In other words, by tracking what you and others do and create (emails, blog entries, tweets, browsing activity, shopping, etc.), and by scouring the web and analyzing its content, these systems make sense of the web in a much more sophisticated way than the brute force kind of searching that Google does. So it could find correlations, generate connections, optimize searches, make you aware of implicit networks of interest, and generally act on your behalf to both filter the incoming avalanche of data, and provide better/faster means to get to interesting information that you might not otherwise find.

While this idea is related to the kinds of recommendations that Amazon and other sites do, it is stronger because it aggregates a lot more activity and content beyond the silo of a single site. Plus, the ultimate expression of the implicit web (I hope) is that the user will have more control, and can “dial-in” the criteria of a search or automated task to their specific interests at that moment, rather than being stuck with some company’s idea of your interests. This idea relates to my essay on Productive Interaction, where the design of these systems is not about creating enveloping, persuasive experiences (as experience design dictates), but designing contexts where users are empowered to create their own meaning spaces.

Related LINKS below

Some posts by an investor in implicit web companies:
http://www.feld.com/wp/archives/2009/04/the-maturing-of-the-implicit-web.html
http://www.foundrygroup.com/blog/archives/2008/03/theme-implicit-web.php

ReadWriteWeb’s take on it:
http://www.readwriteweb.com/archives/the_implicit_web_lastfm_amazon_google.php

More sophisticated, semantic based tagging

http://www.readwriteweb.com/archives/common_tag_brings_standards_to_metadata.php

Some companies doing it:
http://getglue.com/brief.php
http://www.oneriot.com/
http://www.lijit.com/
http://www.orch8.net/

From the Feld.com post:

We think of the technologies that fall under the implicit web theme as a next-generation set of applications, tools and infrastructure that stitch together a long list of interrelated and overlapping ideas: the academic and theoretical ideas behind the Semantic Web, the utility of social networks and social media, crowd sourcing/wisdom-of-crowds, folksonomy, user attention data, advanced search and content analysis tools, lifestream analysis and numerous others. 

When combined, these technologies offer the promise of a more unified computing environment that spans the applications where a user consumes and creates information (email clients, web browsers, RSS readers, etc) and is aware of the user’s preferences, interests and interpersonal relationships without requiring a ton of heavy lifting on the user’s part to get useful work done.

I think the way to go is to cast it in terms of designing behavior.  There are many principles and concepts of designing interesting, rich, meaningful behavior that I think could be developed, some of which is instantiated in code, other aspects in the mechanical design (the turning of a doorknob or the page of a book for example), and others in the conceptual design.  This shift to behavior design as an overarching concept that encompasses computation may make it more interesting and relevant to designers.