Boffin Claims Microsoft's 'Quantum Leap' Is Invalid Due To 'Basic Python Errors' (theregister.com) 74
A peer-reviewed Nature critique argues that Microsoft's 2025 Majorana quantum-computing breakthrough -- and its claim that it could enable "a truly meaningful quantum computer not in decades, as some have predicted, but in years" -- is fundamentally flawed. According to Dr Henry Legg, a lecturer at the University of St Andrews, the claims were undermined by omitted data, selective plotting, and basic Python errors that concealed alternative results. Microsoft, for its part, says the bugs were minor and stands by its findings and roadmap. The Register reports: "Last year they claimed to be years, not decades from a 'topological quantum supercomputer,'" Legg told The Register in an email. "My feeling is that they are centuries, not decades away. If it works at all -- and, based on what I have seen, the most likely scenario is that it doesn't work." Based on his analysis of the research Microsoft published in 2025, Legg argues that the company's claims about finding and being able to control the elusive Majorana particle to build a topological superconductor do not withstand scrutiny.
"I demonstrate that Microsoft's tune-up software is flawed and that coding errors resulted in incorrect statements to peer reviewers," said Legg. "Raw data, which was omitted from the original paper, also appears to indicate Microsoft's devices contain considerable disorder and are not compatible with the existence of a topological gap. In other words, the prerequisites for Microsoft's claims do not appear to be met, but this was obscured because this data did not appear in the original publication."
Essentially, Microsoft has proposed a Topological Gap Protocol (TGP) that can be used to detect the phase transition deemed to be a prerequisite for conducting quantum calculations using Majorana particles. Legg argues that based on his analysis of underlying transport data (measurements of particle change) -- omitted from the original publication -- Microsoft chose to focus on results that supported its thesis and ignored data that could be interpreted as a negative result. As he notes in his critique: "The TGP plotting code was set to highlight only the largest purportedly topological region."
"The primary consequence was the omission of other regions that passed their tune-up protocol (the TGP)," said Legg. "When peer reviewers asked if other regions existed, Microsoft inaccurately stated that they had investigated the only region passing the protocol within the explored range. This was not correct." Legg also argues that Microsoft mishandled its code. "The code antisymmetrized bias voltage based on array index rather than physical value," his analysis says.
In other words, Microsoft's researchers made a basic programming mistake by evaluating the array index -- the number identifying a value's position in an array -- instead of the value to which the index refers. "There were two pretty basic Python programming errors that hid these alternative regions," Legg explained. "Their plotting software was hardcoded with a filter (zbp_cluster_numbers=[1]) that forced it to display only the single largest region, concealing other successful results from their phase maps. Changing this to zbp_cluster_numbers=[1,2] shows already a second region." Legg added: "The TGP software transformed the data by simply reversing a Python array (x[::-1]) based on its index position, ignoring the actual physical bias voltages."
"I demonstrate that Microsoft's tune-up software is flawed and that coding errors resulted in incorrect statements to peer reviewers," said Legg. "Raw data, which was omitted from the original paper, also appears to indicate Microsoft's devices contain considerable disorder and are not compatible with the existence of a topological gap. In other words, the prerequisites for Microsoft's claims do not appear to be met, but this was obscured because this data did not appear in the original publication."
Essentially, Microsoft has proposed a Topological Gap Protocol (TGP) that can be used to detect the phase transition deemed to be a prerequisite for conducting quantum calculations using Majorana particles. Legg argues that based on his analysis of underlying transport data (measurements of particle change) -- omitted from the original publication -- Microsoft chose to focus on results that supported its thesis and ignored data that could be interpreted as a negative result. As he notes in his critique: "The TGP plotting code was set to highlight only the largest purportedly topological region."
"The primary consequence was the omission of other regions that passed their tune-up protocol (the TGP)," said Legg. "When peer reviewers asked if other regions existed, Microsoft inaccurately stated that they had investigated the only region passing the protocol within the explored range. This was not correct." Legg also argues that Microsoft mishandled its code. "The code antisymmetrized bias voltage based on array index rather than physical value," his analysis says.
In other words, Microsoft's researchers made a basic programming mistake by evaluating the array index -- the number identifying a value's position in an array -- instead of the value to which the index refers. "There were two pretty basic Python programming errors that hid these alternative regions," Legg explained. "Their plotting software was hardcoded with a filter (zbp_cluster_numbers=[1]) that forced it to display only the single largest region, concealing other successful results from their phase maps. Changing this to zbp_cluster_numbers=[1,2] shows already a second region." Legg added: "The TGP software transformed the data by simply reversing a Python array (x[::-1]) based on its index position, ignoring the actual physical bias voltages."
