A central problem with the arguably overhyped field of quantum computing remains the difficulty in objectively ascertaining performance and new developments, as much here relies on indirect measurements. Such is especially the case with topological quantum computing, with its use of Majorana fermions. For a few years now Microsoft’s quantum computing department (Azure Quantum) has made claims here of major progress, which have subsequently repeatedly been shot down in peer review. Their most recent attempt at said progress in topological quantum computing now got a blistering response (PDF) by Henry F. Legg in an article in Nature.
We previously reported on Microsoft’s attempts here in early 2025, when they claimed the detection of the crucial Majorana Zero Mode (MZM), before it faced the criticisms of peer review, including by Legg, which included academically vicious language by some researchers, including terms like ‘essentially fraudulent’.
This raises the awkward question of whether Microsoft’s quantum researchers are just too eager to confirm a discovery, or whether a more benign reason exists.
Majorana Versus Dirac
In traditional quantum computing generally Dirac fermions are used as the qubits for quantum computations, but so far this approach has been fraught with complications and challenges, with decoherence and noise intrusion making long-running computations extremely hard and necessitating the need to run computations multiple times for error-correction algorithms to have a shot at divining a plausible result.
This is where topological quantum computing comes into play, as although it imposes some limitations on its feature set, it would be much more resilient to outside influences. Some confusion here may exist with the referencing of Majorana particles, as fermions come in Dirac, Majorana and Weyl flavors. What is referenced here is actually a Majorana anyon, a quasiparticle that just happens to have the same property as Majorana fermions of being its own antiparticle.
By combining these anyons with braid theory using the intertwining of the anyon world lines it becomes possible to perform operations, which theoretically can be used to create a topological quantum computer.
Essentially, this swaps the very fickle, trapped quantum particles for significantly more stable braided Majorana anyons, which – if confirmed – could herald a significant breakthrough in the world of quantum computing.
Is It Majorana Shaped?
Even if you have created a device that theoretically should create Majorana fermions, the next challenge is to confirm that this is in fact the case. This, roughly speaking, is the challenging point where Microsoft’s attempts the past years have repeatedly ending up beaching themselves. As mentioned earlier, the evidence here is determined indirectly rather than through simple direct measurements or experiments.
When the first semiconductor transistor was demonstrated at Bell Laboratories in 1947 in the form of the world’s first point-contact transistor, it came after many years of theorizing and failed attempts starting in at least the 1920s.
Here the evidence of a working transistor was impossible to ignore, as it obviously worked as an amplifier of current, with even the simple current- and voltage-measuring devices of the era sufficing to establish the simple truth. Subsequently this design was commercialized before eventually being replaced with the bipolar junction transistor and a flurry of other devices that followed once the basic principles had been demonstrated.
In the case of quantum processors, whether traditional or topological, there is no obvious way to replicate such a basic demonstration at this point in time. Even the far more basic case of quantum annealing in the form of D-Wave’s commercial offerings is mired in controversy whether there is any ‘quantum advantage’ to be found here. This is territory where even mighty IBM has seen its quantum advantage claims trolled and outperformed by researchers using a lowly Commodore 64.
Where it concerns Majorana anyons and evidence of MZM, you can of course try to build a finished device that demonstrates a clear quantum advantage, or you can build a more limited device where you deduce the existence of these fundamental elements based on what remain mostly theoretical assumptions.
For its most recent attempt at proving that they had succeeded at creating these anyons and with it topological superconductors, Microsoft’s team used a new procedure they called the Topological Gap Protocol (TGP), which purportedly was able to perform a parity readout from their manufactured devices and use this to prove that they had really achieved their goal this time.
Broadside Peer Review
Consequently, Legg’s most recent critique comes as response to Microsoft Azure Quantum’s paper in Nature which got published as a result of that new approach. In this paper it’s claimed that this time they really did detect topological qubits in this improved test setup with TGP, based on – again – indirect measurements and analysis of recorded data. In Legg’s critique it is this analysis of the measurements that’s being attacked as having been performed incorrectly.
The main issue that he identifies is a selective interpretation of the measurements, focusing on the data that supports the experiment’s assumptions, in what would essentially be confirmation bias. There’s also the argument that Microsoft’s researchers made a number of mistakes in their Python code, where they use the array index rather than its value. After adjusting for said basic Python errors, Legg then got entirely different results based on the same measurements.
As noted by Legg, you can get very similar data signatures from sources like quantum dots. Along with the somewhat fundamental data processing issues, this obviously puts into question just how close the Microsoft team was to actually having created these topological qubits.
