'Solid' light could compute previously unsolvable problems

Possible, but not useful.

Consider a shaped piece of glass. It "calculates" something about how a light-input bounces, is diffused, makes interference patterns, etc. It leverages natural laws directly and quickly, literally at the speed of light. Perhaps with the right shape and the right light-beam, we can answer questions that matter to us.

Now imagine a programmed version, where all those calculations that "just happen because the universe works that way" have to be done by a CPU crunching numbers. (Ultimately relying on its own set of natural laws.) The second case is going to be much slower.

Another analogy might be a bunch of gear-based clockmakers who want to use electronic circuits instead... A gear-based simulation of electrons doesn't really help them.

Quantum physics is - in some sense - strictly more complex than classical physics and therefore simulating quantum physical systems with classical systems, for example a classical computer, causes exponentially growing costs. Besides that quantum physics is non-local and can not be simulated at all if you have only access to local state. There is a prize of a million dollar waiting for you if you can demonstrate the opposite [1].

[1] http://www.science20.com/alpha_meme/quantum_randi_challenge_...

No, not really. There are many reasons. One, the outcomes for quantum theory computations have counterintuitive properties, like entanglement and the observer effect. Two, computations to predict quantum outcomes for aspects we do understand are very difficult. As one example, we can model a hydrogen atom and predict its spectral lines, but any more complex atoms are beyond computational prediction. This is true in spite of the fact that, in principle, we can say what rules and equations should be applied.

The tl;dr: quantum theory computations are very complex and become more complex very quickly.

Quantum systems can be simulated by classical computers, but the growth of the problem is exponential. If you can leverage actual quantum systems to do the work, then you can take a shortcut, for certain classes of problems.

> In a way, it is like answering questions about airplane design by studying a model airplane in a wind tunnel – solving problems with a physical simulation rather than a digital computer.

Unless you're kidding, that will never happen. The use of "solid" in the linked article is a figure of speech. Photons have no rest mass, so they cannot form solids.