving down into the weeds, just a little, there are three main technical stumbling blocks (among others!) which completely sap any enthusiasm I might have for the potential for quantum computing:

1. Inability to reach and zoom past three nines of qubit and gate fidelity. Some research efforts have touched (barely!) three nines in some niche cases, but my estimate is that we need four to five nines for any serious quantum computing, even for fairly low-end applications. Vendors simply aren’t putting in the effort or even expressing serious commitment or even any interest in this essential aspect of quantum computing. 3.75, 3.5, and maybe even just 3.25 nines might enable some interesting applications, but… nobody is getting anywhere near that… and not even seriously trying to!
2. Lack of full any-to-any (all-to-all) connectivity between qubits. Some modalities have achieved this but most have not. Worse, most vendors simply aren’t putting in the effort or even expressing serious commitment or even any interest in this aspect. Zero interest. Zero. Z-E-R-O!!
3. Extreme uncertainty as to how fine-grained probability amplitude and phase can be. AKA fine-angle rotations. The math of real numbers in the range of 0.0 to 1.0 is fine, even for infinite precision; real phenomena with that precision? Not so much. Dozens or hundreds of gradations may be a slum dunk, but beyond that? Thousands? Millions? Billions and beyond? Completely unknown at this stage. With no theory, analysis, or experimental evidence to base any assumptions on. Zero. Zip. Nada. Zilch. Metrics, benchmarks, and disclosure? Again… nothing! Advanced applications, such as using wide quantum Fourier transforms (QFT) and quantum phase estimation (QPE), will require very fine-grained or even extremely fine-grained probability amplitude and phase, but there is essentially no serious discussion of this essential aspect of quantum computing. Vendors and academic researchers alike simply aren’t putting in the effort or even expressing serious commitment or even any interest in this aspect. Zero interest. Zero. Z-E-R-O!!

The extreme lack of progress, commitment, or even interest in these three critical stumbling blocks by vendors and academic researchers alike make it very easy for me to shake my head, sigh, and throw up my hands in despair… and walk away.

Maybe the only thing keeping my attention is that I am a student of phenomena, any complex phenomena, and watching a trainwreck unfold quickly or even very slowly is worthy of study, even if simply to analyze it for lessons to be learned to avoid such phenomenal and colossal failures in future endeavors.

What might it take to bring quantum computing back to life? None of the above! We need a true breakthrough, that is head and shoulders categorically distinct from what we currently have. Or as they would say on Monty Python, and now for something completely different!

Alternatively, it may not be that the raw technology per se is the issue, but that the science and technology are overwhelmed by inept, incompetent, and uninspiring management.

Maybe we should just relegate quantum computing to being… a conspiracy theory. That would certainly keep it alive and funding flowing in!

This is starting to feel like the fable of the Emperor’s New Clothes, where I can clearly see that there is nothing there, but everybody else is confidently proclaiming that what they (think they) can see is extravagant, wonderful, and of great value.

Although quantum computing as we know it may be dead, a dead end, maybe this is just Quantum Computing 1.0, and maybe in the years and decades ahead quantum computing will be reincarnated as a Quantum Computing 2.0 which is radically different from Quantum Computing 1.0 as we know it and actually shows some sign of life. Maybe. Hope does spring eternal, so they say. But that won’t help until and if it eventually does happen, so it’s not something we can rely on today in the here and now. That’s a fantasy. Wishful thinking. Not a living reality.

For years now, I have held the belief, the position, that current qubit technologies are far from ideal and that the ideal (or even simply practical) qubit technology has yet to be discovered or invented. That does beg the question as to whether a reasonable qubit technology will ever be invented. Only time and action will tell.

But what about the stock market for quantum computing?

Hah! We can talk all we want about facts, merit, and value, but the stock market is only partially driven by such factors, called fundamentals.

