ContentThe thread on social networks and email got me thinking about the real future of email-like correspondence. My suspicion is that SMTP-based email won’t go away completely for a long while, but we will stop thinking of “email” as our primary message delivery mechanism.
Email has a number of advantages: It scales very well. It costs very little to send or receive an email. Email messages can be queued up by mail servers for delayed delivery. Recipients are easy to locate and easy to identify via email addresses.
But there are some problems with email as well. Because it costs so little to send or receive email, unsolicited email (spam) has run rampant and dwarfed the number of legitimate messages. (Addressing this is apparently the premise behind the UseBestMail service that was mentioned in the writebacks in the previous thread — a technique commonly known as “sender-pays.”) While SMTP has robust mechanisms for ensuring that messages are delivered to a server, there is no mechanism in place to guarantee that they are read by the actual recipient (ignoring Microsoft’s use of “return receipts” for a moment.) Email clients are typically built around a delayed correspondence model — back and forth conversations are a little awkward — more like rapid letter writing than a phone call. And the email infrastructure is more suited for text than binary media, particularly media such as audio or video that could be streamed.
Other commonly used messaging protocols (whether you think of them this way or not), are instant messaging, voice mail, SMS, phone calls, blogs, radio and television, etc. What is common to all is the notion of taking some message and delivering it to a recipient or set of recipients.
Looking forward a bit, I see a tiered delivery system being built up and supported by the universe of communications devices. This model would rank all available delivery mechanisms according to the type of media being delivered, the recipient’s capabilities, and the urgency of the message. Moreover, the clients would have a built-in mechanism for encryption and digital signature, something that has surprisingly not caught on with email.
For example, if I wanted to send a quick note to a co-worker that said “the meeting has been moved to 11,” I could type it into a client running on my desktop machine, and the client would digital sign it as me, and forward it to a tier-delivery server (probably running on the same machine as another process). The server would then look up the recipient’s capabilities based on their address (which could look like an email address, but would probably just be a unique id associated with a real name). If the recipient (in this case a co-worker) had registered their AIM or MSN address, the message would be delivered immediately to their IM client if they were logged on and not idle. If they were not available on IM, the server would check to see if they had a phone number registered for SMS. If SMS was unavailable then an email would be sent.
Taking this model further, you could envision this multi-tiered client ported to handhelds, cell-phones, desktops, web appliances, kiosks, etc. The client’s capabilities aren’t that important — only that they are IP aware and can eventually connect to the network. (And in the future they would be IPv6 aware, as unique identification of the device would be useful.) In fact, the client itself could also be an IM application or email reader, so that messages can be delivered directly to the user without switching applications.
Future protocols can easily be added on a per-user basis. Streaming media protocols can be added — if both the client and sender support a protocol, than it can be automatically selected as the delivery mechanism. You could integrate this with text-to-speech capabilities to deliver a message to a voice mail. Or someday in the future, we could use speech recognition technology to go from voice to text. In fact, the media of the message is largely irrelevant — the tiered system will attempt to deliver whatever media to the best of its abilities. For example, a movie could even be posted to a website and a URL sent to the receipient if that recipient wasn’t capable of receiving it in real time.
Privacy and security (and anti-spam measures) can be achieved via a straightforward friend-of-friend network. Based on trust-level or degree of separation certain groups can access your IM address or SMS address, and even modified dynamically based on your current environment. For example, co-workers could reach you via any protocol during work hours, but on weekends are shunted to email. Or strangers (people not in the network) can leave messages in email, but only friends can get you on the phone or on IM.
Add to this the potential of the ad-hoc networks made possible via bluetooth and zero-conf protocols (e.g., Rendezvous) and suddenly the tier-delivery architecture could rapidly adapt to the user’s needs, by making it not only possible to get in touch with people immediately and effectively, but also shut down access when it’s not desired or appropriate.
The key idea behind this is that you want to be able to write (or say) your message and name a recipient (or group of recipients). You really don’t care how it gets delivered, just that it does get delivered in the most expedient way. And the cool thing is that all of the technical pieces are already there — and the tiered model effectively incorporates new technologies as they appear.