In a massive follow-up on the famous Six Degrees-of-Separation Stanley Milgram experiment by Columbia sociologists Peter Sheridan Dodds, Duncan Watts and their colleagues have published the first results of the Small World Project, and their findings do not bode well for the social schmooze-fest networks such as Linked-In and Ecademy.

the problem of throughput

While Dodds, Watts et al do confirm the close connectedness of social networks, in the context of leveraging social networks to forge new connections, they also find a 97% attrition rate in the lines of communications: You may connect with your first Degree-of-Separation band of associates, but only a few will relay your request to their band, and very few those second-order associates will comply, and so forth and so forth -- the throughput of the network is virtually nil.
This agrees with my experience in online social networks engendered by Ecademy, Ryze, Linked In and even just the everyday interactions I have through mailing lists, emails and, yes, this blog: The local circle of first-order associates is attentive and thoughtful, but just as Dodds and Watts report, given a request for a connection, those few (1%!) who do respond will say they broke the chain because they "could not think of anyone to send the e-mail message to" -- when first-order associates know (or subsequently meet) someone who may be interested in my work, they often recommend me but there's a problem of transmission when the members of that first-order group believe they must selectively retransmit my request.

Therein a flaw with the Linked-In's: Circumstances would need to be remarkable before I would burden the whole of my network to push out a received request, and thus, while you may find that, once a connection is made, the degree of separation is under 6 steps, the probability of connecting to that same someone is astronomically small.

This is maybe another advance with blogging: the majority of those thousand people reading this page today did not arrive here because I invited them or even because someone I know specifically invited them, but because some other blog listed me in their blogroll, cited me in their post, and very often even because Google or Blogdex picked up some relevant keyword and returned this link in its results.

Introduction by blogging is passive, initiated not by my effort to meet you, but by your curiosity to meet me. It may be a World of Ends but it is a world of asymmetrical flow, it's a pull process, driven by the consumer, not by the provider

Six Degrees experiments fail to garner net traction because they only try push processes; Watts admits a frequent excuse for transmission attrition was because their experimental message in a bottle was often mistaken for spam.

And ditto for trying to pump your message through the schmooze-fest sites. Attempts to pull "customers" out of a Ryze or Ecademy comes off like a troll. Outside of the focused and intentful discussion forums, the typical "feature item" posts are not 'communications' in any sense of dialogue, but very gently fishing for sales prospects. Informative, clever and sometimes timely, but still an albeit friendly sort of spam, cluttering the login page instead of cluttering your inbox.

And like spam, these posts may collect some infinitesimal percentage of "hits that convert to sales" (prompting Thomas Power to believe the secret is in "Power Networking", ie being as ubiquitous as possible) but those are only first-order effects and unlikely to see any viral second-order returns; while highly infectious viral memes do happen, by the Dodds/Watts attrition law, third or fourth order returns would be about as likely as scoring big in the national lotto.

send in the blogs

Here's maybe were blogs upset the whole applecart because blogs do in fact rebroadcast what comes in to an associate out to that associate's whole 'network' -- the tradition of attributions on cited items (even for relayed stories, we most often cite the blog where we found the link in addition to the content): A case in point is my post yesterday about the Bradbury Birthday Greeting -- the Planetary Society's message reached me by a second-order network aggregating my rolodex of science-related blogs, and now reaches my audience, some of whom may also relay the request; it's still a highly atrophic transmission because most of you reading my blog won't blog the item, but we do improve on the Six Degrees attenuation because those who do blog it relay to their entire network.

Back at the pull vs push notion, blogs are also more effective at building social networks through leveraging the passive side of the pull process. I may use active and person-to-person voice when I tell you I'm contract-trolling, but I didn't really tell you, you came in here and by so doing asked for that information. My role is passive, you called the shot.

Although it's a passive connection, blogs are also more effective at building the first-order network: My RSS casts my items far and wide through all sorts of aggregators -- Radio Aggregators shows up as the number one consumer of bits on my website -- other weblogs blogroll me (where my pings to weblogs.com will flag my content as fresh) and these pings are also fed into other services such as Feedster, Blogdex, Daypop, ... each further extending the breadth of this target I've left for your curiosity to find -- I've therefore found an outlet where I can actively increase my probability of passive reception of your kind attention.

