There is not. That has been outside the scope of this project, but there are several projects for a future "secure wallet protocol".

Here are the best practices in the meantine: https://en.bitcoin.it/wiki/Securing_your_wallet

Sure is! Hopefully the devs have talked about more officially securing the wallet file, but it's nothing we can't do manually until then. Plus you can dump copies of the truecrypt volume on dropbox or wherever so if your PC goes kaput you don't lose the coins!

Is there any mechanism currently in place which would prevent someone from writing a worm or virus who's purpose is to seek out and steal your banking password?

Seems like if someone gets that password you're completely fucked...

I was thinking earlier that it doesn't even have to be a widespread attack.

For example, I have seen threads here about specific users who claim to have a lot of coins. In some cases you can find enough out about a person to design an attack specifically for that person. Be it tricking them into visiting a web page with an exploit that gives you access to their computer, or even just finding out their address and stealing the whole computer.

In the meantime. I am looking forward to seeing what good solutions people come up with for making the wallet less vulnerable to being stolen or lost.

+1
fully agree

Encryption is not enough to solve this problem. It solves it is someone randomly steals your computer physically if the container isn't mounted, but if a worm gains access to your machine or you are targeted for an attack to steal your bitcoins in meatspace putting it in a wallet like that wont help. Need to find a way to add air gaps so that the wallet is not on a machine with internet access, and you can still send coins from it. The only sure way to protect from hackers is to have no internet connection on a machine and nothing that has been on the internet ever going from the machine back to the internet. Check out reddit for more details.

Do you want BTC to be a currency for geeks or one for the masses?

If you want the latter you have keep in mind that most people like cheese... 

This would at least raise the bar. Security is all about raising the bar. Don't take an all or nothing approach. Don't wait for the perfect solution before closing the holes. Having the wallet.dat file completely open is asking for trouble. It's an easy way for a critic to ruin the reputation of bitcoins.

The solution to a keylogger is to have the client present an image of the keyboard with the position of the keys jumbled differently each time and let the user enter the keys based on this layout.

It is impossible to fully secure modern computers because they are too complex to fully understand (top-to-bottom) yet are not formally proven correct at each abstraction layer. In fact, since at least 1996, the trend has been to hide implementation details from the end-user. I personally think that the computer industry won't be mature enough to sustain a stable crypto-currency for another 150 years.

There are two competing security concerns when talking about the wallet.dat:

• Security from attackers who want to spend your wallet
• Security from data destruction

Encrypting the wallet.dat by default will help protect against the first security concern, while making the second one worse.

I plan on using full drive encryption, as well as an encrypted back-up. The passphrase would be written down in two locations. The back-up would be stored in a safety-deposit box and never move with the decryption key, though I may store it with the decryption key (in the safety deposit box).

BTW: for secure encryption, the term is passphrase, not password. Passwords are simply not long enough. A passwords made up of random ASCII numbers, letters, and symbols has about 6 bits of entropy per character. I have seen it reported that a 12 character password is "enough." That works out to about 72 bits of entropy. 64 bits of entropy can likely be cracked by a fast computer within a year. 72 bits increases the difficulty by a factor of 256. 128bits is believed to be computationally infeasible to even count during the lifetime of the universe (energy constraints). If your password was ever published at any time during human history, the entropy is probably less than 64 bits (I don't think more than 1.84x10^19 words/phrases have ever been published, even on the Internet).