[Cryptography] Signal chat fallout.

Kent Borg kentborg at borg.org
Fri Mar 28 15:55:31 EDT 2025


On 3/28/25 11:59 AM, Ray Dillinger wrote:
> The thing that pisses me off most about cell phones is, no matter how 
> secure you personally may have your own set up, when you set up a 
> supposedly secure chat, if ANYBODY on the list has accidentally […]

Beyond securing the endpoint devices, securing the endpoint people is a 
fundamental problem of any group. Cue Dr Franklin "Three can keep a 
secret, if two are dead.". (From memory, I hope I have the pretty correct.)

Key design goals of classified communications systems are (1) secure the 
communications, (2) secure all the endpoints, and (3) secure the 
individuals who have access to those endpoints.

Signal does #1 (apparently) very well. It does what it can with #2, 
which is not very much. And for any open system #3 is something between 
a UI problem and doomed, and that is where this blew to pieces: If you 
can't control *membership* in the group the integrity those members is 
beside the point.

-kb

P.S. How good is Signal at #1?

P.P.S. UI design died of its own success: Make something simple to use 
and people mistakenly conclude that it IS simple, merely graphic arts 
painted on top, and can't we make it look really Cool? No. Good UI 
design requires careful thinking into the structure of the thing, it 
goes far beyond skin-deep. (Pun intended.) But in a 
best-practices/design-patterns/feature-velocity/break-things/web-world…no 
one cares.



More information about the cryptography mailing list