Welcome to the inaugural post to this e-newsletter and blog.  For my background, please read my profile on LinkedIn.  (In brief, Microsoft, Oracle, Samsung, start-ups, industry analyst; business development positions, although spanning strategic planning, corporate and product marketing, new business development and global account management.)

[This is a letter to a good friend about Twine features and usability.  It implicitly describes why Twine is better than any other social network, although each existing social network could try a Twine-like play since it's unlikely the core technology can be patented and the application is really quite obvious -- at least to anyone in the AI sector who has used a social network.  This post is slightly modified from its original post as one of my AlwaysOn Network "Letter from China" network blogger posts/columns.]

I was thinking about our conversation on Wednesday about Twine.  You made the point that you're most interested in finding people who are not necessarily like you, but complementary to you.  Actually, none of the social networking systems are very good at this.  To really make this work would require someone to enter a helluva lot more info than they're likely willing to do and/or to give up a lot of privacy to permit the system to track their interactions with their connections (although this should be doable).  For example, LinkedIn could see who you are connecting to, notice that it's complementary, and then suggest other complementary individuals.  Lots of AI, I suspect, to make this happen.  In theory, genetic algorithms would be better, but each person would get a lot of false contacts for the first several tries while the system learns about them.  On the upside, users would have to input less data about themselves (although the more data they input, the better the matches).  OTOH, with an expert system, the only way to make it work is if a lot of data is inputted by a critical mass of potential matches.  (As a general rule, expert systems, either rule- or knowledge-based, are not as adaptive as genetic algorithms or neural networks.)

But let me point out where Twine could have a lot of value, something I didn't mention on Wednesday:  Partnerships and alliances.  Although what you may want to find is somebody who really should be a 2nd degree connection in your expanded network (i.e., somebody who is complementary to you), a lot of times people are looking for partners -- and this would likely be a perfect match 1st degree connection, i.e., somebody who would be a match on Twine.

Obviously, people have to be careful about divulging information to direct competitors (a likely match on Twine; on the upside, this might also lead to potential employment opportunities).  But it should also be possible (and very likely) that you'd find potential partners, especially since we're in a geographically desirable country [note to readers: I'm referring to China], but a country that is also a bit detached in many ways from innovation and emerging technologies.

I could see Twine also having a lot of online dating value, playing to points that we've discussed.  If somebody wants to find someone they can really, truly converse with, well, Twine could be an excellent way to do this, to find a great match.  Matter of fact, Twine combined with eHarmony would be an extremely powerful combination, the only issue (and it's not a small one to overcome) is getting critical mass in a combined system. 

What we don't know just yet is if Twine is easy to use.  What I would prefer is a digital library like Furl, but something that can automatically grab my bookmarks (think Foxmarks or Google Bookmarks) and THEN upload the data to Furl.  From there, the Twine-type engine could do its stuff, using semantic processing for matchmaking.  Matter of fact, just have a bookmarklet (to make it really easy) to make a certain piece of content Twine friendly, eventually also enabling the input of documents, like through Scribd.  (A bookmarklet-equivalent for OpenOffice/IBM Lotus Symphony and Microsoft Office, I guess, as well as for Outlook and Gmail -- if necessary, added as a Better Gmail or CustomizeGoogle feature -- would be best.)  [Note: Among the three options, IBM Lotus Symphony, OpenOffice or Microsoft Office, IBM has the best suite -- and it's free!!]

The point is to make it easy, EXTREMELY EASY, for users to upload or reference content.  They should be able to do so with the simplicity of a bookmarklet and WITHOUT the need to wait as with Furl.  It should be pretty much instantaneous.  EXTREMELY FAST.  If a bookmarklet that uploads content real-time is too slow (and it might be), I'd use the bookmarket and then have a Foxmarks-type syncing function combined with a Furl-type uploading of bookmarked content -- and then let Twine do its thing.

To the user, this is very, very easy.  See something you like, just click on the Twine bookmarklet.  Matches will follow.  Good for finding potential business partners, life partners, and employers.  In comparison, Facebook looks a lot more like Plaxo: Perhaps nice for managing contacts, but not nearly as effective for finding new, desirable contacts.

Based and living in China for the past 4 years (both in Beijing and Qingdao), David Scott Lewis is SVP with Startech Global Corporation, the outsourcing hub for Tsinghua University (China's MIT and Hu Jintao's alma mater) and Zytech Solar, a Going Green 100 winner.  Lewis attended his first IJCAI conference nearly 30 years ago and chaired the session on web-based agents during the First International Conference on Autonomous Agents in 1996.

AO Tags:   twine,  facebook, myspace, furl, foxmarks, google bookmarks, linkedin, plaxo, eharmony, social networks, social networking, semantic web , artificial intelligence, neural networks , genetic algorithms, Alliances, partners, partnerships

--
David Scott Lewis, Editor & Publisher
Interactions 2.0: Beyond the User Interface & Mere Communications
Beijing & Qingdao, China
LinkedIn: http://www.linkedin.com/in/davidscottlewis

Blog: http://interactions20.blogspot.com/
RSS/News Feed: http://feeds.feedburner.com/interactions20
E-Newsletter: http://www.interactions20.com/