I find usernames a lot easier to communicate than phone numbers, and they are cheaper to move to another phone provider. And don't require me to share my phone number, which is an extremely high priority channel, with everybody I might want to send a message to.
And if it is face2face, scanning a barcode like e.g. Threema supports is even quicker.
It would be nice if we could avoid that fragmentation, and to be clear, I am not against Signal offering phone number sync or even making it the default.
I'm not convinced that "just put my phone number 0394859489 in" is so much easier than "I don't have my phone number in there, just type in horse-battery-staple" that not having the second option is a large usability win. Or "give me your phone number and i'll add you" ;)
It's very laudable that Signal tries to be perfectly easy, with as few options as possible, but on the other hand I feel the overdo it in many cases where a slightly more manual, hidden way would help to mitigate issues. If it works for you it's fine, if it doesn't (either through convention or actual bugs, the latter ones are a bigger issue for me) you are out of luck. That's not great for adoption either.
> I think signal:wtbob works just as well as tel:+15551212
That doesn't work as well for auto-discovery of other Signal users, though. The latter enables auto-discovery based on people whose phone numbers are already in one's contact books. With the former, one needs to collect other people's usernames. It's not a tough thing to do, but it does make auto-discovery difficult.
> That doesn't work as well for auto-discovery of other Signal users, though.
It does if one's Contacts app stores Signal contact information for one's contacts; then one could simply query using non-Signal information as the key (e.g. one could ask, 'give me Signal contact information for tel:+15551212, mailto:fizzbuzz@barquux.org and http://plus.google.com/SomeUserName').
> one could simply query using non-Signal information as the key (e.g. one could ask, 'give me Signal contact information for tel:+15551212, mailto:fizzbuzz@barquux.org and http://plus.google.com/SomeUserName').
That doesn't seem like a very good tradeoff because then it requires Signal to store/compare _more_ personally identifying information (though I guess you could make many of those things pseudo-anonymous).
> That doesn't seem like a very good tradeoff because then it requires Signal to store/compare _more_ personally identifying information (though I guess you could make many of those things pseudo-anonymous).
Actually, it's possible to do even better that that: one can execute a private set intersection protocol so that you & each of your contacts can discover mutual contacts without revealing others.
Signal aren't required to store any personally identifying information: they choose to do so.
> Can you expand on that? I'm not sure quite what you mean.
Sure! It derives from the SDSI/SPKI work of the late 90s, which demonstrated how one can use petnames (i.e., local nicknames) for people, and then refer to a contact's petnames via one's petname for that contact.
Nope, not fudged: the names are hierarchical. So (name mom) is mom in your namespace, (name mom dad) is dad in your mom's namespace, and (name mom dad June) is June in your mom's dad's namespace.
Agree that the majority of users desire for anonymity is without merit, but so is requiring users to identify themselves.
If a user is willing to put up with a less useful and usable version of Signal - they should be allowed to use it without identifying themselves.
Beyond this issue, there are others that me show a troubling pattern. For example, Moxie has refused to post a "warrant canary" for US National Security Letters.
In spirit, I like Moxie, in practice, I feel like he's the captain of ship flying a false flag.
Anonymity isn't Signal's goal, simple as that. If you're looking for anonymous messaging, you'll have to search elsewhere.
OWS recently received a subpoena and fought to be able to publish it (and their response), the only data they could provide were registration and last contact timestamps.
And if it is face2face, scanning a barcode like e.g. Threema supports is even quicker.