While intentional collaboration is good it is not where the bulk of untapped collaborative potential lies. Accidental collaboration is. But the challenge is to intentionally facilitate accidental collaboration. For the full list of 10 requirements see the original post. Last time I wrote about requirement #3: why usage and context patterns of information are so important.

This week we continue the series investigating requirement #4 where we change gears a bit and move from our previous automation focus and consider the humans. After all it is we-the-meat that actually create and use information. It is the meat part of life which can transmogrify data to information to knowledge to action. So our topic is how and why human revisions of information, annotations to and classifications of information must be enabled and preserved.

In enterprise settings the goal and purpose of our consumption and creation are focused by the tasks at hand, our line of business, or even the company "acceptable use" guidelines. But we are still consuming and creating. No one wants to have a purposeless job. Few enjoy doing throw away work. Meaning is one of those key intangibles that allow us to keep doing drudgery, "I will do this because even though it stinks, it is useful." For knowledge workers - pretty much anyone who consumes and creates information as part of their job - this need for meaning is even greater. In many cases we will even put up with perfectly horrid tasks or working conditions if we know, even in the abstract, that someone somewhere someday will find this useful. So all of us crave meaning or at least that our work matters.

Now enter in Enterprise 2.0 and social media. It has been widely found that recognition is the currency of social media. You are someone if others retweet what you say, read, subscribe or comment on your blog, and listen to what you have to say. This is influence. Social media has wrought is the democratization of influence. No longer is a title, nice car, or tailored suit enough to demand influence (though they still have impact). But when a crowd is listening and responding you what someone says, whether they are a CEO or an Intern, that face commands respect and awareness if not actual influence.

And therein lies the phenomenon of enabling the humans in the enterprise. Businesses can ill afford to ignore good ideas, patterns, trends or brainstorms because of their origin. Yet it is surprising that so many businesses do just that. Ideas, after all, are not all hatched by suits around a board room table. They are embedded in project documentation, they are present in a bespoke integration or customization done for a customer, they exist on blogs. As organizations globalize the hallway and water-cooler conversations of earlier eras have moved online. And yet how much of your company information is filed away in a shared network drive somewhere? How many ideas are embedded in a project proposal or a statement of work or a technical architecture project document for a particular customer? How many good ideas happen to be present on one slide of a power point deck that is 44 slides long?

If your company is anything like most then the answers you have given in your head are probably not encouraging. But the next question you are probably asking is, "Well sure, there might be good ideas in there but how do we know?" Yep. That is a good and fair question. But the answer is in the title of this blog post and in the theme of the series. If you want to get more out of your information then you have to create ecosystems that enable information re-use and accidental collaboration. Rather than relying on serendipity ("Hey, I stumbled across this old power point deck the other day and on slide 18 there is something I think we can do for that new customer!") you foster and environment where people are invited, incented to participate and recognized for their participation.

For example: A consulting project manager might have initially classified a presentation as "consulting", "customer_name", "January 2008", "records management compliance". A sales representative might have a customer who wants some case-file management features. If that sales Rep saw the project manager's presentation returned in a search result set chances are he would ignore it thinking, "its not relevant because its 1) from consulting 2) old 3) for a customer not in my industry 4) a records project and mine is case file management for real-estate." Yet if the ecosystem allowed for better (intentional) human interaction such as social tagging (folksonomy), rating (binary good/bad or scale 1-5 stars etc), social comments, linked content, discussion threads, bursting of the deck into each slide and templated previews then the technology ecosystem starts to make it more likely that the sales rep will find that elusive slide 18 describing the "related items" feature. The rep then has not only some information on where and how that requirement has been met before, but knows who created it, who implemented it, and has a customer who he can call for a potential reference. The trick here was that the technology ecosystem made accidental collaboration more likely and less reliant on luck.

So what are the specific "human enablers" that should exist in such an ecosystem? Here are 6 for you to think about. The list is not comprehensive. But it is a good starting point:

1) Information should be centrally manged though not necessarily centrally stored or centrally indexed. This means that an enterprise content management system is absolutely vital. Distributed storage of the bits and bytes is fine. But there should be one place that is "aware" of the existence of business information. I've talked with others in the enterprise 2.0 community about this before as this interview with Sameer Patel of Pretzel Logic describes. Enterprise search advocates may be wanting to get in on the conversation here but it is important to recognize that enterprise search is complementary, not self-sufficient. It can do a good job of hunting for elusive information provide that it knows where to look. Furthermore as information proliferates, run-of-the-mill enterprise search solutions suffer from the same info-glut that other search systems do. Information ecosystems should have a central repository that is aware of the diverse locations throughout the enterprise where business information is located.

2) Keeping track of how information evolves in it's primary intent is valuable. We have written a lot about how tracking and measuring how others use information is valuable. But it should not be ignored that information originators (creators, authors, collaborators, etc) always have a purpose in mind when they create. Those purposes change over time and understanding how the information and structure of a purposed information artifact changes gives important clues to the context, market drivers, problems and opportunities with the intended audience of the artifact. Information ecosystems should have the ability to store and track and chart the evolution of content and information. For transaction histories this is a fairly straight forward business intelligence activity. For unstructured content items it is more difficult since data is trapped within its content item container. Fortunately new entity extraction, text analysis, sentiment awareness, and semantic indexers are coming online that help free data from its container. Once that critical step is achieved then longitudinal analysis of data can commence and previously hidden patterns will emerge.

