On Apr 13, 2016, at 4:16 PM, Jerry Leichter <leich...@lrw.com> wrote:

>>> Yes, make it significantly smaller than the current form factor.
>> 
>> Ah.  OK, well, that is certainly doable, though how small you can make it is 
>> ultimately limited by the size of the display.  How small do you want it, 
>> and how much are you willing to pay?
> I wonder if one could get rid of the display per se and add some kind of MEMS 
> steerable laser to it.  The output would be projected onto some nearby 
> surface.  This could be physically much smaller.

Hm, that is an interesting idea.  But I think it’s a little more than I want to 
bite off for version 1.

> In another message, you suggested using a passphrase to unlock the thing, so 
> even decapping wouldn't reveal the secrets.  That requires a secure input 
> device.

Only if you have an adversary that pwns your client machine *and* then obtains 
physical control of the device.  For either of these attacks in isolation, a 
non-secure input suffices.

> Going all the way to a virtual keyboard might do the trick.  The keyboard 
> doesn't have to be very good, just functional for this one purpose.
> 
> Of course, this would add significantly to cost, though the one I listed 
> above only costs $40.   What size you could end up with isn't clear.

I’m not trying to protect against every conceivable attack, I’m just trying to 
design an 80/20 solution (actually I think what I have is closer to a 99/1 
solution, but it’s early yet).  One must always keep Munroe’s law in mind:

https://xkcd.com/538/

rg

_______________________________________________
cryptography mailing list
cryptography@randombit.net
http://lists.randombit.net/mailman/listinfo/cryptography

Reply via email to