This suggests a different approach before we apply the Dodds/Watts/Milgram methodology in the blog-network ecology. In the blog space, instead of asking each node to make the forwarding decision, we would invert the information flow and turn it instead into a scavenger hunt, and that, of course, is just a web-crawler, and has already been automated, and that suggests there's more to this game of connections than just making the link.

the return of the trust problem

You may find my work, you may read through it all and be very impressed and write me kind letters (as many of you do) even pop a note in my tipjar but ... will you risk your own reputation by recommending to your boss that they spend six figures to engage my services on their next k-blogging social software intranet project?

You don't have to answer that. The Dodds/Watts result already forgives you because you just don't know any boss to do the recommending to, but even if you did, you'd still have a problem of trust and reputation and how to translate this blog-proxied meeting of ours into a real live and trusted business relationship.

Where this does happen, apparently with regularity, is on Ecademy, but the success there is not from the blogs or the find me more people like this power networker links. Those links help break the ice, but the deal-clinchers are because of Ecademy hosts meet-ups.

Like it or not -- and ignoring whether or not we might someday change this -- big deals happen between the people who have made first-order social connections, and by people who have been introduced by trusted advisers.

For example, if the respected manager of a neighbouring department or corporate division were to tell you ...

"... [his] perspective is one I trust, as [he is] in the middle of it all and keeping pretty good tabs on what is going on ... I get lost trying to keep track of all of it all. Fortunately, I now have an expert to whom to turn!" Marilyn Cavill, Industry Canada

You are far more likely to risk your own reputation recommending me to your own boss.

Linked-In as a verifier

This suggests a role for schmooze-fest link sites not as a discovery network, but as trust verification systems, as an adjunct to the blog. While I can and do pepper my pitch pages with high-profile endorsements, it's still accolades from people you don't know, people you didn't actually hear making those songs of my praise. What's needed then from a Degrees of Separation networking service is not the means to relay a message out from any one participant, but only as a means to draw some personal connector to some participant, as a means to gauge their distance from those you already know.

Dave Pollard mentioned something like this recently in a post about how corporate k-blogs needed to retain the individuality of the person, to accommodate different styles, different models, different taxonomies of the individual while fostering aggregation, trust and communications between them to effect a knowledge transfer. K-blogging needs some means for you to know that I'm the one who knows what it is you need to know, and it needs to do that without confining both of us into some standard stylesheet communications mode.

kiss this guy

You see, here's the problem:

• May does reflexology and gives a pamphlet to Bob
• Bob's daughter's neighbour can't find a reflexologist

How often does it happen that you find out post-hoc, "Gee if only we'd known we could have made the connection ..."

We have tools to manage our own Rolodex, ways to exchange Vcards, format them, index them and retrieve them in a thousand ways, but the Six Degrees says we need to bridge our social networks across two to five separations, and we have no tools to resolve those interior layers between May's links and those of the neighbour's. So near and yet effectively infinite; it's not enough to find a bridge point between social bubbles, we often need to find up to 5 serially strategic bubble connectors to bridge the gap. We might induce Judas to kiss the Christ, but we first need the Paul we know to kiss a Matt who'll kiss a Thomas who'll kiss us a Judas, and we've no means for this in the present state of affairs save packing us all into some event, slapping identification name-tags on us and keeping us captive in a conference hall until we start talking.

In the tradition of other great ideas here's a solution as simple as a sidebar chicklet, only it has an awkward technology twist that's going to require some clever innovation: You place a marker on your blog page that will trace a path back from the blog page to the reader's network, a sort of Six Degrees Blogrolling that reads your blogrolling.com cookie (or some such) and then compares your blogroll to my blogroll and reports all the shortest paths.

That's one implementation. For another, also harking back to a Pollard post about mapping Instant Messaging, this Trust Me chicklet could cross-product the paths between IM buddy lists, or between aggregator inputs, or between VCard collections, or from any one (or all) of the ways we track our first-order networks to give a multi-dimensional matrix view of this author's trustworthiness.

It would work.

Trust me.