Quantum Leap is invalid? (Score:5, Funny)
Quick! Somebody call Scott Bakula!
Re: (Score:2)
Quick! Somebody call Scott Bakula!
Dr. Sam Beckett never returned home.
Re: (Score:2)
"Dr. Sam Becket never returned home."
You forgot to misspell his name.
Re: (Score:2)
Sir Clive enters the room
"the most likely scenario is that it doesn't work" (Score:5, Informative)
At this time, this is the only rational stance left. There is no indication that QCs can ever scale to useful size, but a ton of indicators that they likely will not. There is not even solid proof that QCs work at all, because the longest, most complex complex calculation ever done successfully is apparently factoring 29 with a specialized algorithm for 29. That is easily in range for a conventional analog computation by non-quantum mechanisms. Hence while I think it is unlikely, the computation mechanisms that QCs rely on may still turn out to be hallucinations. Also note that even very, very, very minor deviations from the theory (and we _always_ had those in the past as soon as we had equipment to verify theory against reality precisely enough) would completely kill the QC idea. The precision required to do, say, a 128 bit calculation precisely, is unimaginable and a digital computer only reaches it by extreme measures.
Re: (Score:2)
Someone has implemented
bool isEven(uint32_t n);
with this method. See https://andreasjhkarlsson.github.io/jekyll/update/2023/12/27/4-billion-if-statements.html [github.io]
Re:"the most likely scenario is that it doesn't wo (Score:5, Informative)
the longest, most complex complex calculation ever done successfully is apparently factoring 29 with a specialized algorithm for 29.
Shor's algorithm has been used to factor 21, not 29. 29 is prime, and people don't usually talk about factoring primes because the factors are trivial. But otherwise, yes.
Re: (Score:2)
Ah, sorry. I get confused about the number, because it does not really matter, only that it is exceptionally small. You are right, it was 21.
Re: (Score:2)
Re: (Score:2)
29 is prime, and people don't usually talk about factoring primes because the factors are trivial.
I don't think Microsoft will let something as trivial as primality to get in the way of their quantum computing research. REAL men can roundhouse-kick any number into factors if they so desire.
"The obvious mathematical breakthrough would be development of an easy way to factor large prime numbers." -- Bill Gates, The Road Ahead (1995)
Re: (Score:2)
I'm quite sure quantum computers are valid. Whether they're useful is another question. I'll agree that it's not clear that general purpose quantum computers will ever be useful. (I won't agree that it's clear they never will be useful.)
OTOH, specialized quantum computers are already useful. DWave sells one design.
Re: (Score:3)
I'm quite sure quantum computers are valid.
That just means you have no clue what you are talking about...
Re: (Score:2)
"Everybody" agrees that you can build them using more energy than is consumed by conventional algorithms, and that they can be scaled with enough energy.
The question, the debate, is if they provide a "shortcut" that gets around the minimum energy required for a calculation under traditional information theory. If they always take more energy (as seems likely) then they're useless. But still valid as a type of device.
The default position should be skepticism, because we do have a well-established, well-teste
Re: (Score:2)
Do you realize who you're talking to? Slashdot's very own gweihir is the most knowledgeable person in the world on the viability of quantum computing. Indeed, everyone else is an idiot.
Re: (Score:2)
That makes sense, he's an expert in everything else too, so he's gotta be pretty amazing.
Re: (Score:2)
I weep for all the scientists wasting their lives working on quantum computing, which will never work. They'll wish they had listened to gweihir. Same goes for all the companies wasting money on this technological dead end. They need to get their heads out of the sand.
Re: (Score:2)
That is because you have absolutely no clue how research works. There are useful things coming out of QC research. It is just not QCs.
Incidentally, that is the explanation a quantum computing researcher gave me about 35 years ago when I asked directly, in private.
Re: (Score:2)
A field can change a lot in 35 years.
I can see someone saying something like that if they don't feel progress is being made. I met someone at a conference who said, "Everyone thinks their own research is bullshit." That rang true to me.
Re: (Score:2)
OTOH, specialized quantum computers are already useful. DWave sells one design.
