[Cryptography] Why full-fledged quantum computers might always be five years away

[Henry Baker]

-----Original Message-----
From: Jon Callas <jon at callas.org>
Sent: Aug 1, 2026 4:04 PM
To: Cryptography <cryptography at metzdowd.com>
Cc: Jon Callas <jon at callas.org>
Subject: [Cryptography] Why full-fledged quantum computers might always be five years away


 or


[Excerpt, see links above for full article. Well worth reading, as this is along with the sort of things that a number of us have been saying for ages. -- jdcc]


Why full-fledged quantum computers might always be five years away
31 July 2026


To predict when we’ll have a truly useful quantum computer, we need to know precisely what one is. Columnist Karmela Padavic-Callaghan finds that’s harder to pin down than you might think


What is a quantum computer? The question feels deceptively simple. It is a computer that uses quantum effects to run calculations – or at least, that is what I have been saying to friends and acquaintances for years. Because I report about the latest advances in quantum computing nearly constantly, I have felt confident in this answer. Recently, however, I have had to re-examine that confidence. A more detailed question makes clear why: though I can name many different components and functions of a quantum computer, could I identify exactly when they all become a quantum computer? One recent philosophical study suggests that this is at the root of the difficulty of forecasting the future of quantum computing.


For Christophe Jurczak at investment firm Quantonation, it all started at a conference that had nothing to do with quantum computing. Researchers were discussing a different emerging technology, and Jurczak realised that he had first heard about it 30 years ago. Back then, everyone was saying that it was only “five years away”. Déjà vu. Then, a worry set in – could quantum computers also be such a “perpetual five-year technology” (PFYT)? This drove him to do a year-long deep dive into philosophical literature. The paper I read was its outcome.
In it, he argues that, right now, quantum computers are firmly in the PFYT category. All attempts to forecast when quantum computers will truly arrive are misguided so long as they presuppose that we already know what quantum computers are, he writes. “If the identity of a quantum computer is still being settled as the machine develops, that presupposition fails.”


[...]


In order to understand quantum computers, you first have to understand classical computers.


Classical computers come in 2 flavors: analog and digital.


Analog computers were traditionally attempts at *linear* systems, where signal and noise were both amplified simultaneously, and there was no way to (linearly) filter out the noise from the signal.
(This is important to understand, because the "state" of quantum computers also "evolves" linearly.)


Digital computers overcame this problem with analog computers by utilizing *non-linear* components to separate & filter signals from noise.  In particular, "signals" consisted of levels and pulses that could be "squared up" -- i.e., convert substandard signals (e.g., 90% of a standard signal) into *standardized* signals (at 100% of standard).  Thus, in thermodynamic terms, each digital element (e.g., "gate") was effectively a Carnot refrigerator, which removed noise from the important degrees of freedom of a system and exhausted it as "heat" (into other degrees of freedom).  Anyone paying attention to the current power needs of large AI systems might have noticed the energy cost of such "refrigeration".


When engineering a Carnot refrigerator, what elements are required?


We need *insulation*, which *isolates* some degrees of freedom from *interference* by other degrees of freedom.  We also need a Carnot *engine*, which is able to remove the "heat" (energy + information) from one set of degrees of freedom and move it to other degrees of freedom.  The worse the "insulation", the stronger the "engine" has to be.


It is nearly impossible to *isolate* quantum systems from one another, as they love to interfere with one another through "spooky action at a distance".  Some isolation can be obtained by going to extremely low absolute temperatures (hundredths of a degree K); some isolation be be obtained by going to extremely low "effective" temperatures for the important degrees of freedom of the system.


So-called "quantum error correction" can be useful at isolating important degrees of freedom by encoding them in a multiplicity of other degrees of freedom with huge amounts of *redundancy*.  For example, *bosons* love to be in the "same" state, and therefore can provide redundancy; *photons* can possibly be our friends in the battle against noise/interference.


Traditional electrical systems have always had *insulation* and *shielding*.  Insulation takes the form of *non-conductors*, while shielding takes the form of *excellent conductors* which *reflect* electromagnetic waves away from important calculations.  We have already commented on insulation for quantum systems; now consider shielding for quantum systems.


There is a quantum effect called the "Aharonov-Bohm" effect which basically says that it is impossible to shield a region of space from quantum effects.  One interpretation of this effect is that "quantum radar" is strictly more powerful than classical radar.  Unfortunately, the A-B effect also makes building a quantum computer incredibly more difficult.


https://en.wikipedia.org/wiki/Aharonov%E2%80%93Bohm_effect


So, our current best hope for QC is more effective "quantum error correction" (QEC).


Analog computers were replaced by digital computers before "analog error correction" was invented (did it ever get invented?).  So perhaps a good "baby step" towards QEC would be to invent AEC ?


AI systems would be a good place to start on AEC, as the bulk of their calculations are linear, and there have already been many proposals to use analog linear systems to save energy/power in these linear portions of AI calculations.