[fr] Après LIFT l'an dernier, un questionnaire a été soumis au participants dans le but de déterminer quel impact la conférence avait eu sur leur réseau. J'y ai répondu, avec 27 autres personnes (un assez petit échantillon, à mon avis). Il se trouve que je suis la "super-réseauteuse" de l'étude. Quelques remarques.
Eleven months ago, I participated and encouraged you to participate in a survey which aimed to map social networking between participants of the LIFT’07 conference. As I was browsing around after submitting my workshop proposal, I saw that the report based on that survey had been published. On the LIFT site, you can see screenshots of the graphs (yes, this is what I call a “social graph”!) before and after the conference.
Go and look.
Notice the node somewhat to the left, that seems to be connected to a whole bunch of people? Yeah, that’s me. I’m “lifter 20”. How do I know? Well, not hard to guess — I have a rather atypical profile compared to the other people who took the survey.
So, as the “star” networker in this story, I do have a few thoughts/comments on some of the conclusions drawn from the survey. Don’t get me wrong — I think it’s very interesting, and that we need this kind of research (and more of it!) but as Glenn says himself in the 1Mb PDF report, it’s important to bear in mind the limitations of this study. (All the quotes in this blog post are taken form the PDF, unless I say otherwise.)
The limitations of this study needs to be understood before considering the findings: This
study maps networks from the point of view of the 28 participants. Consequently, it is
only a partial map of the networks established at LIFT07.
In this study, I’m the “star” networker: the person with the most connections before and after the conference.
Before the conference, participant Lifter20 had the largest network (59 attendees)
which was increased by 25 attendees after the conference.
Bearing that in mind, I would personally have removed myself from the “average” calculations (I don’t think that was done), because I’m too a-typical compared to the other people in the survey. Typically, I would find it interesting to be given figures with extremes removed here:
There was a large range in the size of the individual networks before LIFT07 (from 0 to
59) and a smaller range in the number of people added to networks after the conference
(from 0 to 28). However, on average, participants had seven people in their network
before LIFT07 and added nine more people after the conference – leading to the
conclusion that people at least doubled their network by attending LIFT07.
As mentioned earlier, 28 people took the survey. I know I’m not the most networked person at LIFT. In my “network of red nodes” (people not in the survey) there are people like Robert Scoble, Stowe Boyd, or Laurent Haug — who clearly did not take the survey, or I wouldn’t be the “star networker” here. So, they are a little red node somewhere in the graph. Which makes me take the following remark with a big grain of salt:
Before the conference, several “red” attendees (i.e. those attendees nominated as
part of the network of the 28 participants) were significant relay nodes in the network
receiving considerable incoming links – notably the red node to the right of Lifter 12
and the red node to the left of Lifter 16. In both cases, the number of links to these
nodes increased after the conference.
What’s missing here is that these red nodes might very well be super networkers like Stowe or Robert. The fact they receive significant incoming links would then take a different meaning: only a very small part of their role in the global LIFT networking ecosystem is visible. (Yes, the study here only talks about a small part of this ecosystem, but it’s worth repeating.)
I think that most heavy networkers are not very likely to fill in such a survey. The more people you know, the more time it takes. I’m easily a bit obsessive, and I think this kind of study is really interesting, so I took the trouble to do it — but I’m sure many people with a smaller network than mine didn’t even consider doing it because it’s “too much work”. I suspect participation in such a survey is skewed towards people with smaller networks (“sure, I just know 5-10 people, I’ll quickly fill it in”).
Here’s a comment about the ratio of new contacts made during LIFT’07:
For example, the “star” networker, Lifter20 has a ratio of 1:0.4. In
other words, for every third person in her existing network, she met one new person.
Whereas, Lifter18 had the highest ratio of 1:7. In other words, for every person in her
existing network, she met seven new people.
I think it’s important to note that, as I said in my previous post about this experiment, knowing many people from the LIFT community beforehand, the increase in my network (proportionally) was bound to be less impressive, than, say, when I came to LIFT’06 two years ago (I basically knew 3 people before going: Anne Dominique, Laurent, Marc-Olivier — and maybe Roberto… and walked out with a ton of new people). I’m sure Dunbar’s number kicks in somewhere too, and I would expect that the more people you know initially, the lower your ratio of new contacts should be.
On page 8 of the survey there is a list of participants and the number of before/after contacts they entered in the survey. So, if you took the survey and have a rough idea of how many people you knew before LIFT, and how many you met there, you should be able to identify who you are.
This is interesting:
The “star” networker, Lifter 20 had seven links to other participants before LIFT07
which grew to ten after the conference, giving her the most central position in the
network of participants.
So, basically, 10 people I know took the survey — out of 28 total. I know I blogged about the survey and actively encouraged people in my network to take it. This would skew the sample, of course, making it closer to “my network at LIFT”. If we know each other and you took the survey, can you identify which number you are? it would be interesting to put faces on the numbers to interpret the data (for me, in any case, as I know the people). For example, if you’re a person I brought to LIFT, chances are your “new connections” will overlap mine quite a bit — more than if you came to LIFT independently.
A chapter of the report is devoted to the “star” networker (in other words, little me).
Interestingly, many of the
people that she connected to, both before and after LIFT07, were not part of the
networks of the other 27 participants of the study, indicating a certain isolation of parts of
Before the conference, a significant number of contacts (35) of Lifter20 had no
connections with any of the other 27 participants of the study.
After the conference, a number of contacts (14) made by Lifter20 had no connections
with any of the other 27 participants of the study.
The first remark be turned the other way: maybe all these “unconnected” people are actually quite connected within the “global LIFT network”, and it is the sample of 28 people who answered the survey which have isolated networks. Of course, isolation is a relative notion, but the way things are phrased here makes it look like I have an isolated network… which I don’t really believe to be the case — a great part of my network is actually very interconnected, only it doesn’t show in the graph because the people in question did not take the survey. My friend wheel (see screenshot) from Facebook gives a better impression of what it looks like. (No, no, I’m not taking this personally! I’m not.)
Lifter20 shares a number of contacts with one other participant (Lifter13 – the blue
node horizontally to the right in the “after” diagram).
Who is Lifter 13? (14 before, met 7 at LIFT’07) Somebody I knew before LIFT’07. I’m curious.
I’d also love to know who Lifter 18 (the “booster” networker) and Lifter 11 (the “clique” networker) were, though the graph indicates I know neither.
In conclusion, I’d say this is a really interesting study, but the anonymized data would gain to be interpreted in the light of who the actual people were and what their networks were like. I think it would allow to evaluate where this kind of analysis works well and works less well.
I think 28 people is a rather small sample for such a study — it’s a pity more people didn’t participate in the survey. How could we motivate people to participate? I think one of the issues, mainly, is that people don’t get anything directly out of participating. So… maybe some goodie incentive for doing it, next time? Also, I remember the interface was a bit raw. What I did is go through the participant list and type the names. It’s almost impossible to just think back at “so, who did I meet at LIFT this year?” — either you’re going to take a stack of business cards your brought home, or you’re going to go through a list and see what names ring a bell.
Maybe the survey organisation could take that into account. Provide participants in the survey with a (searchable, ajaxy) list of attendees with checkboxes. Then you could add smart stuff to help out like Dopplr’s “travellers you may know” (based on a “contacts of your contacts” algorithm).