Mark Granovetter's seminal paper, The Strength of Weak Ties (summary), revealed the difference between friends and acquaintances and how useful acquaintances can be for certain tasks like finding a job. The difference between a strong tie and weak tie can generally be revealed by time commitment underpinning the relationship. Strong ties are better for action, weak ties for new information.
But time has changed with new tools and social networking models at our disposal. For the first time many social networks are being made explicit, often without the knowledge of participants, at an accelerating pace and dramatically lowered search costs. This newfound transparency may very well make strong ties weaker.
Top-down vs. Bottom-up Social Networking
Top-down social networking models such as Spoke, Plaxo and Visible Path attempt to capture existing social networks through bulk processing of contacts and information flow. They circumvent what Valdis Krebs called the Achilles Heel of social networking data entry. A company or person joins the network by submitting contacts and/or allowing information flow to be monitored. Traditional social network analysis relied upon surveys and interviews. Analysis has already benefited from the computational and visualization advances afforded by PCs, but now the Internet has rapidly decreased the transaction costs for performing analyses as well as allowing dynamic analysis that is closer to actual current state of the network.
Top-down models, or weak tie systems, are particularly good at the breadth of sample and search. Most are focused on the business opportunity of intelligence. Enabling a sales team to navigate a complex organizational sale to an enterprise by identifying buyers, influencers and potential paths to reach them. This newfound business intelligence can shorten sales cycles.
But several problems are obvious which make me doubt the sustainability of the top-down approach. First, salespeople perceive their value is in their Rolodex and how they use it. Its a personal value that they lease to a company during the period of employment. The company benefits from the relationships they gained in previous jobs, they expect to gain more and offer this increasing value with their next gig. Demanding your salespeople upload their contacts is antithetical to the existing social contract and creates disincentives for accurate data entry. Even without consideration of privacy issues, the difference between personal and organizational opt-in is significant.
Second, the objective of keeping data current for dynamic analysis has led some models to burden people who dont benefit from the system to perform data entry. We have all been Plaxoed by now in the name of virality. If the transaction costs for updating contact information were lower or if non-participants gained value we would all be Plaxo users. But we are not -- stop spamming us.
Third, as people become aware of the intelligence that may be driving people to make contact with them they become skeptical. When you get a B2B sales call from John saying he was referred to you by Jane, even if you email with Jane daily, it begs the question of how strong the relationship between John and Jane is. Janes social capital with you has been risked by Johns action. New transaction costs are pushed to the receiver of the call to fact check existing trust. The existing tie may rest upon one of many facets of a relationship, may be inappropriate for the kind of action being taken and the transaction has the potential to make existing strong ties weaker.
That said, weak tie systems provide real insight we didnt have before. They have vast potential for revealing structural holes and other methods of enhancing social capital. So long as participation is voluntary, information flows monitored is public and constraints keep in check how people act upon intelligence these systems will become a critical competitive advantage for any company.
By contrast, bottom-up systems are built for people instead of companies. Within Friendster, Ryze, Tribe.net and LinkedIn connections are made through individual decisions upon of invitations, requests and confirmations. The network grows organically, decision by voluntary decision. I have written at length about the differences between these bottom-up systems and their relative utility. What they lose in network growth until their viral properties take hold, they gain in utility, culture and participation.
The key difference is bottom-up systems do not have the breadth of sample, yet, and generally have less powerful search and analysis capability. But the information they provide is decidedly actionable.
LinkedIn, for example, constrains connections to stronger ties. When you make a connection you offer the other person an option to act through you and your network to act through you to reach the new connection. This is an obligation of time and a risk to your social capital as an up-front constraint. As the requirements for what is a connection increases, connections diminish, as Duncan Watts would say. What remains are actionable connections -- stronger ties are more likely to pass information and do so more rapidly than weak ties. Referrals, a way of taking action, pass through a chain of stronger ties.
Top-down and bottom-up social networking models are actually ideal complements. A sales organization could benefit from the search and intelligence a top-down system provides on who to sell what. How to sell, or taking action, is better performed through a bottom-up system -- with voluntary and confirmed ties meant for this purpose. That is, until bottom-up systems become grow in breadth.
Top-down and bottom-up models dont directly compete today, but have competitive unintended consequences. Bottom-up models are demonstrating how networks are personal. Top-down networks are devaluing ties because involuntary transparency is subject to abuse.
The Great Iteration
When I attended a talk by Granovetter last week I sought an answer to a single question, how does this newfound transparency change behavior? The answer is its unknown how contamination effects behavior.
We can see the beginnings of new social patterns within bottom-up systems where people know what is transparent and how networking games enable different kinds of inter-personal connections. But it has largely gone unstudied.
We have yet to see the feedback loop complete itself for top-down models. How will people act when they recognize their actions are being observed to the advantage of some? Will people keep two address books and email accounts to game the system?
For all of these models, now that we have critical mass of participation, you can be confident that new patterns will emerge and systems will have to iterate in compensation.