I'd respectfully disagree and say it's pretty bad from the computer science side of things. There's plenty of experimental evidence that indirectly suggests FTL information propagation is impossible, if you consider that allowing it enables the efficient computation of NP-complete problems, for which I'll cite Prof. Aaronson at Shtetl-Optimized.

An open source version of a content-based image search engine is already available in the Carnegie Mellon-Intel joint research project called Diamond. (http://diamond.cs.cmu.edu/) It has a facial recognition application called SnapFind which can already perform arbitrary face matching and other filtering (given the image data), and it is available for free. Disclaimer: I work on Diamond, which has many other applications than this, and thought it was appropriate to post.