> The whole "telephone number as identifier" bit, and the network discovery it provides, is the primary reason Signal has had the level of adoption it has.
I agree. Signal absolutely should not abandon this. Rather, it should add other user identifiers that can be used alongside phone numbers.
> Storing anything on systems accessible to the Internet is risky. Plain text is much worse than encrypted blobs, but there's definitely still a non-zero risk. That alone makes it unsuitable for those whose life may depend on their ability to maintain secure communications.
I strongly disagree. A lot of critical work has been done with email + PGP, and that's about as leaky (in terms of metadata) as it gets. Obviously there are use cases where you do worry about this, but "storing data on the Internet" is not as such always a problem for those who need the highest security guarantees. Signal adopting alternative user identifiers would open it to use in some of these extreme cases, but would not (of course) make it usable in every situation - and that's okay.
>I strongly disagree. A lot of critical work has been done with email + PGP, and that's about as leaky (in terms of metadata) as it gets. Obviously there are use cases where you do worry about this, but "storing data on the Internet" is not as such always a problem for those who need the highest security guarantees. Signal adopting alternative user identifiers would open it to use in some of these extreme cases, but would not (of course) make it usable in every situation - and that's okay.
I'm old school. If it's connected to the Internet, eventually it will be compromised.
Yes, strong encryption can (and does) make data compromise immensely more difficult in terms of time and resources (much longer than our star will exist -- about five billion years -- which isn't really that big a deal, since the Earth will be uninhabitable in a billion years or so), but once that centralized server(s) is compromised, all bets are off.
I don't disagree that strong encryption is a valuable tool for maintaining data privacy and integrity, but it absolutely does not reduce the risk to zero.
I agree. Signal absolutely should not abandon this. Rather, it should add other user identifiers that can be used alongside phone numbers.
> Storing anything on systems accessible to the Internet is risky. Plain text is much worse than encrypted blobs, but there's definitely still a non-zero risk. That alone makes it unsuitable for those whose life may depend on their ability to maintain secure communications.
I strongly disagree. A lot of critical work has been done with email + PGP, and that's about as leaky (in terms of metadata) as it gets. Obviously there are use cases where you do worry about this, but "storing data on the Internet" is not as such always a problem for those who need the highest security guarantees. Signal adopting alternative user identifiers would open it to use in some of these extreme cases, but would not (of course) make it usable in every situation - and that's okay.