On the Nature of Things

Where's your Ed at | The Generative AI Con

Quite.

HPCWire |
Microsoft’s Big Bet on Majorana Pays Off with New Topological Quantum Chip

Majorana fermion

We'll see. Claims of observing a Majorana fermion have been made before and later retracted.

Typhoon wrote: ↑20 Feb 2025 04:17 HPCWire |
Microsoft’s Big Bet on Majorana Pays Off with New Topological Quantum Chip

Majorana fermion

We'll see. Claims of observing a Majorana fermion have been made before and later retracted.

Nature | Microsoft claims quantum-computing breakthrough — but some physicists are sceptical

At this point I'm thoroughly bored with the machine-learning a.k.a AI hyper-hyperbole.

Science | The End of Disease?

First off,
(1) the huge majority of what people are calling AI these days is in fact machine learning. Nothing wrong with that - machine learning can be great stuff, when you have a large enough and well-curated-enough data set to feed into such systems, and when the problems you're trying to work on have well-defined boundaries. But I would like to add that in my opinion

(2) machine learning does not create any new knowledge. It rearranges information you have already obtained and combs through it looking for correlations and rules and patterns, and it can do a far, far better job at this than any human could. If the problem space you are working in has few enough degrees of freedom in it, it can use these patterns to make extremely useful analogies and predictions - the protein structure work that Hassabis references is a shining example of this.

But (3) I strongly believe that all machine learning is done by looking for patterns in some sort of language (be it a native human language (or a coding language) as with chatbots, mathematical symbols, the language of protein sequence and structure as with AlphaFold et al., and so on. The analogies between letters, words, sentences, and grammar hold up quite well across such systems. And I believe that Wittgenstein was right when he said that in any language there are things that cannot be said.

Quite.

The big IT vendors and government are all in now, its only going to get worse, not better.

doesnt matter what we think.

they are perfectly happy for creatives to slip through the cracks - the perfectly statistically average , predictable , automated idiot, on current "best data" is the dream worker.

innovation is so rare, its best not to worry about - the person who comes in second with a big business behind them will be the one that makes it happen.

noddy wrote: ↑22 Apr 2025 07:55 The big IT vendors and government are all in now, its only going to get worse, not better.

doesnt matter what we think.

they are perfectly happy for creatives to slip through the cracks - the perfectly statistically average , predictable , automated idiot, on current "best data" is the dream worker.

innovation is so rare, its best not to worry about - the person who comes in second with a big business behind them will be the one that makes it happen.

Perhaps.

On the other hand, machine learning [AI] has a long track record of overpromising and underdelivering.

Typhoon wrote: ↑27 Apr 2025 02:17

noddy wrote: ↑22 Apr 2025 07:55 The big IT vendors and government are all in now, its only going to get worse, not better.

doesnt matter what we think.

they are perfectly happy for creatives to slip through the cracks - the perfectly statistically average , predictable , automated idiot, on current "best data" is the dream worker.

innovation is so rare, its best not to worry about - the person who comes in second with a big business behind them will be the one that makes it happen.

Perhaps.

On the other hand, machine learning [AI] has a long track record of overpromising and underdelivering.

The technocratic control grid wants to replace the MIC.

Photonic quantum computing is here.

The Biggest Breakthrough in Quantum yet

Nonc Hilaire wrote: ↑27 Apr 2025 12:58

Typhoon wrote: ↑27 Apr 2025 02:17

noddy wrote: ↑22 Apr 2025 07:55 The big IT vendors and government are all in now, its only going to get worse, not better.

doesnt matter what we think.

they are perfectly happy for creatives to slip through the cracks - the perfectly statistically average , predictable , automated idiot, on current "best data" is the dream worker.

innovation is so rare, its best not to worry about - the person who comes in second with a big business behind them will be the one that makes it happen.

Perhaps.

On the other hand, machine learning [AI] has a long track record of overpromising and underdelivering.

The technocratic control grid wants to replace the MIC.

MIC = "Monkey in a Cage"?; "Military Industrial Complex"?

Doc wrote: ↑16 May 2025 16:53 Photonic quantum computing is here.

