4 ms·

Please tell me one thing that a user (power or just every day user) really needs that is not supported by a reasonably recent email client? Sure, here are some

Please tell me one thing that a user (power or just every day user) really needs that is not supported by a reasonably recent email client? Sure, here are some ideas I'd like to see: - Effortless encryption - Verified identity - Pull model rather than push model (like twitter, requires verified identity, would eliminate spam and work for many email addresses which you don't want to be public) - Attachments which are uploaded once, not duplicated and sent over the network - Better filtering (this one isn't bad currently, but can always be improved) - Integration with other forms of messaging - Better support for public messages like mailing lists or mass mailings, e.g. URIs for messages or parts thereof (attachments etc). - Single signon (get rid of passwords on websites by producing an email replacement so awesome people use it as a central identity) ...etc. There are plenty of possibilities, and the ones at the top of this list I think would benefit a lot of people and make things like spam a lot harder. I'm not sure this particular service is going to succeed (it reminds me of app.net in that it targets developers, which is probably a bad idea for mass-market tech), but there are plenty of things we could improve about email and email clients.

VonGuard 12y ago
A lot of these are protocol changes, not necessarily client changes... And I don't see many protocol changes coming for email...
- jrlocke 12y ago
  Yep, totally, email is probably as good as it will ever get. Pack up the bus guys, let's go home.
- worklogin 12y ago
  I think that's the point of the project.
Someone1234 12y ago
> Pull model rather than push model How would someone know to pull from you? Or is every email server (or whatever we're calling it) pulling from every other server all of the time? If your theory is that everyone in the whole world pulls from your specific service, then that kind of contradicts your whole "open source replacement to email" model. > Attachments which are uploaded once, not duplicated and sent over the network So you're depending on the remote party to AV scan incoming attachments? Seems flawed. > Better filtering How? What tech' is in place to make filtering easier/better.
- grey-area 12y ago
  If your theory is that everyone in the whole world pulls from your specific service Think pull as in twitter - as in the recipient controls who can send to their stream. Email is currently completely open to any spammer/idiot to spam your stream of incoming messages - for most people and most email addresses that is suboptimal. Many internal addresses do not require everyone in the world to reach them without permission, and many orgs could have just one point of contact which anyone can push to and leave the rest to be by permission only. So you're depending on the remote party to AV scan incoming attachments? Seems flawed. Scanning email attachments is flawed. What tech' is in place to make filtering easier/better. Bayesian filtering not just for spam but for all messages, filtering by person/org etc etc. Sometimes possible with fiddling today but email clients and the protocol itself could clearly be much improved. I do think the suggestions of encryption and verified identity are more important than any of the above though, which would require fixing the email protocols.
jude- 12y ago
Shameless plug: I'm working on such a system right now. It gives you the security of PGP, but makes key management transparent to the user without compromising security. The gist is that we designed an automatic key distribution and revocation algorithm that requires an adversary to compromise a bunch of different user-chosen services to break it. The (alpha) white-paper is at [1], poster (USENIX ATC 2014) is at [2], and the source code is at [3]. Also, the design is transparently backwards-compatible with IMAP. [1] https://www.cs.princeton.edu/~jcnelson/steak-whitepaper.pdf https://www.cs.princeton.edu/~jcnelson/steak-whitepaper.pdf [2] https://github.com/jcnelson/syndicatemail/blob/master/papers/poster-atc2014/steak-poster-atc2014.odg https://github.com/jcnelson/syndicatemail/blob/master/papers... [3] https://github.com/jcnelson/syndicatemail https://github.com/jcnelson/syndicatemail
- grey-area 12y ago
  Sounds great, thanks for posting the details.

VonGuard 12y ago

A lot of these are protocol changes, not necessarily client changes... And I don't see many protocol changes coming for email...

jrlocke 12y ago
Yep, totally, email is probably as good as it will ever get. Pack up the bus guys, let's go home.
worklogin 12y ago
I think that's the point of the project.

Someone1234 12y ago

> Pull model rather than push model How would someone know to pull from you? Or is every email server (or whatever we're calling it) pulling from every other server all of the time? If your theory is that everyone in the whole world pulls from your specific service, then that kind of contradicts your whole "open source replacement to email" model. > Attachments which are uploaded once, not duplicated and sent over the network So you're depending on the remote party to AV scan incoming attachments? Seems flawed. > Better filtering How? What tech' is in place to make filtering easier/better.

grey-area 12y ago
If your theory is that everyone in the whole world pulls from your specific service Think pull as in twitter - as in the recipient controls who can send to their stream. Email is currently completely open to any spammer/idiot to spam your stream of incoming messages - for most people and most email addresses that is suboptimal. Many internal addresses do not require everyone in the world to reach them without permission, and many orgs could have just one point of contact which anyone can push to and leave the rest to be by permission only. So you're depending on the remote party to AV scan incoming attachments? Seems flawed. Scanning email attachments is flawed. What tech' is in place to make filtering easier/better. Bayesian filtering not just for spam but for all messages, filtering by person/org etc etc. Sometimes possible with fiddling today but email clients and the protocol itself could clearly be much improved. I do think the suggestions of encryption and verified identity are more important than any of the above though, which would require fixing the email protocols.

jude- 12y ago

Shameless plug: I'm working on such a system right now. It gives you the security of PGP, but makes key management transparent to the user without compromising security. The gist is that we designed an automatic key distribution and revocation algorithm that requires an adversary to compromise a bunch of different user-chosen services to break it. The (alpha) white-paper is at [1], poster (USENIX ATC 2014) is at [2], and the source code is at [3]. Also, the design is transparently backwards-compatible with IMAP. [1] https://www.cs.princeton.edu/~jcnelson/steak-whitepaper.pdf https://www.cs.princeton.edu/~jcnelson/steak-whitepaper.pdf [2] https://github.com/jcnelson/syndicatemail/blob/master/papers/poster-atc2014/steak-poster-atc2014.odg https://github.com/jcnelson/syndicatemail/blob/master/papers... [3] https://github.com/jcnelson/syndicatemail https://github.com/jcnelson/syndicatemail

grey-area 12y ago
Sounds great, thanks for posting the details.