And it's been very useful in making money for DWave, so you are in fact correct. It's also been useful for academics being able to publish blog posts and papers arguing over whether it is or isn't actually a "quantum computer", whatever that really is.
Re: (Score:2)
The DWave is a very nice long-running scam that depends on people not understanding what a QC is. Because actual QCs are so incredibly weak (with a factorization record of 21, and 35 still failed), the DWave essentially does something like chemical computing (in the widest sense) to handily beat that and then the scammers pretend they have a powerful QC.
What really gets me that we now have 50 years of no useful computing mechanisms from QC research. And people are still in denial. A lot of really long-term
Re: (Score:2)
As for the rest of your post, topological or Majorana qubits that Microsoft favors have never been reliably proven to exist at all, much less be useful for quantum computing. B
Re: (Score:2)
I like the term "hallucination", because with AI it now describes something made up because of a lack of understanding.
every single performance metric has been improving exponentially for most of the last two decades
Now you are hallucinating. The exact opposite is true. All scaling in QCs has been sub-linear and very likely inverse (!) exponential. Incidentally, factoring 21 is the current, actual, not-outdated-at-all record for a real quantum calculation.
If you want, I can also call you "delusional" instead of "hallucinating". Would fit about as well with what you just claimed.
Re: (Score:2)
All scaling in QCs has been sub-linear and very likely inverse (!) exponential.
Please provide a citation to demonstrate this claim in even a single metric.
Re: (Score:2)
If you actually understood your link you're realize that is basically says, "Quantum computers help us understand the engineering involved in quantum computing."
What it doesn't say is, "Quantum computers are known to provide a shortcut that bypasses the minimum energy requirements predicted by information theory." And that's what you'd need for quantum computing to be useful.
Re: (Score:2)
You have to understand that gweihir knows more than all of the experts actually working in the field. It's a shame that they are wasting their time working on this technology that very obviously isn't going to work, and it's all because they did not seek out gweihir's superior knowledge.
/s
Now, were there errors in the code behind Microsoft's paper? Errors in academic code is something I believe in a heartbeat. Glad someone reviewed it at this level of detail.
Re: (Score:2)
Nope and I did not claim that. What I do know is that many of the non-commercial researchers into QCs are quite aware they are not doing it in order to build a QC. They are doing it for side-results and because, due to idiots like you, it is easy to get grant money when you claim you are trying to build a QC. That makes the research quite worthwhile, but it does not make it produce a QC.
Re: (Score:2)
Typical. (Score:3)
Microsoft doesn't have a Legg to stand on.
Boffin (Score:5, Insightful)
A quick perusal of his published papers [google.com] seems that boffin would even be understatement, but a well published researcher who is wholly dedicated to this field of research. And whose papers are appear to be written with well respected researchers in the field like this one [google.com] and several others written with Daniel Loss [wikipedia.org].
From the Daniel Loss article:
His 1998 paper (jointly with David DiVincenzo) proposing the use of spin qubits in semiconductor quantum dots is the foundation of one of the main approaches towards the realization of a quantum computer and (as of 2025) has been cited more than 9000 times.
This doesn't appear to be a critique that should be easily dismissed, guy clearly knows his stuff.
Re: (Score:2)
Re: (Score:2)
That's a fascinating take.
Re: Python ? (Score:3)
Re: (Score:2)
decade chasing Marjorie particles.
I think that's a different type of particle used in space lasers.
Re:Python ? (Score:5, Informative)
What you don't understand is the Python is often used as a method of invoking libraries that are written in more efficient languages. And for the layer that it handles it doesn't introduce unacceptable inefficiencies. E.g., you wouldn't want to do ray tracing in Python, but it's fine for calling a library that does that.
Re: (Score:2)
Re: (Score:2)
I've recently discovered lambda functions in Excel. Between dynamic arrays, LAMBDA(), LET(), and the advanced formula environment module of the Excel Labs add-on (import functions form a GitHub gist), it is possible to do some actual programming in Excel. https://arxiv.org/abs/2309.00115 [arxiv.org] discusses some of that.
I'm in electrical engineering and have started using Excel with these features more often because at least now I don't have to copy long, unreadable formulas into hundreds of cells, and my coworkers
Re: (Score:2)
You can also do that in LibreOffice.
Re: (Score:2)
Last I looked, it didn't have support for LAMBDA. Here's a post indicating it wasn't a feature as of 15 days ago, but someone is working on it: https://ask.libreoffice.org/t/is-a-lambda-function-in-calc-possible/75866 [libreoffice.org].
