Shocked!
I'm shocked the Microsoft has buggy software!
Boffin claims Microsoft's supposed quantum leap does not compute due to 'basic Python errors'
Prestigious journal Nature has published a peer-reviewed critique of Microsoft's claims to have made quantum computing breakthroughs – and the scientist who wrote the paper has essentially said Redmond got it wrong. Microsoft made its claims of a quantum breakthrough in February 2025 when it revealed tech called Majorana and …
-
-
-
-
Friday 26th June 2026 05:23 GMT DaveK23
Re: The experiment will only work while we do not observe the cat.
No, the experiment will only work while we do not observe those who have observed the cat.
I mean, the experiment will only work while we do not observe those who have observed those who have observed the cat...
And so on and so on, until the end of the universe, when Wigner's Friend makes the final observation ever and the whole thing finally collapses into a single determinate state.
(Tip o' the hat to Baxter's Xeelee cycle!)
(Also, no icon because I am unable to post a superposition of thumb's-up/thumb's-down!)
-
-
-
-
-
Monday 29th June 2026 14:18 GMT CrazyOldCatMan
Re: Too late.
Maybe it's a superposition of too late and not too late...
The Cornish have a phrase for that: direcktly.
Means: We've heard of manana but that *far* too urgent for us. Besides, if it annoys the emmets [1], it's all good.
[1] Equivalent to the Devon 'grockles' aka tourists. Emmet is the old Kernewek word for an ant (I think..)
-
-
-
-
-
-
Monday 29th June 2026 14:15 GMT CrazyOldCatMan
Re: Quantum computers revealed to be Schroedinger's Cat.
the first useful quantum computer will only work while not being observed by a cat
So no chance of working in this house then..
(Meeow! You bought a super-expensive computer which will only work when not in the presence of a cat! Hmmm.. warmz.. I require my food to be brought to me while I rest on my new expensive, humming throne. Oh, and a nice silk cushion while you are at it. The static is playing havoc with my fur!)
-
Monday 29th June 2026 19:30 GMT DaveK23
Re: Quantum computers revealed to be Schroedinger's Cat.
(Meeow! You bought a super-expensive computer which will only work when not in the presence of a cat! Hmmm.. warmz.. I require my food to be brought to me while I rest on my new expensive, humming throne. Oh, and a nice silk cushion while you are at it. The static is playing havoc with my fur!)
Schrödinger's-cat-like-typing both detected and not detected!
-
-
-
Wednesday 24th June 2026 22:15 GMT Paul Hovnanian
Re: Quantum computers revealed to be Schroedinger's Cat.
The CS folks call those "side effects". And look down upon them in their search for the ideal.
Oddly enough, "performing I/O" is considered to be a side effect. So I guess any program for which you are able to obtain a result is right out.
On a related note: Generating heat would also be a side effect. So I guess software is hopeless.
-
-
-
Thursday 25th June 2026 14:21 GMT RegGuy1
Re: Quantum computers revealed to be Schroedinger's Cat.
Teaser? You mean a rich kid with nothing to do?
-
-
-
Thursday 25th June 2026 18:42 GMT find users who cut cat tail
Re: Quantum computers revealed to be Schroedinger's Cat.
For start, they at best observe Majorana quasiparticles. Meaning, not fundamental particles, but collective behaviours in solid state that are well described by the same equations. This always riles me. It may seem like nitpicking and sometimes it kind of is. But is is a slipper slope, because then you read people made black holes somewhere – and of course they just made objects following the same maths in some manner.
-
-
Wednesday 24th June 2026 15:31 GMT MattieD
Childish - but expected - response from Microsoft
"Critically, Legg offers no alternative physical model capable of reproducing the capacitance signal or the RTS phenomenology, and does not constitute a substantial scientific challenge to our findings."
Ah, the old idea that you can only offer criticism if you're able to do better. It's such a intellectually-weak, nonsense argument. It's only purpose is to deflect the conversation away from the criticism, rather than address the points made.
I'm not a Michelin chef, but I can tell when a piece of chicken is undercooked.