A huge fraction of the stock market, especially for any new technology is driven by what Wall Street calls animal spirits. Noted economist John Maynard Keynes coined that term. Much — or even most — of Wall Street is driven by emotion, passion, popularity, sentiment, and all manner of psychological factors. Hopes, dreams, aspirations, optimism, and yes, even the much dreaded hype run wild on Wall Street. Sometimes they call it irrational exuberance.

As famed investor Warren Buffett has put it “In the short run, the market is a voting machine, but in the long run, it is a weighing machine.” Meaning that in the short run popularity, sentiment, and all those psychological factors run the market, but in the longer run fundamentals kick in.

That’s in bull markets. But there are also bear markets, where the psychology and animal spirits run in the exact opposite direction.

There are many schools of thought on Wall Street, a spectrum from pure irrational exuberance on one end, a casino, if you will, to strict fundamentals on the other end, with hybrid methodologies along the way.

New and young companies can indeed thrive in the stock market for some time based on promises and claims of potential, but somewhere down the line investors will want to see real, fundamental results, with substantial revenue, substantial profits, clear business value for customers, a healthy growth rate, and consistency of performance.

For now, Wall Street is rather bullish (favorable) on quantum computing and quantum technologies in general. That may not last forever, or may wax and wane like the weather, but for now there is a definite bullish attitude towards the potential or presumed potential for quantum computing.

So, while I have serious concerns about the technical potential, the merit, and the actual value of quantum computing for production-scale practical applications for real-world problems, I’m also betting that Wall Street will remain enthusiastic for this new technology for at least a couple more years.

I wouldn’t bet the farm on it, but there is money to be made in the relatively near term.

In fact, I have stock positions in every company which is a player in the market for quantum computing, and I’ve had these positions for several years now. I bought IonQ stock as a SPAC back in 2021.

How these companies will fare on Wall Street in the coming years is a real crap shoot, but for now I’m enjoying the ride.

That said, don’t be surprised if two, three, five, seven, or ten years from now it all comes tumbling down as the old dot-com bubble did in 2000. Just as with dot-com, there may be an afterlife, a Web 2.0, and some of the current players may survive, but a whole new wave of players for a Quantum Computing 2.0 may in fact rule the roost in that new world.

And if you think you can predict what the future will be for either quantum computing technology or quantum computing stocks, please be very mindful of the prescient words of famed economist and equally-famed investor and trader, John Maynard Keynes:

• Markets can remain irrational longer than you can remain solvent.

The only guarantee is that it will be a wild ride. So, enjoy… for now!

For more on the stock market for quantum computing, including a list of publicly-traded stocks related to quantum computing, see:

• The Stock Market for Quantum Computing
• https://jackkrupansky.medium.com/the-stock-market-for-quantum-computing-ddd9ce3cc3f6

I do have hope for the younger generations. Most of these kids have too little innate talent, education, expertise, experience, or initiative to break out of the box, out of the mold, to break free from all of the tired old dinosaurs (and their loyal disciples!) who are holding the quantum computing sector hostage to their tired, old, dated ideas and distorted visions, and to chart a new course and blaze a trail to the promised land that I always wished quantum computing would be, but… kids can surprise you sometimes. I’m counting on it. And expecting it.

I look forward to being surprised. Even now at the moment I pen these words, I suspect that there are one or more budding Bill Gates, Steve Jobs, Larry Ellison, Elon Musk, or Larry Page or Sergey Brin somewhere out there, maybe in college, maybe in high school, even junior high school, maybe even in elementary school or even nursery school, or maybe just being born at this very moment, who really are able and committed to completely changing the game of quantum computing from the absolute mess that the quantum computing sector finds itself in at present.

Tick, tock; tick, tock.

Over and out. For now.

To be continued? Possibly. Maybe not. Flip a coin.

This isn’t the absolute end of the line for me, but is more of a place where I hit pause and wait for reality to play out a little more fully, and get more selective.

For more of my writing: List of My Papers on Quantum Computing.