I need to come up with a new term for "Inbox 2.0" because I find it pretty annoying. Maybe next-generation messaging or the Intelligent Inbox or something like that. But on to the topic at hand…
Matt Blumberg made a great post today on Automated Relevance. The entire post is well-worth reading, but I think the most important statement is this one: "I think of relevance as the combination of Relationship and Context." He also writes some good stuff about the significance of the channel/medium of communication, but imo while it’s important, that is very much secondary to relevance.
A lot of what Matt wrote touches on three themes I’ve spent a lot of time on around messaging. Following up on the Three I’s I wrote about, here are four important A’s when it comes to next-generation messaging.
Automation – As the volume of information we are presented with continues to increase. Forget about just increasing email volumes – I now check Facebook updates at least once a day, check Google Reader blog feeds a couple of times a day, have various LinkedIn communications to manage, and occasionally take a peek at Twitter – all new channels of information that just plain didn’t exist in my information processing pipeline until relatively recently. Managing that volume of information requires automation, plain and simple. The simplest piece of that is automating mundane processing steps like filing and categorizing information. More interesting automation can take place at a higher level by intelligently utilizing the context and relevance of information (including the context in which the information is being accessed) to figure out what types of things users are likely to want to do with that information. I give some examples in my post The Context Web.
Aggregation – Two facets of aggregation are important when it comes to processing ever increasing volumes of diverse information. One is aggregating together content from the same streams/channels. Message threading is an obvious basic first step, which is becoming increasingly common. Aggregating messages around related content is a next step. The second facet of information aggregation involves combining information across different channels. Pulling together related email information, combining it with relevant schedule and project information, external information streams, and related content on the web. Zimbra (now Yahoo) is one company that has already started to create that sort of integration to external content from within email.
Accuracy – This is an often overlooked, but critical point, when creating intelligent information solutions. At ClearContext we spend a lot of time balancing the line between accuracy and automation. The more structured and well-defined information is, the more accurate you can be at doing intelligent things with it. In many areas, though, accuracy needs to be effectively 100% or else the solution loses the bulk of its value. If someone needs to double-check their spam filter, the utility of that tool has dropped tremendously. If an application causes you to miss a big meeting because it deemed that information unimportant, you can pretty much kiss that program goodbye. A classic solution to this problem has been to require large amounts of markup/metadata to be provided. However, adding a big burden on the user’s normal workflow just doesn’t work. A more pragmatic approach is to increase the level of automation based on the confidence levels of accuracy. Newsletter? Yes, file that away in the appropriate folder. Note from an important contact that is 80% likely to be a business context and 20% to be personal in nature? Bring that to the user’s attention and let them do the final step in filtering.
APIs – Brad Feld wrote about email APIs today. Programmatic access to next-generation messaging information is definitely important, but I think of APIs in a fundamentally different way when thinking about this stuff. Rather than thinking about them in the traditional sense of data/process access methods, I think the more important ones are information-oriented and context-rich APIs. What do I mean by that? To be honest, I’m not exactly sure. But I don’t think what’s really needed to take solutions to the next level is a bunch more APIs to grab messages and contacts. Instead, I think we need constructs that let us deal with information, relevance, and context. Much of that can even be data-driven, using information contained in the message itself. Looking at services like TripIt on one side and notifications coming into the inbox from various Web 2.0 apps/sites, one can definitely make a case for email itself being the API for email – or at least an important self-contained component of the next-generation messaging API.
We’re currently prototyping a lot of features in/around these areas. I look forward to sharing more specifics in the upcoming weeks and months. In the meantime, here’s an overview of stuff I’ve written recently touching on these topics:
Semantic Web is great, but… – The basis on which a lot of this stuff will be built. Best part of this post is the set of links to good writing on the structured/semantic web.
Contact segmentation… – This will be the basis for a lot of the context/relationship stuff Matt wrote about in his automated relevance post.
Email is sexy again – Contact relationships and priority – a key building block in analyzing email communications.
The Context Web – more thoughts and links on the semantic web, and my take on what to call it.
Inbox 2.0 – Response to NYT post that rekindled a lot of email-focused discussions.
Three I’s of Inbox 2.0 – Information, Interface, and Integration
And of course, this post, the four A’s – Automation, Aggregation, Accuracy, and APIs.