5 ms·
Ok, you got me thinking and now I have to assume that someone must have already thought this up, because it would be so immensely useful! Imagine a USB key that
by jballanc 15y ago
Ok, you got me thinking and now I have to assume that someone must have already thought this up, because it would be so immensely useful! Imagine a USB key that behaves like a normal USB key, but if you invert it exactly 3 times in < 10s, then it activates a secondary storage partition. Put mundane family photos or music or such on the first portion, secret documents on the secondary. Sure, someone could disassemble the device to discover the hidden documents, but the point is that most people, after inserting the device and finding the photos and music, would just discard it as uninteresting...
- dekz 15y agoHopefully they don't struggle plugging it in and accidentally rotate 3 times in that time frame, but your point is still valid. Also encrypt with truecrypt and this gets very interesting.
- wnight 15y agoEspecially if when mounted this way it kept much of the regular data and the true-crypt partition (appearing as random noise) was 1-in-4 interleaved throughout making it look like a uninteresting read error on a bank of chips.
- shabble 15y agoStick an accelerometer on there, and detect spikes corresponding to someone tapping the device. Then implement some sort of rhythm-based security as desired :) Turning it around would be pretty hard if it's plugged into a PC, and it can't really do the sensing unplugged unless it's got some sort of internal power, but tap-detection should be fairly robust. You'd probably want to have it remain in whatever secret-mode until unplugged, to stop spurious jolts from accidentally un/re-mounting it (unless you're really paranoid)
- RodgerTheGreat 15y agoThe inherent problem with rythm-based security is that everybody already knows shave-and-a-haircut. :)
- jrockway 15y agoA good passrhythm needs more than two bits of entropy.
- Zash 15y agoBut then you need a battery or something. How about switching to the secondary memory if you plug it in, pull it out and plug it in again? Then you might be able to get away with a small capacitor that would get charged on the first connection, and could activate secondary paths afterwards.