Is device independence and non local games only studied for cryptography purposes? I started reading device independent approach on quantum mechanics from here Device Independent Outlook on QM. I am still a beginner in this field, but out of interest I just
browse papers related to it. Most of the papers I find are related to device independence, deal with non-local games ( not exactly sure what they mean ). And generally the papers discuss the some winning probability for a specific game in classical and quantum case and compare them. Other papers deal with the complexity classes related to the games and interactive proofs  related to them (  not sure what they exactly mean ). But the only application upon googling I found were related to different cryptography algorithms. So is device independence and non-local games only studied for cryptography purposes or to bring out the difference between winning probabilities in case of different strategies ( like quantum vs classical ) ?
 A: I'd say that "device independence" is only interesting for cryptographic purposes (at least at present). These are my reasons:


*

*If you want to study a physical experiment, you'll always need to consider the device. You can compare different devices, but you can't do an experiment without specifying the device unless you want to run into philosophical conundrums. Therefore, device independence can only be interesting for non-scientific purposes, i.e. for engineering. 

*Now that we settled that this question is interesting for engineering, we can have a look at different applications. It is immediate that device independence is interesting for cryptographic/security purposes, because this enables you to be sure that the people building the device didn't already trick you. This question is irrelevant for other applications.

*Finally, device independent implementations can help you test whether a device is actually doing what it is supposed to do, and they can do so in a rather simple manner. This might be interesting in other areas, but most applications of QI I can think of (simulations, computation) would be tested by performance tests anyway (do they outperform classical computers?) and thus device independent implementations could be irrelevant - "could", because we'll need better engineering first to be able to see further problems. 


Finally, note that there is a difference between device independent quantum information and nonlocal games. "Nonlocal games" are thought experiments much like regular game theoretic multi-party games with the additional catch that some parties involved in the games share "nonlocal" resources (usually entangled particles). The study of nonlocal games, at least as far as I can see, is done to obtain more insight about how entanglement can outperform classical physics. It is thus a theoretical field. Device independend quantum information will need the ideas developed in this theoretical area, but it only needs a certain number of them and it will deal with applications. 
I'm not an expert, so the following is speculative: I could imagine that you find a connection between nonlocal games and device independent QI very often because it's hard to get money for nonlocal games, while there are "potential applications" for device independent quantum information and since there is a connection, one will exploit the second to be able to do the first.
