I've ranted about this before, and I'll resist the urge to ramble about it again, but the gist is: if there is an encrypted backup option, everyone who's not thinking deeply about it will just use it because it sounds better, and they will end up with no plaintext backup anywhere. In reality though, if you have no plaintext backup, you have a brainwallet. Your coins go with you to the grave, or when you forget the decryption passphrase in 10 years (the first time you ever need it). I believe that it's best for everyone to have a plaintext backup somewhere, and I don't usually support "protecting user's from themselves" (like the drug war, etc), but in this case I think it's preventing a lot of pain. Though, I could probably make some money setting up a service to help people recover their wallets after they forget it...
This is why I was excited about that M-of-N fragmented backups. Because it really opens up the possibilities for backing up your wallet without effectively creating a brainwallet.
One thing I was thinking of doing was having a screen that says something like: "Print a paper backup with a printer-protection key: create a passphrase that is required to restore your wallet from the paper backup, so that the backup information cannot be stolen by a compromised printer. Please write the passphrase in the specified area on the paper backup after it is done printing". This would hide the capability as an extra protective measure, and most users would probably just follow directions and write it on the paper (along with adding extra protection for the Samsung printers with known root exploits). But an expert user could choose not to write it on there. That might be enough to sooth my nerves.
This is all coming with the new wallets... if I ever finish them. It's turning out to be a complete overhaul of some previously-well-tested code, and so it might be a while before I can get them working again (and I probably have to re-write my 1,000+ lines of unit-tests, too). But I think it will be worth it.