A HUGE sign of incompetence is if a company tries to "hide" the way their algorithms work. If they don't know enough about crypto and security to know that this is pointless, they CERTAINLY don't know enough to write secure software.
On that note, I am sure a lot of you use ssh keys.
Do you password protect them?
What about keys used for automated server administration tasks? Surely you can't password protect those. (Do you see the init process typing in a password? ;)
I think USER-CENTRIC KEY MANAGEMENT will be a big trend in the coming years. Not just for key management, but for login to any web service.
Imagine a future where all the "social network" does is transfer opaque encrypted packets from one place to another. The User, with his "keychain" (held on his machine) can browse the "social network" from anywhere and decrypt the messages intended for him.
Using current technology it would be quite inefficient: sharing a new photo would mean encrypting a copy for each of my friends thus transferring an order of magnitude more traffic.
Perhaps new crypto is needed?
Maybe we use AES for the data and send an auxiliary crypto header with 100 copies of the AES key encrypted for each of the 100 friends you wanted to share the picture with.
Research plug: Stefan Brands has invented a very cool upgrade to the basic public-key signature schemes. His protocols allow for "partial disclosure" of only certain parts of a certificate signed by a third party.
(unlike the current sertif. schemes in which I have to show you my entire certificate cleartext so you can hash to check the signature)
Keepass has two levels of access control: "you can access this database" and "you can't". You can use a single database per access group, but once you go beyond two or three groups that's infeasible. You really need a system with a highly granular ACL built-in.
Works well for me - we then sync our password file between us using Dropbox. It creates a lock file that also syncs and stops multiple people from editing simulaneously and messing things up.
Been using 1Password across a PC and a couple of Macs myself and I love it.
As per [1] their data format is pretty open and based on the OSX Keychain format. Uses PBKDF2 to generate unique salts/encryption keys per password. Like any password management system, if you lose your master password, you’re still screwed… But (format being open and based on solid standards) it doesn’t appear (to me [2]) to be vulnerable to a similar attack (i.e, an unintentional backdoor password of sorts).
The Windows version must use some implementation of the same backend, since the data file works across platforms (they’re big on the Dropbox sync support).
[2] i.e., with my limited experience with information security, knowledge of some best practices for password storage, and my interpretation of the data format
