If you happen to have a magstripe writer, then yes.
Those aren't ubiquitous, and neither are portable mechanical key cloners, though both do exist. It isn't even difficult to build a mold to quickly copy mechanical keys, and later clone them at leisure. Yet we still use mechanical keys, as they still have some quite attractive features suitable for everyday use. RFID tag writers are quickly becoming more widespread because the tags are becoming insidiuously widespread, down to passports and bank cards.
NFC, worse luck, comes with a "writer" built right in, as it is but a software-driven RFID emulation. So swapping out codes, once you know what they are, is even easier and can be done with every handset unless someone else (operator, manufacturer) locks your owned device down somehow. And then you jailbreak it because you'd do that anyway.
What you're complaining about isn't the general system but a failure of that implementation. It seems implemented with a naivette that is exactly like what I'm railing about with the pushing of NFC.
To wit: That magstripe system could've been setup better: How many bits can it carry? How many doors and/or door combinations? How sparse a random coding can you make? You could even add a checksum or perhaps a cryptographic hash or some other way to make guessing valid values hard, then flag anomalies loud and clear and drop on the perp's (you know where he is as you know what magstripe reader is being offered invalid cards) with hotel security. Even false alarms are quick and easy to defuse under the guise of being helpful to the guest. That is but a little organisation and fits well with a model where there'll be security-concious personnel in the vicinity anyway. It wouldn't work so well for securing, oh, a holiday cottage or something, and it certainly doesn't work if the surrounding organisation is lacking regardless of where it's deployed.
The security of such a system, in fact and when implemented properly, is resilient to easy changing of keys. So magstripe cards are "good enough" once the model and its failings are well understood, and cheap to replace. Altough for a new system now I'd probably use chipcards instead of magstripe cards, both are in use as are completely mechanical "hole" access cards and a variety of other cards. The idea being to get "good enough" security from cheap keys that are easy to disable as loss happens too often to contemplate having to replace locks or cylinders every time.
I just don't see NFC adding much to the party, except a lot of "it'll be great when everybody has it!" agitprop and convenient ignoring of the drawbacks and complexity. Let's work on resilient security models independent of the actual technical implementation details first, shall we?