The Biggest Breakthrough in Quantum yet

Quite interesting.

A somewhat less saguiine review of photonic quantum computing . . .

Typhoon wrote: ↑17 May 2025 01:21

Nonc Hilaire wrote: ↑27 Apr 2025 12:58

Typhoon wrote: ↑27 Apr 2025 02:17

Perhaps.

On the other hand, machine learning [AI] has a long track record of overpromising and underdelivering.

The technocratic control grid wants to replace the MIC.

MIC = "Monkey in a Cage"?; "Military Industrial Complex"?

Military Industrial Complex.

"There's no school like old school".

FT | Lone coder cracks 50-year puzzle to find Boggle’s top-scoring board

‘Remarkably old-school’ method establishes supremacy of 1,045-word grid

Wikipedia| Boggle

The Register | ML Large Large Language Model [LLM] collapse

GIGO.

Companies are attempting to cross the Nvidia GPU moat with ASICS.

Digitimes | The AI chip wars: ASIC uprising challenges Nvidia's crown as TSMC takes all

AI goes to war in a big way. (the first few minutes are about North Korea but stick with it. 9:45 in.
it is pretty shocking)



North Korea Retrieves a Sunken Warship, Dozens of Balloons Float at the Site, Exposing a Big Secret

Substack - Gary Marcus | A knockout blow for LLMs?

Quite. I'm beyond bored of the hyperventilating hyperbolic hype of LLMs by promoters and their media sycophants.

The light at the end of the silicon tunnel.



This Breakthrough Technology Will Power Everything

The only alternate option to AI powered "We see what you see. We hear what you hear" computing



Deciphering the Mumbo Jumbo of New Laptop Technologies. We'll Simplify



Find a COMPATIBLE Linux Computer for $200+: Roadmap to Linux. Part 1

Reuters | How PR China built its ‘Manhattan Project’ to rival the West in AI chips

The history behind the ASML EUV lithography machine is remarkable.

Construction Physics | How ASML Got EUV

Typhoon wrote: ↑20 Apr 2026 04:36 Google Has a Secret Reference Desk. Here's How to Use It_.pdf

ChatGpt. I imagine LM would be better.

[!scholar] = search only Google Scholar / scholarly-paper results

A couple of caveats:

* I can try to restrict research to Google Scholar-style results, but I cannot guarantee perfect Scholar-only coverage every single time.
* I may not be able to open full text for paywalled papers.
* Sometimes I can get the title, abstract, authors, year, and citation trail, but not the full PDF.

A good way to phrase it is:

* [!scholar] Find recent papers on protein folding benchmarks
* [!scholar] Summarize the evidence on creatine and cognition
* [!scholar] Only use peer-reviewed sources

Quantum Computing and the New Speculative Mania

In the early seventeenth century, a rare flower captivated the Dutch imagination. The tulip, and especially the elusive black tulip, became more than a flower. It became a story, a promise, a symbol of untold future value. Prices rose to extraordinary heights. People bought not for what the flower was, but for what they imagined it might become. Then reality intervened. The market collapsed, and the tulip returned to being what it had always been: a flower.

Today, another black tulip has emerged.
It is called quantum computing.

Billions of dollars flow into laboratories, startups, venture funds, government initiatives, and technology giants pursuing a machine that promises to revolutionize computation itself. The claims are extraordinary. Quantum computers are said to solve problems that classical computers cannot. They are said to exploit superposition, entanglement, and interference to perform calculations in ways fundamentally inaccessible to conventional machines.

The promise is breathtaking.

The question is whether the promise rests upon a solid scientific foundation or upon a particular interpretation of quantum mechanics that may ultimately prove incapable of delivering what investors have been led to expect.

The Hidden Ontology Problem

At the heart of quantum computing lies a philosophical issue rarely discussed outside specialist circles.

Quantum mechanics is extraordinarily successful as a predictive framework. It predicts experimental outcomes with astonishing accuracy. Yet almost a century after its formulation, physicists still disagree about what the theory actually says about reality.

The mathematical formalism is accepted.
The ontology is not.