-
-
-
Thursday 25th June 2026 10:01 GMT Charlie Clark
Re: Childish - but expected - response from Microsoft
Hm, it seems that you haven't read the circular: criticisms both of Major Colonel General Admiral Pete Hegseth, Public Idiot Number One, and Prof Dr Dr Kennedy are now off limits. A team of ICE specialists are on their way to collect, interrogate and deport you back to where you (or your ancestors) came from…
-
-
-
Wednesday 24th June 2026 22:53 GMT Anonymous Coward
Re: Childish - but expected - response from Microsoft
>> Critically, Legg offers no alternative physical model capable of reproducing the capacitance signal
They had two conductors in unexpected proximity to each other? Or some white dielectric material had gotten in from some other Microsoft publication?
-
Monday 29th June 2026 19:37 GMT DaveK23
Re: They had two conductors in unexpected proximity to each other?
OMG, I just spent so long arguing with some dude on phys.org who turned out to be a notorious kook who believes he has discovered longitudinal EM waves, almost entirely due to his failure to understand parasitic capacitance...
-
Monday 29th June 2026 14:24 GMT CrazyOldCatMan
Re: Childish - but expected - response from Microsoft
I'm not a Michelin chef, but I can tell when a piece of chicken is undercooked
It's a good sign if said chicken is still walking around pecking at food..
And, in the case of MS' quantum computer, I'd say it's hiding behind a curtain, projecting in a loud and impressive voice. And Microsoft really, really don't want you to look behind the curtain..
-
-
-
Wednesday 24th June 2026 19:53 GMT Someone Else
Re: Python is not a Microsoft product
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.
The same kind of error I would expect in a High-school coding class (or made by CowPileIt).
-
-
Thursday 25th June 2026 10:05 GMT Charlie Clark
Re: Python is not a Microsoft product
As the article notes, it's the sort of mistake that we all make every now and then. The problem seems to stem from a lack of process to pick it up and that does sound very much like Microsoft's marketing driven approach to software development: who cares if it works as long as enough people buy it.
-
-
Thursday 25th June 2026 13:12 GMT JDX
Re: Python is not a Microsoft product
>>The same kind of error I would expect in a High-school coding class (or made by CowPileIt).
Easy to be sanctimonious but the reality is every developer and coder here has made errors like that. Every one of us has caught them in code reviews of our peers. Everyone one of us has MISSED them in code reviews too.
Anyone who claims they would never make such a trivial mistake is arrogant and somewhat ignorant, because we ALL make trivial bugs and we should be open about that.
-
Thursday 25th June 2026 15:21 GMT Someone Else
Re: Python is not a Microsoft product
Of course we do...happens to everybody. I would do it in high school all the time, until I learned not to. That said, I would not expect a paper that purports to present a scientific breakthrough to have such an error in it. (You don't have high-school students doing the coding for your scientific papers, now do you, Micros~1?)
And trying to rationalize such an error in that environment is more than a little disingenuous.
Code reviews? Yeah, we've heard of them...No, we haven't...
-
Monday 29th June 2026 09:37 GMT Charlie Clark
Re: Python is not a Microsoft product
You need to understand that, for some companies, these kind of publications are essentially part of the marketing. Some US corporations have in the past (IBM, HP, Xerox, Bell Labs, etc.) really have maintained real research departments with minimal inteference (ended long ago). Some still do some, and Google can be credited with more or less continuing the tradition with publications over things like map/reduce, work on network protocols and things like Kubernetes and, more recently, Deep Mind enjoys quite a lot of independence (for a US corporation).
-
-
Friday 26th June 2026 08:44 GMT tiggity
Re: Python is not a Microsoft product
We all make code mistakes & miss things on code review sometimes .
However if code is to be used for promoting a major scientific breakthrough then there should definitely be multiple eyes going through the code with a fine toothcomb. Including multiple experts in that language (as being great in one language does not mean you will be skilled in a different one & may be caught out by syntax quirks - e.g. python is not a heavily MS promoted language so I would expect less MS coders to have core competence in python in than in more heavily MS promoted languages e.g. C#, C++)
In a pharma related coding role I had to get to grips with R (essentially a free stats oriented programming language based on S) and MATLAB.
That was an odd experience as indexes in those start at 1 - so you always ground to a halt on index related code & made sure you were doing it appropriately (not on autopilot 0 based index mindset)
-
Friday 26th June 2026 21:02 GMT ptlambert
Re: Python is not a Microsoft product
They mentioned using AI agents and Copilot. I'm guessing they "vibe coded" some or all of this, perhaps without even looking at the AI-generated code. Also, Python is by now the most popular language out there, thanks to its ubiquity in AI. It's pretty much the lingua franca of AI slop, both in terms of implementing the AI platforms themselves and generating virtually unlimited amounts of AI sludge code.
