Presumably the answer is a result of up-dates to the 2013 version of the Library. When matching peaks in the library there is always an element of judgement involved in any case. I would never blindly accept labels for peaks from a library for an unknown material because what it gives you is the best match it can find. If that has a high probability, 90% say, then it is probably that compound (although not always in my experience). If however the match is below about 85% I would always look at the mass spec of the suggested compound carefully to see whether it matches what is in the data. Sometimes the match is obviously incorrect, but this is a machine and is only as good as the algorithms within it for doing the match. As the libraries are up-dated with more compounds and better algorithms, matches will inevitably change, but should always be checked. The situation is particularly problematic if you have two peaks co-eluting, as the library will never disentangle that for you.
You have also answered your own question to some extent. Different methods of analysis will change what compounds are eluted, how big the peaks are and therefore how good a mass spec you get for each one.