Some interpretations suggest that quantum systems genuinely exist in superpositions of multiple states until measured. Others propose hidden underlying structures. Some invoke many worlds. Others propose nonlocal connections, superdeterministic relationships, or entirely different frameworks.

The remarkable fact is that these competing ontologies often produce identical observable predictions.

Quantum computing is therefore built not merely upon mathematics, but upon confidence that the mathematical formalism corresponds to a physically exploitable reality.
That distinction matters.

A predictive model can be useful without revealing the true nature of reality. History contains many examples of successful theories that were later understood as approximations.

The Ancient Problem of Incompleteness

The uncertainty highlighted by quantum mechanics is often portrayed as a revolutionary discovery.

In reality, incompleteness has always been present.
Newtonian mechanics never provided perfect knowledge of initial conditions. Measurements always contained error. Calculations always involved approximations. Domains of validity always existed.

Classical physics merely concealed these limitations beneath layers of practical success.

Quantum mechanics brought these limitations into the foreground.

The lesson may not be that reality itself is uncertain.
The lesson may simply be that our knowledge is.
The distinction is crucial.

Uncertainty in observation does not automatically imply uncertainty in existence.

Probabilistic prediction does not automatically imply probabilistic reality.

Yet many popular presentations of quantum mechanics quietly move from epistemological statements about knowledge to ontological statements about existence.
That leap remains controversial.

The Quantum Black Box

Quantum computing relies upon maintaining coherent quantum states while manipulating them in highly controlled ways.

The idealized picture is elegant.

A collection of quantum systems enters a coherent state. Superpositions interact. Entanglement spreads correlations throughout the system. Interference amplifies useful outcomes and suppresses incorrect ones. A measurement then reveals the answer.

The reality is less elegant.
Quantum systems are extraordinarily fragile.

Environmental interactions rapidly destroy coherence. Noise accumulates. Errors multiply. Decoherence constantly threatens to erase the very quantum properties upon which the computation supposedly depends.

The result is a machine that often resembles a black box.
Inputs enter.
Complex quantum dynamics occur.
Noisy outputs emerge.
Enormous effort is then required to interpret, filter, and correct those outputs.

Error Correction and the Growing Complexity

The industry's response to these difficulties is quantum error correction.

In principle, error correction is a triumph of engineering. Classical computers also use extensive error correction.
Yet quantum error correction introduces a paradox.

The more practical quantum computing becomes, the more it depends upon elaborate layers of classical correction machinery.

Thousands or millions of physical qubits may be required to produce a much smaller number of reliable logical qubits.
One may reasonably ask:

At what point does the supporting infrastructure become the actual computation?

How much of the claimed computational power originates in the quantum system itself, and how much originates in the increasingly sophisticated classical machinery wrapped around it?

These questions remain active areas of research.

The Strong Counterargument

Critics of quantum computing should acknowledge the strongest scientific response.

Quantum computing is not based solely on philosophical speculation.

There are genuine achievements.

Quantum devices have demonstrated coherent quantum operations.

Entanglement has been experimentally verified.

Quantum algorithms such as Shor's factoring algorithm and Grover's search algorithm provide mathematically rigorous speed advantages under specific assumptions.

Small-scale quantum processors already perform tasks impossible to simulate straightforwardly on ordinary hardware.

From this perspective, quantum computing is not tulip mania.
It is simply a difficult engineering challenge.

The history of technology is filled with skeptics who underestimated what engineering persistence could achieve.
Quantum computing may yet join that list.

The Skeptical Reply

Yet skepticism remains justified.
Demonstrating quantum effects is not equivalent to demonstrating economically transformative computation.
A laboratory demonstration is not a revolution.
A proof of principle is not a business model.
The history of science contains countless examples of genuine phenomena that never produced the technological transformations initially imagined.
The question is therefore not whether quantum effects exist.
They do.
The question is whether those effects can be scaled, stabilized, controlled, and exploited sufficiently to justify the expectations currently embedded in valuations, investments, and public narratives.
That question remains unanswered.

Black Tulip

The black tulip is a symbol of a deeper human tendency.
When uncertainty is great, imagination rushes in to fill the gaps.

Stories become investments.
Possibilities become certainties.
Potential becomes valuation.

