Earlier quoted context omitted.
> But: why bother? Email is just one of dozens of messaging systems available to Internet users. No, it's not. It's the only widely available, decentralized system, with which you can send to anyone, if you know the address. None of the big ones is this open. XMPP tried to address this and failed; now Matrix is trying again.
WhatsApp has over a billion users. There are big places where its market share exceeds that of SMS --- another big centralized service that has a userbase comparable to that of email. My conclusion is that the people who care about "decentralized" systems are a rounding error. I care about non-technologists managing to send asynchronous messages to each other that are well-encrypted by default. That's a solved proble…
Encrypted email is still a pain
141–150 of 450 posts
Re: Encrypted email is still a pain
#142Re: Encrypted email is still a pain
#143Earlier quoted context omitted.
> But: why bother? Email is just one of dozens of messaging systems available to Internet users. No, it's not. It's the only widely available, decentralized system, with which you can send to anyone, if you know the address. None of the big ones is this open. XMPP tried to address this and failed; now Matrix is trying again.
As a happy user of matrix (via riot.im), I sure hope matrix takes off!
Re: Encrypted email is still a pain
#144moxie HAS to agree with his protocol.
thegrugq? Wasn't he selling exploits to the highest bidder? http://www.forbes.com/sites/andygreenberg/2012/03/21/meet-th...
I can't argue about their qualifications, I argue about their morality.
.edit added link.
Re: Encrypted email is still a pain
#145I do not get why everyone thinks that encrypted email is GPG. S/MIME is supported by almost all email clients. S/MIME is far less of a pain (but still some pain and could be improved). It has a model of how to verify that keys belong to the right person, that actually works in practice in contrast to GPG where you basically have to verify keys by hand (adversarial CAs are a problem, but probably only for a tiny amoun…
In an industry niche where people love the phrase "perfect is the enemy of good", S/MIME is better than GPG/etc because its actually somewhat useable out of the box for people.
Re: Encrypted email is still a pain
#146Earlier quoted context omitted.
> No, I fully understand the problem. If Google Mail vanished tomorrow, a pretty large number of people would probably stop emailing altogether. The number of people for whom that's true increases every year. I highly doubt that's true. Email is pretty essential to the functionality of the internet, from signing up accounts to getting notifications, to just plain discussions with professionals. It's pretty much the o…
Wechat is what's doing this is China, and it's working fairly well for them. It's obviously impossible to do the same in the West (companies won't be trusted by people in Europe, nation-level apps won't be trusted in US) but it's not impossible to replace email. Note: mobile is gigantic compared to desktop in China, so this might also be a reason. I still believe email will outlast all the current solutions though, b…
Re: Encrypted email is still a pain
#147Encrypted email is pretty much over in 2017. The emerging consensus among experts is that it's not worth the trouble, or, worse, incapable of doing much more than generating a false sense of security. That's for a bunch of reasons: * An enormous installed base of clients that won't do encryption, meaning that at best you're attempting to tunnel encrypted messaging over an unencrypted transport. * A protocol that leak…
> But: why bother? Email is just one of dozens of messaging systems available to Internet users. No, it's not. It's the only widely available, decentralized system, with which you can send to anyone, if you know the address. None of the big ones is this open. XMPP tried to address this and failed; now Matrix is trying again.
They have to be email users to have an address. Granted, it's so widespread that even signing up for one of the other messengers often requires an email address, so even the non-email users have email addresses.
But I still think email is going to go the way of usenet, non-internet phone calls and the dodo. Perhaps it's too hard fixing email, compared to just moving to something else. Email is obviously hard to solve since it hasn't been solved already.
Re: Encrypted email is still a pain
#148Earlier quoted context omitted.
> But: why bother? Email is just one of dozens of messaging systems available to Internet users. No, it's not. It's the only widely available, decentralized system, with which you can send to anyone, if you know the address. None of the big ones is this open. XMPP tried to address this and failed; now Matrix is trying again.
Forgive my ignorance, but what caused XMPP to fail? Simply the lack of uptake or is there some other reason?
Everyone[0] is using one of these.
None are XMPP or compatible with XMPP.
[0] I don't technically mean the entire planet.
Re: Encrypted email is still a pain
#149Earlier quoted context omitted.
By making the discredited argument that WhatsApp's key-change behavior is a fatal flaw, you're disagreeing with: * The EFF * Moxie Marlinspike * Matthew Green * Bruce Schneier * Isis Lovecruft from Tor * the grugq * Matt Blaze * Avi Rubin * Steve Bellovin * Joseph Lorenzo Hall * Bart Preneel * Peter Honeyman * Jon Callas (who cofounded PGP Corp) * Paulo Barreto ... and about 50 more experts equally respected in the f…
Do you have sources for that? Because just looking at the first two you named the EFF (who marked WhatsApp down for being closed source). And the owner of the Signal protocol (which is what WhatsApp uses). Obviously he's not going to argue against it.
Re: Encrypted email is still a pain
#150Email is still the text equivalent to the publicly switched telephone network. Domain names have to be registered. Mail routing has to be set up via DNS MX records and mail server software. It is like writing a letter on a postcard. Any attacker will be able to read the origin postmark, the destination address, and the ciphertext, no matter what code you use for the message.
All the problems with key exchange and trust chains is a result of trying to send secure messages from one known person to another known person. It's approaching from the wrong direction. People meet in person to grow their web of trust before exchanging secure messages online. But most of the people I communicate with online, I have never met in person, nor do I even know what they look like.
If I use secure pseudonym-to-pseudonym communication to set up an in-person meeting, and prove to the other that the human controls the pseudonymous identity, it should be impossible to determine how those two real people happened to show up in the same place at the same time. It should be indistinguishable from a coincidental encounter. You just can't do that with encrypted email.
I envision an uncountable number of large rocks, and masked ninjas running around hiding encrypted notes under them. The ninjas hang out in their darkened ninja bars, letting each other know which dead drop to use if they want to start a correspondence, and how to signal that it should be checked. But in the network, nothing exists unless someone runs the service, so all those rocks would have to be created and maintained by volunteers, and there is always the possibility that someone is preferentially using one of their own rocks, or setting up a fake rock to spy on anyone that lifts it.
I don't even know enough cryptography or spy tradecraft to even make a helpful suggestion on how to communicate with anyone on the network in such a way that all known attacks by state-level actors may be mitigated. All I can think of is the Black Hand missions of Oblivion (Elder Scrolls 4)--which used dead drops--and I ended up spoilering all the spoilers, because they had been spoilered. Never mind the philosophical implications of questioning how you even know whether a person you are talking to face-to-face, whom you have known for years, is really the person they appear to be. Do we even need to attach an identity to an actual face most of the time?