Cooperative tools need to become ‘engines of meaning’

Had a discussion yesterday with Kakul Srivastava of, formerly of Yahoo, Flickr, and Tiny Speck. Got a peak at Tomfoolery’s coming products, but there are some UX changes coming in the near term, and I still have only played with it for a few minutes, so an actual review will have to wait.

What I want to share is an idea that I’ve been stewing on for some time — years, actually. Somehow these ideas crystalized in the discussion with Kakul.

The subject is following. I believe that any cooperative tool — one that is really designed to support a fast-and-loose style of work, and not just another collaboration tool — will have to rely on following as the central mechanism of messaging. Following is a pull mechanism: the potential recipient of information opts to receive messages from people (and other information producing agents). The default mechanism in collaborative tools is membership in groups, teams, spaces, or other defined contexts: belonging. That is a push model, since the creator/owner of the context has to invite the user, and in order to receive any messages from the context at all, the user in essence has to sign up for a long list of things, including symmetric visibility with the other members.

So, I believe that one of the differentiators of the coming generation of cooperative tools is this: while they will support contexts and belonging, other and better mechanisms of supporting messaging through following will be supported.

I reviewed Azendoo recently, and suggested that some of that product’s features put it in at least past the border between collaborative and cooperative tools (see Azendoo is one of the first cooperative work tools). I wrote about Azendoo’s ‘topics’, which at first glance seemed like just another term for a context, but then I realized they’d done something unique:

Users define topics in the workspace as a means of organizing content, and as a way of managing visibility. In most solutions we see some sort of context — a space, project, or group — to which people are invited in order to symmetrically share access to some store of information and conversational streams. However, if you are not invited you can’t see anything going on inside those closed contexts.

Topics are different. They act as a context and as tag to be followed, at the same time. For example, I can create a topic that people can follow like a tag in Twitter, and all public postings made by members of that topic can be seen by anyone following. This support an open follower style of interaction within a business setting.


Azendoo topics are like light: they are both waves and particles. Both a closed, private context and an open, public stream. I believe Azendoo is one of the first to implement this model, and I have been waiting for this breakthrough. Azendoo is one of the first of what will prove to be the next generation of social business software: cooperative work tools.

Talking about the possibility of overload with Kakul, I had an insight regarding the possible filtering of followed streams using tags. So, imagine that I am following a colleague, Bette, on an imaginary cooperative tool, Koan. Koan (let’s imagine) implements a user tag-based filtering. So I could tag Bette with ‘Jones project’, ‘social business’, and ‘NYC’ and then messages that match those tags would surface in my stream.

This requires the Koan system to be very knowledgeable about tags, and to be able to cluster them, like Flickr does, in order to infer things that I would like to see. So when Bette tags something as ‘Manhattan’ or ‘Times Sq’, Koan should pass it along to me. And, following in the footsteps of Azendoo, anything public, or private that I am privy to, regarding the Jones project, I should see. And if she has some breaking news about Yammer or Tomfoolery to share, the social business tag should suffice to pull that information into my stream.

Note that these ‘search tags’ must be interpreted liberally, and not limited to looking just at the explicit tags offered up by Bette. In particular, tags and terms of other users, commenting on Bette’s messages, would be important, even if I am not following them.

Years ago, in 2006 and 2007, I advanced the idea of ‘groupings’ as a replacement or extension of groups. I made the case that all the people using a tag or searching for it represent a ‘grouping’ which has similarities to a group or a context, but which is very different. No one has to invite me to use the tag ‘social business’ or to search for it. I opt to do so without the tag being ‘owned’, and participation in the grouping is very unlike membership in a group. There may be reasons to allow people to define tags that are ‘private’, meaning they define contexts, but I bet in the future the proportion of closed tags will decrease as part of the explosion of connection as we move to a cooperative world of work.

[Incidentally, it was in a Twitter discussion of these ideas, back in August 2007 where Chris Messina proposed the idea of Twitter channels, similar to IRC chat tags, for which I offered up the term ‘hash tag’.]

The Azendoo topic acts like a context and a tag at the same time, but doesn’t include the fuzzy logic that I envision will be necessary for actual serendipity to happen. With that fuzziness, I could decide to follow ‘social business’ and discover people who I don’t know arguing about a new release of Dachis software. And that acceleration of serendipity, that increase in coincidensity, is the magic needed for cooperative work tools to take us past collaboration’s slow-and-tight models. When we can rely on platform-mediated weak ties to surface the information we have made clear we want to see, then that machinery will become ‘an engine of meaning’, sifting through the social exhaust of thousands to find the eleven things I really need to see right now.