-
-
-
-
Wednesday 24th June 2026 20:26 GMT steelpillow
Re: Help of Agentic AI
Aww, c'm on! Copilot assures us most emphatically that it is absolutely correct in reporting the Majorana as genuine. You just need to ask it to hallucinate some spurious citations to support its scientific rigour. And if Copilot believes it then any MS marketing droid who doesn't regurgitate the same AI confidence will get fired. Give MS their due, they have a consistent track record.
-
Thursday 25th June 2026 07:12 GMT Dan 55
Re: Help of Agentic AI
Microsoft insisted its work is sound and in early June 2026 announced Majorana 2, a "next-generation topological quantum chip" it developed with the help of its own agentic AI.
You can tell they've vibe coded the python script, the chip, and the reply. The chatbot has maybe even convinced them that they're on the verge of discovering a new kind of quantum physics. The chip will be a mess and the project will be quietly buried and never talked about again.
-
-
Wednesday 24th June 2026 16:56 GMT Bebu sa Ware
"We stand by our results and our roadmap."
deeply rooted in (or by) our AI hallucinations as reality crashes by.
I imagine "stand by" has the same sense as saying "I stand by this train wreck awaiting the ambulances."
Microsoft, AI and Quantum Mechanics — the forces of cock·up never enjoyed so many degrees of freedom.
-
-
-
Wednesday 24th June 2026 21:15 GMT StargateSg7
Microsoft REALLY PICKLED THEMSELVES due to the WAY their technology works! Since I have posted in other forums before about our PARENT company (i.e. NCA - North Canadian Aerospace aka a pseudonym for a Vancouver, British Columbia, Canada-based under-the-radar space-systems and computing company!), I can tell you from PERSONAL EXPERIENCE that using that ridiculous -193 Celcius and below cyrogenic cooling will be the RUIN of all Quantum Computation systems as the introduction of errors and quantum decoherence in literally GUARANTEED by the methods used!
To ENTANGLE and then STORE a quantum state, you need to use a technique originally pioneered by Toshiba Japan (in the 1990's!) and then IBM in hard disk manufacturing where they used MOST LIKELY VALUE based sampling of the magnetic field of a GROUP of rust particles used as spinning disk data storage bit, which allows one to AGGREGATE the MOST-LIKELY CURRENT STATE of a GROUP of bits as the final state value, which is a technique that can now be applied to quantum state values be it a charge, a spin or a position.
In our case, we drill a bunch of holes into a 10 cm by 10 cm flat plate of 5 mm thick Borosilicate Glass using a laser OR electron beam and organize those 300 x 300 x 300 nanometre-sized holes into "Quantum Wells" that are an array of 8192 by 8192 (67,108,864 total wells) which contain trapped Xenon gas (or Argon!) and each hole has a micro-needle jutting a 100 nanometres into the well that is made out of platinum which then goes out to a conductive copper-line-trace which then goes out to a large circular copper plate which can be HIT by a UV laser which will excite the copper and then the platinum needle which then causes a spin state change, a charge change OR a position change of the trapped Xenon or Argon gas molecules using an INFERENCE-based value set or read.
You DO NOT DIRECTLY hit the quantum well itself with a laser but rather use an EXTERNAL FORCE (heat or electrical change in the copper plates and traces!) that SOMEHOW forces (i.e. we are still figuring out the HOW!) a group-based state change which then creates Quantum Value that has super-position (i.e. at least three states and up to 1024 states of value) that could be used as memory storage, emulation of von Neumann microcircuitry, a quantum communications modem OR as actual quantum computation devices AT ROOM TEMPERATURE! No Cryogenic systems are needed!
By organizing a set of wells into a COMBINED GROUP-BASED quantum bit, you can then use 30+ year old Hard Drive digital signal processing techniques to arrive at a Group Concensus as what a given quantum state is assigned to that group of trapped inert gas nano-wells! That group-based sampling, setting and getting is what creates a large Meta-Q-Bit that can then be used in general quantum computation. We go waaaaaaay beyond what D-Wave or IBM or Huawei does and we have the internal proof that STABLE ENTANGLEMENT-based quantum communications and general quantum computationis now possible WITHOUT having to deal with that thorny Quantum Decoherence problem of causing destabilization of a single-Q-bit quantum state into a random value while TRYING to read or write to said quantum well!