The modern quantum computing industry may ultimately transform the world.

Or it may discover that nature permits intriguing quantum phenomena while refusing to provide the computational miracle investors expect.

If that happens, future historians may look back upon today's enthusiasm and see not a technological revolution but a familiar pattern.
A rare flower.
A captivating narrative.
A market intoxicated by possibility.
And a lesson as old as speculation itself:

The existence of mystery is not evidence for unlimited value.
Not every black box contains a treasure.
Sometimes it contains only a black tulip.

This is becoming a tragi-comical freak show. A quantum black box with thousands—or even millions—of entangled poo particles that one day perform computations that can be used as output that, as it turns out, is just noisy dog poo.

But it doesn't matter: wrap this nothingburger in classical computing technology and error-correction systems that do the actual work—with no added speed at all. The nothingburger q-poo black box serves another very important purpose: attracting enough investors to keep the people behind these q-projects on a nice payroll.

The donut battery comes to mind.

Good link, Parodite.

Nonc Hilaire wrote: ↑20 Apr 2026 19:24

Typhoon wrote: ↑20 Apr 2026 04:36 Google Has a Secret Reference Desk. Here's How to Use It_.pdf

ChatGpt. I imagine LM would be better.

[!scholar] = search only Google Scholar / scholarly-paper results

A couple of caveats:

* I can try to restrict research to Google Scholar-style results, but I cannot guarantee perfect Scholar-only coverage every single time.
* I may not be able to open full text for paywalled papers.
* Sometimes I can get the title, abstract, authors, year, and citation trail, but not the full PDF.

A good way to phrase it is:

* [!scholar] Find recent papers on protein folding benchmarks
* [!scholar] Summarize the evidence on creatine and cognition
* [!scholar] Only use peer-reviewed sources

There's a lot of good and interesting preprints on arXiv that have not yet been peer-reviewed.

As for LLMs, I find them useful to get a quick overview of a topic in mathematics or physics with which I'm not familiar.
However, with the caveat of the tendency to hallucinate. No substitute for reading papers and books in the field.

For those who do not have access to paywalled papers, there's Sci-Hub, for papers up to 2022,
and Anna's Archive. For books, there's Lib-Gen.

Exactly. LLM’s organize and summarize well. They are better at finding questions than answering them.

They don’t hallucinate if you prompt correctly and train them on the resources. They do find the easiest answer and try to please the user. Most hallucination stems from user error - you can simply ask the LLM to check its work or have a different LLM check it.

I like LLM for clearer thinking, as a spell and grammar checker and also as a ghost writer it has a talent. The fact it can hallucinate I especially find valuable for "writing" fiction

As for quantum computing, it made this claim (of course I fed it first the right ingredients, I plea guilty):

Quantum God wrote:No current small quantum device has conclusively shown a useful speed gain that cannot be accounted for, challenged, or at least interrogated in terms of the classical control stack, benchmarking assumptions, error mitigation, and classical simulation limits.

Well, it surely should be considered an amazing technical achievement to keep a collection of qubits in a superposition (undisturbed) for 20 seconds with very fast quantum gates to read out the quantum states. But that is still not the ultimate test if the core quantum processor computes anything to begin with. That is assumed.

The good news, if I were a skeptic who does not believe that "superpostion" has any real obvious ontological meaning since it is a mathematical description of probabilty, to try falsify the notion of physical superpositon I would exactly do what these developers of quantum processors are doing: try prove it doesn't work, no matter all ingenious efforts. The skeptic believes: there will never be a speed gain that cannot be accounted for by conventional error correction layers and other known tech wrappeed around the mystery quantum box.

Now it will take a while before they admit defeat, and sell the argument to investors who are still wating for the ROI, that it is only technical limiations that are in the way, not the fundamentals. In priciple, it could work! Quantum mechanics after all has proven itself beyond reasonable doubt!

Quantum Donut Processors

Quantum God corrected me:

A decisive, reproducible quantum computational advantage would strengthen the case that coherent quantum amplitudes represent a physically exploitable resource. Conversely, a long history of principled failure—despite overwhelming engineering sophistication—would increasingly motivate re-examining what those amplitudes physically represent.