[avz]

I believe they talk about the computational task described in [1] (published in 2016). Note that this is somewhat contrived: it's a task defined specifically so that it is very difficult for classical computers and relatively easy for quantum computers. It is also defined in a way that ensures that a quantum computer necessary to demonstrate quantum supremacy is relatively small in terms of the number of qubits.

The task is that of generating output samples from pseudo-random quantum circuits. The paper shows that the task has exponential computational complexity on classical computers and that it can be used to demonstrate quantum supremacy using a quantum computer with approximately 50 qubits.

Now, the IEEE Spectrum article indicates that the team has now built a 49-qubit quantum computer and plan to use it to demonstrate quantum supremacy.

EDIT: Replaced direct pdf link with the article page link.

[1] https://arxiv.org/abs/1608.00263

[maxander]

Am I right to read this as "the classical computer and quantum computer compete at the task of being that particular quantum computer?" Scott Aaronson is probably preparing to denounce this experiment already.

[PeterisP]

Not really "that particular", but more like performing a trick explicitly designed to be feasible even for low-qubit quantum computers but unfeasible even for very powerful conventional computers attempting to emulate quantum computing.

Performing such a trick demonstrates that you have a working device that can perform arbitrary quantum computations - unlike, say, the much hyped D-wave quantum annealing device.

[JCzynski]

If you'd checked the article, you would see that actually he's solidly in favor.

But yes, it's basically "represent a quantum computer of this size", which we legitimately can't do classically beyond a certain size. It's not hugely useful but it's legitimate quantum supremacy, if it works.