In a world where the US government is scanning all your electronic communications, and (we'll next discover) searching your OS X- and Windows-based computers at will, how do you, as a practical matter, keep your private key "private"?
Most people using software only solutions won't ever have their keys stolen, but that's because nobody tried to steal them. The compromise of a client os is inevitable if targeted by a competent actor, given enough time.
Smartcards and HSM's may not be infallible, but their rate of compromise appears to be negligable at best and an extremely rare capability for an offensive team to have access to.
Smartcards are surprisingly cheap and easy to work with, and due to their simplicity and long history are quite secure. The only real attack on them involves physical access and causes obvious physical damage that'd be impossible to miss.
http://shop.kernelconcepts.de/index.php?cPath=1_26&sort=2a&l...
https://www.opensc-project.org/opensc/wiki/OverView
this would probably be the place to start, at least to figure out which type of card you'd want. The main choices are a) support pgp and ssh b) support x.509 certificate based signing c) support time or use type tokens (like smartphone 2 factor apps) or d) some non standardized system running custom code on a tiny jvm inside the card.
a) would be what you'd want in the context of this conversation, but b) is much more supported and has a wider set of use cases.
In most cases it amounts to making sure you buy the right card & reader, plugging it in, and compiling the opensc and related packages
It seems quite unlikely the masses would have access to a trusted platform of any kind, especially considering that any secure platforms for communication that have existed, like Skype, have been opened up. Even good old GSM (AS/1 was it called?) voice-talk encryption was designed with a backdoor in mind at the urging of NATO.
Even assuming it's a compromised platform it's still a hell of a lot more likely to keep your key material safe as compared to having it sit on disk or in addressable address space. One presumes backdoors like that are used sparingly as they become considerably less valuable once publicly exposed.
I recommend disconnecting your monitor and only receiving output by having it blinked out at you through your capslock light on your keyboard. Bonus points if you can get your hands on some TEMPEST hardened hardware, and/or tamper-resistant hardware.
Anything less will leave you vulnerable to the black helicopters!
Note: I'm joking obviously, but this is something to take seriously.
Edit: obviously the 24V must come from a battery which is charged only at specific intervals -- otherwise they can interpret your messages by watching mains voltage variation.
Bonus: Anyone surveilling you via audio bugs will need new ears.
As long as you have a flexible hardware platform that lets you crank up some of the voltage regulator outputs, gpios that can be attached to a long trace/external wire as a makeshift antenna and have a decently fast cpu clock you have all the ingredients for a crude but usable software defined radio. maybe not super fast if you can't repurpose a hardware phy or radio interface, but more than enough bandwidth to exfil a secret key or 10 for maybe a couple dozen meters.
Tools to do sdr utilizing only general purpose processors and no radio specific gear are already available here and there as research implementations, and code that uses gpus/audio dacs/ and re-purposed phys to make a radio interface with a different spec or broadcast frequency is already in production use (wifi phy using a dvb radio interface -> tv whitespace communicator).
Using an approach like that to exfil or bridge an air gap is just too tempting for it to not happen. Honestly, I'd be willing to bet there's already an example of that somewhere out there in the wild today.
Examples of genuine vulnerabilities that would make you look paranoid just by defending against:
* Make educated guesses about passwords from a microphone recording of the keypresses. Both the intervals between keypresses indicate the region of the keyboard being touched, and the sound of each key differs slightly. Given a statistically significant sample of typing, you could deduce which keys are which based on the frequency of their use. http://www.securityfocus.com/news/11318
* Read a screen through a reflection, even from far away http://www.schneier.com/blog/archives/2008/05/spying_on_comp...
It's times like these, I'm grateful for limited terms of office, and a politically divided country.
I'm not a crypto type but I believe what you want is a password-based key derivation function such as scrypt, the output of which you can then use as the symmetric key to encrypt/decrypt the private key. (This might even be what GPG/SSH does for you; I'm not at all sure)