What Can Web 1.5 Teach Us About Web 2.5?

Lately there's been lots of buzz and hype about "Web 2.0" (See here for a popular essay by Tim O'Reilly that explains what that generally means.)  In trying to think about how relevant all this might be for clients, I've been thinking a lot lately about "past as prologue".  More specifically, how did "Web 1.0" play out over 2-3 years, and what could that evolution suggest for how "Web 2.0" might unfold?

Six years ago, I joined ArsDigita, an open-source software firm in Cambridge.  In part I was drawn to ArsDigita by a long view of its purpose as a "fifteen-year research project into the best ways for people to collaborate on the Web".  This vision helped attract an incredibly talented group of engineers and some terrific clients like Siemens.  It also dovetailed neatly with my own interests in understanding and managing "costs of coordination" within and between businesses.

In 1999, B2C e-commerce was in full bloom, and predictions for its B2B big brother made B2C just the tip of the iceberg.  Two years later we all stood in the ruins of the New Economy wondering what happened.  While the real lessons were partly drowned out by the sound of the IPO market's door slamming shut and the other, grimmer sounds of 9/11, they were unfolding quietly but surely before us.

In 2001, I co-authored an article in The Journal of Business Strategy entitled, "Deciphering Collaborative Commerce" (March/April 2001, pp. 36–38 ).  The main theme of this article was that the most real, lasting value in the New Economy was being created not by simply opening up a new, highly efficient channel for supply and demand to meet, but in supporting and leveraging "the interaction around the transaction".  As we and many others began to appreciate this, "online communities" moved from the crunchy fringe to the commercial main stage.  But online communities are not quickly built, and require care and feeding to thrive.  So, many failed, and much of the corporate world dismissed them as a fad.  However, closer inspection of successful examples suggests that the lessons were not in whether "online communities" were good ideas, but rather in how they were implemented.

The global ShareNet KM system which ArsDigita helped Siemens' ICN Division build was one such widely-recognized success (see here for the Harvard Business School case study on the initiative).  ShareNet was a highly flexible platform and surrounding process through which this telecommunications equipment division's 15,000 sales and marketing personnel worldwide shared best practices and tools.  Perhaps the most successful KM application of its time, its features and processes embodied several concepts essential to supporting collaboration online:

• intrinsically valuable things to share -- winning proposals and associated tools, reinforced by an extrinsic reward system for contributing ("ShareNet shares", exchangeable like trading stamps for tangible items of real value:  trips, cars, fancy devices, etc.)
• support for subgroups in which users with common interests and needs can interact without being overwhelmed or threatened by an overly broad feature set or information architecture, or excessively large audience. (Research described in "Working Knowledge" by Davenport and Prusak suggests a person's ability to process information effectively falls off significantly when the group from which that information is drawn gets much over 200 people; the hypothesis is that trust plays both a significant conscious and unconscious role in helping us absorb and make sense of what we see, hear, and read.)
• a dynamic "emergent" scheme for the information the system collects and presents, based on a metadata-driven administrative interface that allows the systems maintainers to "stay inside the turning circle" of what users think is valuable to contribute and see.
• active moderation with investment in high-quality "seed content" and offline reinforcement (at conferences and meetings) for online behavior.

In subsequent reflections, a set of ideas for supporting what I've called "Structured Collaboration" emerged.  I wrote them down in a few places:

• this article for Line 56 on social networking software
[note: Line56 is undergoing a makeover and they seemed to have shut down access to old articles; I'll dig up a copy and post when I can.]
• this summary of a KM cluster panel I participated in at IBM Research in January 2005
• this set of tactical lessons from .LRN, an open-source e-learning software project I'm involved with.
• The (true) story that inspired the "Structured Collaboration" idea

So let's fast-forward to 2005.  Today, some of the hottest online business models are described by appending the word "social" in front of something else:  networking (Friendster, LinkedIn), shopping (Yahoo Shopping, Kaboodle), bookmarks (Yahoo MyWeb, Del.icio.us), software (Six Apart, PodShow), publishing (Pluck InSite).  "Emergent" approaches to a semantic structure for the Web are de rigeur.  At a technical level, broadband proliferation has made us all forget about the performance penalties of too much javascript at a page level, and of the significant overhead that machine-to-machine integration at the XML level carries.

But what hasn't changed, and lies in wait for many of these applications?

