An experiment with online social networking

by Andy Oram

A couple people have told me that the hot new VC quarry in the Silicon
Valley is social networking--a whole herd of new services such as
and a number of major companies soon to join them. One friend of mine,
after complaining about the uselessness of social networks, begged me
to invite him into one--because he's doing research to help his a
company he consults with develop a competing social network.

The experiment

Reluctant to add more information overload to my creaking virtual
shelf of resources, I joined a couple social networks some weeks ago
after being pushed by various colleagues, including my boss's boss. I
wrote up my

initial impressions

but didn't feel I had really explored the medium. So I decided to try
a formal project that would acquaint me with the medium's strengths
and weaknesses, and give a first answer to the question of whether a
social network could help me where my usual channels of getting
information could not. Since I'm married and have a job and friends
and leisure activities, the test I chose was to find someone in an
established discipline who could direct me to research in that
discipline, one which I knew little about.

The subject was reading--in particular, whether twentieth-century
media such as radio, television, and the Internet have changed
people's expectations when reading and their ability to absorb
information from books. (I'll probably get more useful responses from
readers of this blog than from anything I've done up to now--but I'm
giving away the ending to my story.) The information was not
arbitrarily chosen for the sake of experiment; I really need it for a
presentation I'm writing on trends in technical documentation, to be
given at the upcoming

O'Reilly Open Source Convention

I had already tried my usual ways of finding information: web
searches, a relative in the field of education, a contact I had
developed from an article I wrote about

Marshall McLuhan
a mailing list he pointed me to, and a couple similar gambits. None of
them helped me, although I got encouragement that information on this
subject did indeed exist. So I turned to searching my social networks.

One of the most popular networks proved right away to be unsuitable;
it allowed searches only on a few characteristics such as whether
someone is married. I assume more flexibility will be added in the
future, but for the moment I can use the network only as yet an extra
(unneeded) mail service, one where most of the messages are in
languages I understand only at the pidgin level.

Another network had a sophisticated search system that impressed
me. It was based on metainformation such as geographic location and
industry, which users are encouraged to fill out when they join the
network. Here is an example of the benefits that accrue when people
take the time to provide metainformation. The buzz over social
networks has apparently seduced them into taking this time, but
they're not likely to do it often. Metainformation would help me find
documents these people have written, but they're not likely to
attached metainformation to their documents. In fact, they're not
likely to update the metainformation they already have here when
changes take place in their lives.

So I did two searches, setting the industry to "Higher Education" and
sorting search results by "Keyword Relevance." I left other
metainformation as general as possible so I could search for
anybody. First I sorted for the term "reading," then for "cognition,"
and finally for "cognitive." The last term proved most fruitful,
judging from the brief descriptions I read of the people it turned up.
I thought the search and sort was intelligent. The top choice that the
search presented to me in terms of keyword relevance was also my top
choice based on the descriptions--and ultimately the person I
corresponded with.

Now for the downside. My top choice was more than four degrees away
from me. That meant I had to contact her cold, just as if I had found
her name in a journal or on a web site. That's fine with me, but it
eliminates much of the point of being in a social network.

On the other hand, another of my choices was three degrees away, which
mean I could not contact him cold even if I wanted to, but
had to go through the two intermediate people. This effectively meant
introducing a delay of several days (and bothering the intermediate
people), which eliminates much of the value of online connectivity.

As it turns out, the network itself introduced an artificial delay
into my direct request to my top choice. So the response from her (she
approved my request for contact) took four days, only one day less
than the response from the person who was contacted through two
intermediate degrees. (The latter person rejected the contact because
he did not have the information I wanted--in other words, I guessed
wrong from the descriptions he provided on the network.) A third
request I made is to someone four degrees away and is still listed as
"en route" after ten days.

So how did I do? My top contact provided me with detailed names,
university departments, and terms to search for. So I think my use of
the social network was worthwhile. So far, I have not found the level
of (possibly simplistic) detail I was hoping for; the people she
mentioned are into abstract concepts like "computational modeling" or
at best "sentence comprehension." I didn't expect that I'd have to get
an advanced degree and learn to do my own experiments in order to
develop some general insights for my talk. But I can't say I didn't
get what I asked for.

Some conclusions

I'm not taking myself off the social networks yet; I'm intrigued by
what I achieved through this one experiment and would like to see how
they develop. But I have a few useful observations based on my small
experiences so far.

The artificial concept of "degrees" actually puts barriers in the way.

For most tasks, I'm more interested in what people know than whom they
know. But the primary criterion by which social networks rate
people--and really their whole raison d'être--is the distance
between people socially. If I have to contact intermediate people to
reach someone I'm interested in, I add a lot of extra time to the
contact while straining my relationships by imposing burdens that
offer my close contacts no rewards in return.

Anything over the second degree is equivalent to anonymity.