We KNOW that a single well has a high liklihood of decohering BUT we also know that at least 25% or more of the wells will STAY STABLE with a value that ON A GROUP-WIDE BASIS points to a specific spin, charge or positional state that IS USABLE as a long-term stable Meta-Q-bit state. At the aggregated throughput of over an Exahertz for the excitation states of individual Xenon or Argon gas molecules trapped within the nanowells, the Stable State Throughput at high sample rates via a pulsed UV laser (i.e. set/get operations) is over a 100 Petabytes per Second for just ONE plate of quantum nanowells!
That is an INSANE amount of computation work that can be done in mere seconds!
Soooooo, Microsoft, D-Wave, IBM, Google, etc should switch over to room-temperature GROUP-CONSENSUS quantum state technologies rather than using cryo-cooled Liquid Helium or Nitrogen to get to stable quantum states of individual Q-bits, and the KEY PART is that the actual device is SIMPLE TO MAKE using common etching systems AND that the materials are inexpensive and that it's all done at ROOM TEMPERATURE!
What's not to like about that?
V
-
Wednesday 24th June 2026 21:16 GMT andy the pessimist
measurementz on only one device
When testing devices on early devices you look for 3 or more devices that have the same behaviour.
With decent process and assembly you should get 5 or more same behavior devices. Hopefully lots more good devices.
Only having one device means cherry picking the device (usually the best one).
It sounds like poor yield and a crap device not somerhing you should make enginrering decisions on.
-
Wednesday 24th June 2026 21:51 GMT druck
Bollocks
I said it was all bollocks at the time, Microsoft couldn't discover their own arse with both hands.
But then again the whole quantum computer field is bollocks- simultaneously representing all possible states and having the solution fall out, violates the fundamental physical principle of:-
"There is no free lunch"
-
Wednesday 24th June 2026 22:56 GMT sarusa
Gee I wonder how that happened?
So what are the chances that buggy python code was spit out by CoPilot? Pretty high, I'd guess, especially since they're claiming v2 was partly co-designed (lol) by CoPilot. And nobody on the team actually has any idea what's actually going on, physically, electrically, quantum-ly, or otherwise, but as long as Microslop is funding them who cares?
But pretty obvious now this is never gonna work. It was all completely hand wavey with only a single 'working' sample and everything was based on the python code finding indirect signals in the data, nowhere near what other qbit tech has demonstrated, which... oops. So basically they have vibe proof (in all senses of 'vibe') that it works. 'You just gotta beweeeeb!'
-
Thursday 25th June 2026 00:33 GMT that one in the corner
Microsoft patent on new subatomic particle in 5, 4, 3...
"We found it and we'll let you have a look at one - if you can afford the licensing fee and sign the NDA*"
Get a couple of gullible governments into a bidding war over the patent and ka-ching!
Not our fault if the winner dropped the entire World's supply when they struggled to get the child-proof lid off; we have sworn testimony that "all of the Majorana Particles currently in Microsoft's possession were poured into the labelled container".
* including the "if I couldn't see it sitting there it is because I forgot to bring my reading glasses, mea culpa" clause.
-
-
Thursday 25th June 2026 05:35 GMT Groo The Wanderer - A Canuck
Re: Majorana
That was just the appetizers before the meth and crack were pulled out at their staff party... it's the only real explanation for why Microslop believes its' own bullshit. Certainly if they had any talented researchers on the project, there wouldn't have been these kind of gaffes in the code, but the fact that they're denying the code errors just proves they're not in possession of a full deck.
-
-
Thursday 25th June 2026 08:28 GMT Taliesinawen
# The "If It Doesn't Fit, Just Slicing-and-Dice It" Protocol
def microsoft_quantum_breakthrough(raw_data):
# 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!
-
Friday 26th June 2026 11:39 GMT ptlambert
"a "next-generation topological quantum chip" it developed with the help of its own agentic AI."
Given this, could these errors be due to vibe coding and AI slop instead of actual scientific research? ChatGPT and its kind will certainly oblige happily and generate large volumes of plausible-sounding text in perfect English, whether or not your prompts make sense. They will even tell you things like "here's a reviewer-friendly version" to describe false results and complete gobbledygook.
Thankfully, the scientific method still involves peer review and falsifiability, so this should all sort itself out over time. In the short term, however, junk science may well impress the easily impressed, mainly because most people can't tell a boson from a bozo or a Majorana from a motorcycle.
