What answer does not depend on the question?

Politician answers. They never answer what they are asked for

"An international research group led by scientists from the University of Bristol, UK, and the University of Queensland, Australia, has demonstrated a quantum algorithm that performs a true calculation for the first time."

Dr Xiao-Qi Zhou, who led the project, said: "Before our experiment, there had been several demonstrations of quantum algorithms, however, none of them implemented the quantum algorithm without knowing the answer in advance.

Are they suggesting that experimenters who already knew the prime factors of 15 cheated, and just published a paper _saying_ they'd done it?

How does knowing the answer in advance stop the quantum computer performing a "true calculation" ?

Not trying to detract from their achievement, but whenever someone starts dissing previous attempts at what they've achieved, I get a hackle-raising reaction. (Possibly this dates back to so-called improved versions of "Willy Wonka and the Chocolate Factory", "Psycho" and "Doctor Who"...)

quoting LizR : "How does knowing the answer in advance stop the quantum computer performing a "true calculation" "
As i misunderstand the article, knowing the answer allowed previous workers to 'tune' their process , nudging it more-efficiently towards the answer they wanted. These workers, however, in their own work, then ignored their own a priori knowledge of correctness of the result of their own assignment task they presented to the 'device'.
So, in their own minds, they did not 'tune' the algorithm.
Unfortunately, nothing about the algorithm,the problem, or the answer is presented (which, actually, might be parts of their own algorithm: the last step of which might be: 'deny all fudgings' ...
Now, the price they pay would be not knowing if their own algorithm's result is correct .. but then, nor do we.

As soon as an open access version is available I'll post again.
Still looking.

Well there's this. http://arxiv.org/...76v1.pdf

For everyone asking what difference knowing the answer makes. If I ask you how many beers are left in the case and you write an app to tell you how many are in there. you can write that app in such a way as to prevent obviously wrong answers, such as omitting values higher than 24. you can have a bad program that won't give you a wrong answer. it also wouldn't scale to answer how many beers are in my fridge? At the gas station, or in busch gardens. it's the difference between build a program that can count to ten or building a program that can count, and using it as a subroutine to count to any value you need. only more complex.

Well there's this. http://arxiv.org/...76v1.pdf

If I'm reading the paper correctly (and I'm not at all sure that I am.) then the pre-knowledge of the answer in previous works has more to do with how the reading mechanism at the end was constructed (the one that transforms the qbits of the answer into regular bits).
It's not so much that this was 'hard coded' to give a certain result but that it would more easily read the expected result than any other (i.e. if the algorithm had 'failed' it wouldn't read anything. It would only read a result if the algorihm succeeded.)

but whenever someone starts dissing previous attempts

They're not dissing previous works. They are just showing that they have made an extension that will read any result equally well - not just the one that is expected.

Previous works were all about getting quantum calculations to work at all. This work is a step towards more general usefulness.

so begins the slow death knell of r.s.a. encryption...