The Future of Email: From SMTP to XMPP

Rabble, I love this idea, but how does a buddy list transate into an email white list?

At scale, won’t I be “spammed” constantly by permission reuqests?

This is a cool idea. And XMPP seems to be getting a lot of attention in another sphere now too – as a messaging protocol, say, in cloud computing applications. http://brainspl.at/articles/2008/06/02/introducing-vertebra

What would also be cool is if someone like facebook did this, and supported xmpp for im and email, and NOT smtp. Then everybody gets a soandso@facebook.com address, using just the shiny new stuff. That plus adoption from gmail/gtalk would get pretty much everybody’s attention :)

Email can already do what you are asking for, the problem is the interface that email providers give you to there spam filtering technology plus mail server white/blacklists.

Looking at this another way, by automatically adding your gmail contacts to your gtalk buddylist, does google make it’s own version of IM more spammy?

Wait, EX-employer? where are you now?

I think leveraging social networks for this is a bad idea because they are insular and are not considered to be viable or particularly productive by all. If GMail/GTalk could manage to push the envelope here, that would be great, but older users and non-tech users tend to shy away from IM technology. Not to say that IM is inherently bad, on the contrary, there are studies suggesting that IM makes people more productive. I would argue that with the point that e-mail is almost identical, but that’s another story.

If it’s just a matter of the underlying protocol, then the change could be made easily and most easily by Google and AOL as you suggested. Why? Because it would be transparent to users as they both control the server and client software that users are accessing. As for the spam issue, there are some good points that arise. How do we allow new people onto the whitelist? Can people request to be added in the manner of social networks, or do we bring our own friends? Do I as a user then need to add a contact every time I want to receive e-mail from a new sender or source? That would be rather painful and not acceptable. Administering a firewall is bad enough without having to micromanage e-mail. Technology is supposed to make lives easier, not more complex. I’ve rambled on enough. Interesting idea, you have.

Let’s see if I’ve got this right… The idea is to switch the transport of messages from SMTP => XMPP. Fine at the core that’s like saying I wan’t JSON formatted data not XML. It’s just wire format.

The problem becomes that while you can transport a message, who do you receive messages from?

If you only accept email from people in your address book (which hotmail supports) then you’re back to needing a way for anonymous communications. The “biggest” user of this is that order confirmation from XYZ corp that’s going to show up in your mailbox, do you first need to add them as a buddy before a message is accepted? That new website that isn’t OpenID compatible sends you a confirmation message, where did it go?

At the end of the day you still need that anonymous communications channel, but adding more and more structure to the display of messages makes a huge difference.

great idea… thinking if spam would cause any problem

great idea… thinking if spam would cause any problem

I think there are actually a couple of good alternatives to the current mail problems we all have, however I see the bigger problem being the backbone providers, core hardware providers and the carriers. Spam is bad for us, but good for someone selling bandwidth. I see that being the problem with the lack of wider adoption of IPV6 at a carrier level as well, because less private address space means less spoofing and in turn spamming but this would cost the bandwidth providers precious revenue in spam traffic. I think the barrier in adoption of something like XMPP is the loss of revenue rather from the problem it will help to solve rather than the know-how or adoption process…

Maybe there could be some sort of crypto challenge/response aspect whereby people can silently add themselves to your whitelist by correctly answering a question.

From a spammers point of view, they could: A) spam the email address with whatever, my mail daemon sends a crypto challenge that they don’t know how to answer/no response/wrong answer, I’m none the wiser as an end user and they stay off the whitelist.(admin would see the attempt obv)

B) spam the email address, by computing the response to a challenge for a single email address.

If the challenge was some sort of fun crypto problem like in DES that uses the problem of factoring very large numbers, the spamming network would drop like a sack of potatoes before sending out any volume of unsolicited spam.

If a spammer wants to be on my whitelist without my intervention they’d have to earn it :)

Now that I think about it, those “secret questions” when you forget your password would make good challenges for other users to respond to to get on a whilelist.

This seems to sound like gmail.

I think there really is a business market out there for such a client. I would personally pay for such a client. It’s cheaper than a spam filtering solution, and would work better.

The only issue is that if you don’t use it exclusively, that is you use it in addition, than people (or solutions) still have to filter out the spam from the SMTP part of the AND equation. The perception, remember most people don’t really understand SMTP in the first place, never mind know there is a protocol to email, will perceive it as one and the same. I still get spam, so what’s the point…

Therefore you’d also most need two separate email addresses. One for XMPP and one for SMTP. It could be as simple as replacing the ”@” with another character. Whatever it is, the client has to show there is a difference between SMTP and XMPP otherwise it won’t be perceived as working.

XMPP does not have any of the email tooling or meta-data. Each program does its own logging / index /search, each one does it diferently, and theres much less context to work with than there is with email (where conversations have subjects and are naturally threaded). The notion that unbroken streams of conversation can replace packaged messages is, to me, laughable. I think XMPP’s dominate in an inevitability, but only after many decades of adding in the kinds of context that allow email to work across long time spans in occasionally-connected environments. You talk about plenty of drivers for convergence, but it really doesnt sound like you’ve given any thought to what role email serves, or have any idea how different IM is in function and form.