I might care whether Bill knows Bruce, should I care to get know
Bruce. But I don't care whether Bill knows Bruce who knows Betty. At
that point, I have no way of judging the value of a personal
connection. The premise of the social network is like that of a
country club that requires endorsements for people to join: when
someone you trust marks someone else as trustworthy, it's supposed to
influence your behavior. But the connection gets attenuated very
quickly. The main difference between recommending members to an
exclusive country club and having friends on a social network is that,
when you recommend a friend for a club, you understand clearly such
things as:

  • what kind of activities go on in the club

  • what criteria make a good club member

  • what rewards flow from being in the club

  • how important it is to be part of the club

  • how important it is to keep inappropriate members out of the club

and many other subtle social aspects of the club. All of these
criteria are missing from social networks; they're too broadly
defined. I'll repeat what I said in my
earlier article:

I know a lot of people because they're interesting for one specific
reason or another, and none of that means they're fit for some
particular task that somebody else has defined.

How do you say that something relevant to one relationship is relevant
to a different relationship with a different person?

For what I needed in this experiment, search was more important than
social networking. I think the goal of the social networking services
is to become more than search services, and to turn into true
communities. (In fact, one offers a service called "communities,"
which appear to be much more than newsgroups.) I don't know whether
that will work, what the services need to make it work, and whether
the communities are enhanced by the "degrees" model of social

Let me end by suggesting an exciting new networking opportunity that
is not online and is very direct:
Beyond The Soundbite.
This political discussion was started by a relative of mine who was
upset by the inability of most people to find good sources of
information and good fora for discussing keys issues of our day.
Examples of such issues from Beyond The Soundbite's home page

  • Is "No Child Left Behind" good for education?

  • Should the USA adopt protectionist measures to save jobs?

  • Has the new Medicare bill made the crisis over health care better
    or worse?

  • How important is the deficit?

Beyond The Soundbite is intensely local and personal. The organization
brings in speakers and conducts roundtable discussions. The goal is to
build up trust among people with different political backgrounds and
keep them coming back over and over so they can deepen their
understanding of each other's views. This is an old-fashioned kind of
community-building, quixotic perhaps in an age where people don't even
walk to the edge of the block their house is on. But I think we need
it and I hope it spreads.

What are your results with online social networks?


2004-04-05 19:29:02
Missing the point...
I'm glad your experiment was a qualified success, Andy, but I think you're missing a a couple of key points...

First of all, regarding your top choice being more than four degrees away, how many confirmed first-degree contacts do you have in that system? 13? And who are they? Are any of them super-nodes, like Thomas Power, Joi Ito, or Reid Hoffman?

Just as the profile information is only as good as the metainformation people put in, so, too are the relationships. I have yet to actually be four degrees away from anyone -- often I discover that someone along the path is someone I already know, and just don't have a record of it in that system.

Furthermore, data collected by LinkedIn shows that four-degree introductions are successful more often than you would think, and three-degree introductions are frequently successful. I even have a story of a VC who got his current position at Ariadne Capital in London through a four-degree introduction -- trans-Atlantic, no less!
Trust is transferable, if it's actually communicated along the chain, and the intros aren't just passed along without personal comment.

Regarding "the artificial concept of degrees"...

People participate in this networks for different reasons. Those who are looking to increase their visibility and who are highly receptive to inquiries are perhaps better off in a more open environment like Ryze or Tribe. LinkedIn is better suited to those people who use their network more as a collaborative filter to make sure that only the interesting and worthwhile requests actually make it to them. The cost of that is slower communication, but sometimes it helps you get through to people who might be less accessible otherwise.

That's probably not a situation you yourself would often face as a Senior Editor at O'Reilly, which carries some weight, but for the typical person, stepping through those hoops may give them access to people who would be inaccessible to them directly.

Glad to see you're giving it some more time, though.

Scott Allen
Online Business Networks Blog

2004-04-06 04:39:51
Good Grief!
Andy, I understand. But, Scott what on earth are you talking about? "confirmed first-degree contact", "super-nodes", what's all that about? And let's be honest about it: the whole concept of Metainformation, as good as it is in theory, is problematic in practice for all too human reasons (Metacrap) All this social networking palaver is reminiscent of the online behavioral modeling and marketing bollocks of the .com excesses - an idea in search of a market rather that a product to fill a need or provide an advantage. What are the metrics? How does one measure the benefits of a social networking product? What can I do better? cheaper? faster?
Degrees of separation seems to me a kind of fuzzy measure of a relationship. Is the one degree which exists between me and my brother equivalent to the one degree between me and my ex-wife? I wouldn't think so.
Perhaps the products will be sucessful within pre-existing real communities such as business enterprises, industry groups etc. At least that might provide a common context for the relationships and allow greater consensus regarding the value of those relationships.
2004-04-10 19:18:24
Another view
Andy, an article I wrote on social network based on my own experience is recently out: .

I had some of the same criticisms of "leading" systems as you plus a number more.