Microsoft Strikes Back
Of course, Microsoft’s team got in their reply (paywalled) after taking that broadside salvo. Their main arguments seem to be that TGP has no role in interpreting the RF results – being just a tune-up procedure – that form the basis of the original conclusions, nor do they recognize the issues with TGP that Legg indicated as being valid.
Another point is that Legg offers no alternative physical model that is capable of reproducing the capacitance signal or the RTS phenomenology, and thus the response basically seems to boil down to a curt ‘nuh uh’.
They did acknowledge an off-by-one pixel bug in the TGP processing, but insist that it is only a minor issue.
Effectively, the criticism is rejected, with the original 2025 paper maintained as being valid. This would mean that these topological qubits were truly detected, and with this knowledge a functional topological quantum processor could be constructed and integrated into a larger system.
The Science Continues
As much as academics and science in general can often appear to resemble a shooting gallery where the parties involved are happy to do some sniping, ultimately the scientific method has to prevail. This means the publishing of results, of experimental setups and methods with sufficient details that other researchers can attempt to reproduce the results from fundamentals.
If the Microsoft researchers are correct, then this might be a point-contact transistor moment within the world of quantum computing, which would naturally quickly be confirmed by other teams who would create their own devices and run their own tests, making it a historical fact.
Of course, just in the past few years we saw the Korean LK-99 room temperature superconductor and the controversial EmDrive meet a dismal end at the uncaring hands of peer review, while cold fusion is clinging on in a continuous state of limbo, even as it’s now called ‘low-energy nuclear reactions’.
Perhaps the best part of science is that even if nothing comes out of a research direction, it still offers a fascinating opportunity to learn more about physics, mathematics and so much more. Just in the course of writing this article I had to expand my knowledge of some subjects and refresh it on others. Ultimately this makes even something as controversial as topological quantum computing such a delightful topic to occasionally dive into.
How about they show me a demo that isn’t all graphs and vague explanations about what it does?
A blinky, perhaps? A hello world? Anything?
Quantum Evangelism, not quite as prominent as AI Evangelism, and yet scientifically more out of touch and off base. Evangelism has 2 flavors, 1 public anticipation of the technologies potential, and 2 an investors selling point (which assumes ROI based on flavor 1).
It is a hugely dishonest paradigm, the dogmas of Evangelism will see similar outcomes to other “prophesized” events. 2012, y2k, llm sentience, majorana particles
It used to be said that “there’s no such thing as a failed experiment” because you still learned stuff, possibly more stuff, if you got unexpected results.
But if these days experiments are driven by computer programs and these computer programs have blatant bugs in, I’m not so sure.
Two ‘symposia’ i was fortunate to experience as an undergraduate (at the end of the last century) come to mind. The first was on analog computing, where the presenter pointed out that many basic circuits compute values which are a little bit difficult for a digital computer to replicate efficiently. With the obvious caveat that mostly those computations aren’t terribly interesting to people who need computation. The second was on quantum computing, where someone made essentially the same claim but without the caveat. In place of the caveat, they said that quantum computing would solve every numerical problem in only the time it takes to state it. They used as a specific example, factoring of large numbers (thus ending many kinds of cryptography).
The thing is, the quantum guy didn’t make any attempt to prove equivalence between the spooky quantum effects they proposed to measure, and the desired operation of factoring large numbers. And the longer he spoke, the more i noticed he was simply waving his hands around. He had no math, he had no physics, he had no computational theory, he only had this claim that he kept finding ever more excited ways to state. He did not know how it was supposed to work any more than the audience did!
So i still hold out the possibility that there might be some ‘there’ there, that maybe there’s a guy working on it who is not as much of a mouthbreather as the presenter i witnessed. But it’s been a bunch of years and my impression has only gotten deeper. And that’s the environment the professional teams are working on. They’re being given great sums of money to research snake oil, and so long as their work never leaves their lab, it’s golden. The moment it leaves their lab, everyone can see, obviously, the emperor is not wearing clothes.
Add that funding environment into the existing scientific ‘quantum’ quagmire and there’s just no way any of these groups will ever follow anything but a dead end. I hope someone eventually comes to a clear explanation of spooky quantum effects but i doubt these money-chasers would even recognize it if they stumbled across it.
Very well put. I believe quantum physics research has proven more useful in chemistry than in computing. The term quantum has very heavy baggage and peoples belief systems and values get mixed in making it unintelligible and unscientific generally.
Certainly quantum computation exists in very limited practice, and yet MS has imbued real R&D with quasireligious imagery
Well….I understood the part about the transistor.
I’m glad this debate is capturing public attention now, because it’s been a long time since the hoopla around the announcement of the Majorana One chip and I’ve been wondering what the results were. “Still inconclusive” was not my expectation, however.
^… got lost after the 1st sentence. But thanks!