Absolute gold
"However, some scientists and researchers have questioned whether quantum computing really exists."
Tip top!
IBM has won funding from the US Intelligence Advanced Research Projects Activity (IARPA) programme to develop its quantum computer research, despite ongoing concerns about the technology's practical application. "We are at a turning point where quantum computing is moving beyond theory and experimentation to include …
"There are a lot of doubts that certain quantum technologies could ever work."
I think one will find that Dr Ross Anderson is rather alone his "findings". It's pretty fringe and I hear a bit on the not-particularly-well-thought-out side, being instead on the crank "everyone is wrong except me" position.
Also
"despite ongoing concerns about the technology's practical application"
Dear El Reg, please specify EXACTLY what these "ongoing concerns" are. This is like claiming that there are ongoing concerns about the practical applications of an SSTO vehicle because hell, it doesn't exist just yet.
I think one will find that Dr Ross Anderson is rather alone his "findings". It's pretty fringe and I hear a bit on the not-particularly-well-thought-out side, being instead on the crank "everyone is wrong except me" position.
Ross Anderson has a deservedly high reputation in info-sec, but I admit that after skimming his quantum-mechanics stuff I was pretty skeptical. That's really not my area, but extraordinary claims and all that...
Dear El Reg, please specify EXACTLY what these "ongoing concerns" are. This is like claiming that there are ongoing concerns about the practical applications of an SSTO vehicle because hell, it doesn't exist just yet.
Surely concerns about economic viability of applying QC to a wide range of problems are legitimate. We might not be that far away from very-well-funded actors being able to field QC machines big enough to solve interesting problems, but I haven't seen any signs that we're anywhere near making them practical for widespread use, or even routine use. There's a difference between "the NSA can solve a given instance of ECDH fast enough to be useful" and "a university lab can run interesting-size universal quantum simulations for researchers upon request".
And I'm not sure what in BQP is particularly interesting for, say, business applications. Database search is the case that's usually proffered but implementing the quantum-search oracle required by Grover's algorithm (or enhancements like QPS) to real-world databases in general seems non-trivial. Maybe the interesting business applications are in BPP (which is contained in BQP).
But again this isn't my area, so I may be talking complete rubbish.
Is quantum physics real ? Unquestionably. We can prove that quantum effects absolutely take place on the quantum scale.
But when you scale up, those quantum effects are unobservable. An individual PHOTON may be in a quantum superposition of states, but a LIGHT BULB is either on or off and you cannot invent a practical, usable light bulb which is both on and off at the same time.
The fact that someone got money for research into X does not mean that X will ever deliver anything practical. Researchers live for and are paid for research. Their only interest is in securing funding for that research and they will make any sort of claims for the eventual practical applications that they are sure, given enough time and - crucially - enough money their research will lead to. One day. Possibly. Because all of those applications are possible until proven impossible. Of course, if the research proves things to be impossible (or just too difficult to be practically possible) then when the next funding round comes up, a whole slew of NEW hitherto unforeseen possibilities, not yet proven impossible, will be laid out before the funding bodies.
If there were actual, concrete possibilities to be explored there would be no need for research funding because you would have an alternate source of money: INVESTORS; keen to make a return on their $ and convinced that your development projects are going to lead to such a return.
If there were actual, concrete possibilities to be explored there would be no need for research funding because you would have an alternate source of money: INVESTORS; keen to make a return on their $ and convinced that your development projects are going to lead to such a return.
And that's why pure research has never produced anything useful.
Oh, wait.
Ah, market worshipers and their magical thinking. So cute!
Big Irons are good at number crunching, while Quantum computers are good at decryption.
Well, no.
Mainframes are typically good at I/O and processing high transaction loads. IBM z-series CPUs are not designed for number-crunching; they're CISCy processors designed (for reasons of backward-compatibility) for ease of writing applications in assembly and convenience for business processing, and thus have features like BCD arithmetic. Those are not supercomputer attributes.
Quantum computers are (or would be, if there were any) good at quickly1 solving problems in BQP. Some of those applications happen to be relevant to cryptography, such as factoring and solving the discrete-logarithm problem over various fields, which are useful for breaking asymmetric encryption; and finding a symmetric key using Grover's algorithm - though that simply cuts the effective key length in half, and doubling the length of the keys you use is trivial, so this is really just a matter of QC potentially breaking some older cryptosystems.
But QC is not inherently about decryption, and indeed effective QC-based decryption would involve some hybrid QC and conventional system.
There are likely to be business applications of QC, if and when it becomes economically feasible to apply it to business problems that are large enough to get some benefit from it.
But you're correct that it's unlikely to cannibalize mainframe sales. People aren't going to move their CICS and IMS workloads to QC. That'd be like jetliners cannibalizing the sales of motorcycles.
1Where "quickly" is used in the rather special sense of "having a growth curve that's analytically better than the best one available with computational algorithms, assuming P != NP, and making no claims about actual performance on a given problem".