Let's assume for the moment that whatever's being exchanged in these applications has value (leads on dates, crime statistics presented geospatially).  Web 2.5 will be about creating safe, supportive, and effective places for structured collaboration to take place:

"Safe" will mean not only that information contributors are protected by mechanisms that insure that "only people who should see x, see x", but also that there are mechanisms to appropriately build trust in those contributors among consumers of the information.  Seller ratings on eBay are the poster child for what's meant here.  A prediction -- we're bound to see, at some point, a "portable online credibility profile service" that is analogous to a financial credit rating.  People with an interest in being perceived to be credible online will register their personas on different sites with a third-party service, credibility ratings of these personas will be aggregated (scraped or parsed at minimum if APIs or RSS extension feeds aren't there for this), and as liquidity builds in such a service, sites will incorporate this into their platforms (as Truste vouches for the security of a site, these services will vouch for the reputations of participants in a site).  At an even more technical level, platforms must and will evolve to support complex "subsite" structures with sophisticated permissioning schemes.  .LRN provides a good example of an advanced architecture for this.

"Supportive" will mean that both contribution and consumption of information is coached and encouraged.  Assuming "safe", many more people could be using blogs to share information to their own and their colleagues' and friends' benefit, but they may feel awkward or unsure about Why/ what/ how.  Well designed applications will provide not just "defense-in-depth" help mechanisms for when people have problems, but more extensive tours and ongoing coaching for users as they progress from new to more experienced in their familiarity with the system.  For example, earlier incarnations of the ArsDigita Community System featured a "Curriculum System" module to help people track what they've already learned and what they still need to know to use a system effectively (See http://philip.greenspun.com/doc/curriculum.html).

"Effective" will mean that the contribution and consumption of information in applications is made both simple and obvious.  Applications like our restaurant review experiment are especially useful because the information is easy to contribute (the provided structure prompts, but is flexible enough to accomodate orthogonal descriptions) and consume (the data can be viewed geospatially on the map, or via different sort options).

Let's take a specific example and make some predictions about how it will evolve. 

"Tagging" via services like Del.icio.us is all the rage at the moment among the digerati.  But look closely at Del.icio.us and you will note that the popular tags suggest only a very limited, homogeneous community is using it at the moment.  This is partly a factor of where we are in the adoption cycle for technologies like this.  but it's also partly due to the reality that when people see services like this, the "look and feel" of the language and the data are appropriate for certain communities but not for others.  This "safety"-related insight is a critical realization for both the success of these services publicly and for enterprise applications as well. 

We're experimenting with Yahoo MyWeb, a (beta) social bookmarking service with tagging capabilities that allows users to interact in password-protected subgroups with shared interests.  This service works well for small groups, but the relatively unsophisticated structure of how it handles groups limits its flexibility for rolling it out and maintaining it in large groups with fluid memberships and sub-structures. 

One potentially powerful application of "enterprise tagging" would be for call center representatives who are helping customers use web sites to tag pages and documents.  Thousands of these representatives working together could potentially create a very powerful emergent metadata supplement that could be recycled for other representatives or even into the search tool customers use themselves.  But tags will make even more sense when understood and applied in the context of groups who interpret them in similar ways.

As for supportiveness, Once you "get" tags, you can appreciate how valuable they can be.  But, simply explaining how to tag may be insufficient to getting people to do even this little extra bit of work.  Brief tutorials about "what" and "why", reinforced by "win of the week"-type stories that reinforce the value of using this technology will be essential, as will "seed" tags to get people.

Finally, from an effectiveness perspective, while tags are relatively simple to add, a frequent obstacle is thinking of what tag to apply.  Today's free form approach (whatever keyword(s) you want), may be supplemented usefully by "drop-down" lists of suggested tags in some cases, especially when one dimension of classification of certain pages or documents -- like industry or geography -- could benefit from applying a normalized tag set.  From the consumption side, make it easy to subscribe to tags.  Collaborative filtering could be helpful here -- "users who subscribe to this tag also found these related and useful".  Or, perhaps suggest "starter sets" of tag subscriptions related to a topic or "supertag" that encompasses subsidiary classifications.

In summary, for me Web 1.5 was all about the realization that the real value of the Web for business lies in supporting and leveraging collaboration, not simply opening up a new channel for transaction.  The keys to doing this well: 

• focus "communities" on high-value exchanges;
• structure communities so logical subgroups based on powerful affinities can be created maintained, and easily extended -- not only by site sponsors, but by community members as well;
• support users through the "why" and the "what" of collaborating, not just the "how"; and,
• make contribution and consumption of the information as easy as possible. 

Applications that successfully evolve into Web 2.5 will adapt to these imperatives.

(Some friends ask, "What then is Web 3.0, and when will we see it?"  Clearly the subject for a different post, but think "rich" --non-page-based -- Web applications and content available on non-PC devices via ubiquitous wireless broadband networks, with multi-modal human-device interaction -- keyboard, speech, etc.  We're seeing bits of this now, but it will be at least five years IMHO before we see good integration and wide adoption.  Eventually, I'll re-post this article, and change the 1 and 2 to 2 and 3.)

Comments

[dave]

good,It's good idea