Internet-mediated social networks can enhance the role of relationships in business decision-making.

Seth Grimes, Contributor

April 2, 2004

6 Min Read

Orkut and Tribe? They're warmed over, closed versions of Usenet, the global network-news conferencing system. Orkut and Tribe networks center on discussion forums rather than on networks of interlinked individuals. You're no more connected to another forum participant than you are to someone whose syndicated blog you read. Orkut nonetheless has buzz because a Google staffer started it as a side-project and you have to be invited by a member to join. But digerati exclusivity isn't enough.

Here's how Ben Bederson, a University of Maryland computer scientist and director of the university's Human-Computer Interaction Laboratory, put it: Orkut has "not enough support for real tasks. There isn't much I can do with Orkut. Why can't I combine Orkut with my 1,000 [Microsoft] Outlook contacts, birthday lists, student groups, etc.? Why can't I search my friends of friends for keywords or find others that are 'similar' to me? I'm not even sure what else I want to do with Orkut, but at the moment, there isn't much." I suppose that bulk contact-imports and free searches would conflict with Orkut's exclusivity. Oh well.

And the Winner is...

It makes no sense to crown winners in a race that's barely begun, but I can say that LinkedIn is the only social-networking system I looked at that currently deserves enterprise consideration. The company appears to be taking a careful approach that emphasized network quality. The LinkedIn system lets you import your electronic address book to search for contacts already on the system and to issue invitations, which you can also do manually. Links are purely intentional, and unlike other systems, both parties must agree to link. This ensures that connections are real and current. You can search your network, which includes friends-of-friends up to four links away and people open to direct requests, based on locations, industry, company, job title, and profile.

You can send requests to interesting people you find within your network. The requests are transmitted via the intermediate links in the chain, who can choose not to forward the request. My only complaints about this mechanism are that it can be slow and the middle person in a four-link chain serves as a gatekeeper without knowing the request issuer or target.

I also see where decision techniques could be applied to improve both a user's understanding of the network and also the network's behind-the-scenes routing algorithms. For example, while you can "endorse" other members, you can't numerically weight your connections to reflect the closeness of relationships, which must surely vary widely, especially for those users who have hundreds of links. I received several link requests myself from people with whom I have no relationship but I have no way of knowing if others have accepted such low-grade links.

It would also be nice if a valuation could be applied to users within the system — the network nodes — to account for reputation and authority, whether assessed manually or through data mining. I dare say that most users would value higher a request gated by, say, DBMS pioneer Michael Stonebraker than one that came through me. I'd infer that lacking node weighting, LinkedIn exploits only the simplest request-routing algorithms. Pilot Software's Jonathan Becher agreed with my view, saying, "There are lots of ideas from network flow analysis and Web site analytics that would apply here."

Despite my assessment of Spoke results, I do plan to keep my eye on Spoke because it has made an effort to integrate with office environments and to apply more advanced algorithms to network analysis routing. The quality of its results, however, is simply too low. I arrived at this conclusion by trying real-world tests, such as searching on the name of a former employer, a dot-com that folded in 2001. I got 10 results, only one of which was current: myself, a Spoke user.

Andy Halliday, Spoke vice president for business development, kindly ran the same test himself — he's clearly using the system to its fullest possibilities — and got the same results, which he spun by saying that "one of the interesting outcomes of collecting data from email records is that you get lots of historical data that is timestamped and some that is just out-of-date, but all of that 'bad' data accumulates to some interesting history on people." He was unable to explain why he and I saw the worse-than-worthless results given that I hadn't uploaded any email myself. In addition to the good hit for me, there was another that identified someone who supposedly knows me. I contacted that person, whose name I didn't recognize, thinking that maybe he has me listed in a press database. Not even that close: He speculated that he got my address in a rented list some place. On the plus side, I do accept Halliday's statement that Spoke would prove useful as a sort-of network-mining tool.

None of the systems I looked at go beyond friend-of-a-friend connections. I would like to see a commercial system that would seek out profile- and interest-based matches and generate referrals rather than simply passively facilitating manual requests. Such a system would be a Yenta — Yiddish for busybody — as described in a 1997 paper by Leonard Foner of the MIT Media Lab. Currently available systems provide a start with much unrealized potential.

The Network Effect

LinkedIn marketing vice president Konstantin Guericke wrote me in March 2004 that his system "currently has 290,000 active users who have uploaded over 11 million contacts from their address books. In mid-November [2003]... we had 40,000 users." I'd expect that other systems have had similar growth curves, adding to social networking's value due to the "network effect" postulated by Ethernet inventor Robert Metcalfe: The usefulness of a network is in proportion to the square of the number of members.

There will inevitably be standardization and a shakeout but also interoperability among the systems, just as email function and content are mostly independent of the mail programs being used, and just as instant-messaging systems should soon interoperate. We should soon see acquisitions as CRM and human resources system vendors begin looking to include social-network hooks directly into their systems. The net result will be a new valuation of relationships in enterprise decision-making.

Seth Grimes [[email protected]] is a principal of Alta Plana Corp., a Washington, D.C.-based consultancy specializing in business analytics and demographic, marketing, and economic statistics.


"Recommending Collaboration With Social Networks: A Comparative Evaluation":

"Yenta: A Multi-Agent, Referral-Based Matchmaking System":



Pilot Software:


Spoke Software:

Tribe Networks:

About the Author(s)

Seth Grimes


Seth Grimes is an analytics strategy consultant with Alta Plana and organizes the Sentiment Analysis Symposium. Follow him on Twitter at @sethgrimes

Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.

You May Also Like

More Insights