3) Tagging and social bookmarking are expressly intentional and participatory acts. They allow human consumers of information to classify the information and to save its location (and maybe some comments) for future use. When information systems tap the power of social collaboration they are literally tapping the "enterprise brain". Just like James Surowiecki writes in The Wisdom of Crowds a crowd (or enterprise!) becomes wise when diversity, independence, decentralization of knowledge and aggregation are all put together. This is precisely what folksonomy systems (as popularly implemented as tagging solutions) are designed to do. Similarly, when social bookmarking is enabled and combined with user profiles then priority recommendation patters can easily emerge. If I want to know what the experts in my field are reading then I first must know who the experts are and then what they're looking at in aggregate. For these reasons, information ecosystems should have a folksonomy capability built into them that foster social classification and sharing of artifacts.

4) Similar to social classification but more basic (though no less useful) social rating systems are designed to allow consumers to "vote" on how good a content item is. Whether voting on a scale of 1-5 or with a simple "thumbs up" or "thumbs down" voting allows consumers of information to give back a quick impression that doesn't serve them at all but serves the next set of consumers who want to find the best information and avoid the worst. This kind of "good vs bad" classification is the most basic kind of classification but it serves a very useful purpose. Again you have to be mindful of the context in which business information exists - there is more content than ever. Even specific search terms or narrow topic documents and pages abound. The ability to get right to the best items is vital for enhancing organizational as well as personal efficiency. Ratings are a wonderful human filter when combined with other filtering criteria such as folksonomy classification and content evolution. For instance, version one of a document (lets call it a "Product Feasibility Study" document might be rated 4 out of 5 stars. Version 2 may be rated 3 out of 5 stars. If you are going to use one of them as a template for all future "Product Feasibility Study" documents, which one would you select? The first one of course. You could only do so if ratings and version evolution are tracked by your information systems. Information ecosystems should have human ratings capabilities built in.

5) Comments and Discussion Threads that are tied to each information artifact are vital as well. This allows humans to communicate with other humans over time about the content object. The audience here is not the machine (as it is with metadata) but rather other people - your colleagues and associates and partners. By empowering people to talk about the information in "meta-conversations" that are linked with the original artifact two things are achieved: First additional re-use brainstorms and value-adding contexts are enabled and stored. I can return over time to find out how colleagues used my document, what worked for them and what did not. This in turn will allow me to enhance the content item and version it thereby adding value to the rest of the organization. Second the fact that discussions and comments are linked to the original item is important. After all, we can send emails about documents, projects, images, songs etc. But without immediate access to the information we are talking about in our meta-conversation, we are requiring our audience to use their imagination. When the conversation is close in time to the document or content item, the impact is largely mitigated. But if I ask you to have a conversation about the first Sony Walkman product and I don't share an image or product details with you, we are stuck imagining different things and there is a high likelihood that what you have in your mind is not the same thing that I have in mine. In the worst cases of thise we end up conversing about completely different things. This happens too often in business where we only reconcile with, "Oh! I thought you meant..." Any time we have conversations that start like that time is wasting, efficiency drops, opportunities are lost. Information ecosystems should have discussion threading and comment capabilities that are associated with the content items they refer to built in.

6) The value of linked information cannot be understated. It is what Web 1.0 was built on. Web 2.0 linked people with people and people with content. We have talked in previously about finding value in automatic and hidden linkages. Yet we should not overlook the obvious and intuitive: people are still by far the best brokers of meaning and intent. When we link information together, whether images and text, documents to other documents, user IDs to author metadata field values to discussions we are creating relationships. Those relationships indicate (broker) context and context is intrinsic to our hunt for meaning and understanding. When we link information artifacts together we say implicitly, "these two things are related". More explicit linkage patterns can define the relationship the link indicates: "These items are related to each other because they are all feature requests from customers." The more explicit the description of the linkage, the less guess work others have in understanding original intent. Nevertheless, in too many businesses the only linkage between items we see is that they all appear in a search query. But such an approach ignores vast amounts of information that may not have shared the keyword you were looking for but is nonetheless the keystone in the structure you were attempting to build. For this reason information ecosystems should have content association and linking capabilities built in.

These six items are not the only items that are required for an appropriately human friendly information ecosystem. But they do represent a good-enough starting point. With these technologies in place, people are enabled to participate with the information they consume and create in-the-flow of creating and consuming it. It is important, in thinking about how to secure adoption, that employees not be made to leave the context of their tasks in order to "participate". Research and lost ROI show that it simply wont work. Enabling people to participate with information is critical if we are to consistently achieve relevant automated pattern finding and trend spotting. If we want systems to deliver knowledge to us then we have to be able to first identify the times and contexts and relationships inherent to an information artifact. From there, if we filter on goodness / badness, classification, and then on extracted data, usage patterns and inferential relationships we start to achieve a knowledge aware, understanding information ecosystem. It is this kind of an ecosystem that can spur collaboration with others, unknown to us across the globe and through time.

Next week we will continue the series investigating requirement #5 where accessibility will be considered: The information must be portable, referencing and accessible for people and computer systems.