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

[John Levine]

It appears that Roland Ionas Bialke <webmaster at pro-wrestling.biz> said:
>    What is a quantum computer? The question feels deceptively simple. It is a
>    computer that uses quantum effects to run calculations [...]
    
>    --------------------
>    
>    Isn't a oracle this exactly? You have two markets with different prices - and then you take the average price of both markets, that everyone (on both markets) can trade under the same price.

No, that's arbitrage.
>    
>    A quantum effect would be, that one person is selling a water melon here on earth and million light years away on planet X  a water melon is also sold. And we can trade on both market with the same
>    price. Exept that you look from earth, in the time you sell your water melon, you just see on Planet X to plant and water the seed.(Out of the earth perspective..

That sounds more like relatavistic time dilation. Fortunately, Paul
Krugman analyzed this scenario in 1978:

https://www.princeton.edu/~pkrugman/interstellar.pdf

I slogged through the Jurczak paper and I was underwhelmed. It is
turgid and pretentious to the point of unreadability and if I see the
wordoid "concretization" again I will scream.

As far as I can tell, the big news is that building quantum computers
is not a straightforward engineering problem, there are different
approaches that advance at different rates, and even as some
approaches stall, what they learn is often useful in other approaches.

Different approaches work better or worse for different applications,
which somehow turns into an assertion that nobody knows what a working
quantum computer is. That's silly. For cryptographers we know exactly
what it is: one that can run Shor's algorithm at large enough scale to
factor realistic RSA or elliptic keys. Or I suppose there might be
some other algorithm, but we know what the result has to be.

He compares quantum computers to ENIAC, claiming that the path from
ENIAC to modern computing was straightforward engineering since vacuum
tubes were well developed in the 1940s, while quantum is not. That
shows a fairly poor understanding of computing history. It was in
retrospect sheer luck that transistors were invented at the same time;
if all we had were tubes, computing would have stalled in the 1960s
hitting limits of power consumption and reliability.

Then at the end there is the requisite handwaving about AI, where he
says that a great application will be to invent AI training data.  Well,
I suppose you could get funding for that.

R's,
John

PS: He mentions cold fusion, dancing around the fact that it doesn't
exist, but never mentions high temperature controlled fusion, the
classic technology that is always N years away (10 <= N <= 20.)
The basic physics of the 1951 Stellarator was fine but the
engineering required to contain a hot enough plasma has turned
out to be vastly harder than anyone imagined, needing its own range
of theoretical advances.  There's been a lot of different approaches
advancing at different rates, with frequent breathless reports that
aren't entirely backed up by actual results.  Sound familiar?