Re: (Score:3)
I can't help thinking adding features like this to a spreadsheet is a bad idea. Don't all the problems of incorrect range bounds, and being difficult to maintain remain? Doesn't it just encourage people to use spreadsheets for tasks they are not really suited for?
Similarly, I like to think of myself as a bit is Bourne shell expert, especially in its bash variant. I start off with some fairly straight forward script, and then decide to "engineer" it a bit better. Refactor, introduce some functions, make use
Re: (Score:2)
It can be quite dangerous, yes.
The biggest benefit of the new features is the ability to create named functions that can be reused. Which means I can write it once, debug it, put it in a commented text file in a GitHub gist, and then rely on it in the future. That's a lot better than copying some huge, unreadable formula across hundreds of cells.
This can improve the readability of existing spreadsheets. It can also be used judiciously to add nice features that still make sense in spreadsheets. I implemen
Re: (Score:2)
"Quantum computing" is usually a bit of a lie where they're actually just running simulations, written in python, that they believe would valid on a real quantum computer.
And usually what they're simulating is not an actual algorithm, but just the quantum computer itself... "this is the random state the outputs take with no algorithm, and look, the output is equally random!"
Re: (Score:2)
Re: (Score:2)
But what if? (Score:3)
I'm just waiting for the day when I can have a Beowulf cluster of Majorna chips to simulate Natalie Portman and some hot grits.
Re: (Score:2)
But by the time you get that, she'll be over 50, and not at all as appealing.
Re: (Score:2)
Apparently the simulation will be enough for some...
Re: (Score:2)
a LOT of celebs at 50 are as hot as ever
Re: (Score:2)
My wife is over 50 and everybody guesses about 35.
Re: (Score:2)
Re: (Score:3)
With my four-digit ID, I'm sure she'll be age-appropriate for me no matter what.
Re: (Score:1)
From the summary:
Microsoft, for its part, says the bugs were minor and stands by its findings and roadmap.
IOW, they're sticking with their marketing pitch.
# The "If It Doesn't Fit, Just Slicing-and-Dice It (Score:2)
# Step 1: Invert reality using standard index slicing
physics_defying_data = raw_data[::-1]
# Step 2: Filter out all the inconvenient physics
return max(physics_defying_data) # Centuries of progress achieved!
British slang (Score:3)
Apparently "boffin" is a British slang term for a scientist/engineer. I had never heard the term, and with the word being at the beginning of the headline I thought it was a person's name. Then was confused when TFS and the Register article failed to tell me who they were.
Re: British slang (Score:1)
Yes it is indeed a shame that the English no longer teach their children how to speak [duckduckgo.com] and we Americans must pick up the slack.
Re: (Score:3)
Re: (Score:2)
The normalization of anti-intellectualism is sad to watch. When I was a kid it was only tabloids that used that sort of language.
Oh, wait... The Register did turn into a tabloid. That explains it.
Hrm... it appears all the media has turned into tabloids.
The BBC is no longer broadcast for free all around the world, now it's paywalled. (Semi-literates haven't noticed because they give a couple free articles before the paywall turns on)
Maybe Batboy's grandchildren can clean up the mess.
Re: (Score:2)
Apparently "boffin" is a British slang term for a scientist/engineer.
I recently watched a documentary on the British space program, and I recall someone saying something like "this project needs more engineers and fewer boffins". So "boffin" is specifically a scientist in an ivory tower, much like "egghead" in US slang.
Re: (Score:2)
Its original use dates to WWII and was more akin to "military R&D engineer/technician": the people who developed radar, for example; or Barnes Wallis of the bouncing bomb and Lancaster bomber.
The use has generalised over time and would certainly include madcap inventors like Doc Brown from Back to the Future, who is very much not an ivory tower dweller. I would guess that that's the sense which was contrasted with engineers in the documentary.
In the vernacular of The Register, which the headline comes f
Re: (Score:2)
Apparently "boffin" is a British slang term for a scientist/engineer.
Yes, and many boffins died to bring us that information.
They made a leap... (Score:1)
... a quantum leap. ::removes sunglasses::
Maybe ChatGPT 6.7 will produce a better paper (Score:2)
sound like someone had some help from our AI overlords when writing their Python code and our AI overlords are always very helpful to produce the desired outcome.