Hubs and Connectors: Understanding Networks Through Data Visualization

Yesterday a bunch of friends were tweeting about the new LinkedIn InMaps web app (part of LinkedIn Labs), so I had to check it out for myself.

Wow, wow, wow! InMaps are data visualizations of your professional network, based on LinkedIn connections, with you as the center node. I’ve been waiting years for LinkedIn to finally get its act together and start offering up analytics based on all the data we willingly store within it — but rarely make use of.

Here’s what my LinkedIn InMap looks like:

Immediately I started to explore all the hubs and nodes. You can easily see that Porter’s cluster theory is alive and well. The geographic, horizontal and vertical clusters in my network are clearly delineated, some more isolated than others.

After hovering over enough of the dots, I figured out what each cluster represents and was able to fill in the labels on the handy legend they provide:

So here’s my map with labels applied:

This amazing visualization has helped me confirm something I’ve always innately known about my network, and something I’ve always valued about myself: I’m a social floater. I’m a hub and a connector. I mingle in many different circles (both personally and professionally) and am the node between clusters. In more ways than one, this is the key to my professional success and something I actively cultivate.

Albert-László Barabási’s Linked is probably the best book you can read on the power of networks — it’s much more technical than Malcolm Gladwell’s The Tipping Point. In Barabási’s chapter “Hubs and Connectors,” he writes:

“Indeed, with links to an unusually large number of nodes, hubs create short paths between any two nodes in the system. Consequently, while the average separation between two randomly selected people on Earth is six, the distance between anybody and a connector is often only one or two.”

“When it comes to networks, size does not always matter…As networks are clustered, nodes that are linked only to nodes in their cluster could have a central role in that subculture or genre…The truly central position in networks is reserved for those nodes that are simultaneously part of many large clusters…They are the people who regularly come into contact with people from diverse fields and social strata.”

It’s safe to say that I’m a user experience evangelist. I spend a lot of time writing on this blog, Twittering, speaking at conferences, writing for publications — all on the topic of user experience; all in an effort to heighten awareness, understanding, and practice of the UX discipline in the greater tech and business communities. Barabási succinctly writes: “Hubs coordinate the communication between the many parallel functions.” I cherish my role as liaison and facilitator, and am excited to be able to explore these topics further.

I’ve played with my LinkedIn InMap for hours, and here are some intriguing things I uncovered:


The size of the node is an indication of the size of that person’s network. Therefore the bigger the dot, the more connections to it. The smallest dots on the map belong to my family members and non-tech-savvy friends who are each connected to fewer than 50 people. The biggest dots are hubs, which I’ll go into much greater detail on.

Distinction and Isolation

The most distinct clusters on my map are composed of my contacts from the three companies where I worked full-time (prior to going independent): Digitas, Tribal DDB, and Liquidnet. Digitas and Tribal DDB are both digital marketing agencies, so they are expectedly interconnected and thus close to one another on the map. On the other hand, Liquidnet is a financial software company — and the only finance job I’ve had. As a result, its cluster is highly isolated and distanced from the others. Its lack of connection to the rest of my network tells me that the finance industry is naturally exclusive, and highly specialized.

Clumpiness and Assortativity

Networks can be measured by their “clumpiness coefficient” and their “assortativity coefficient.” In laymen’s terms, these relate to the density and exclusivity, respectively, of each cluster. The two coefficients are combined to create four classes of networks, three of which I can observe on my map.

The “clumped assortative” areas of the network are UX and Liquidnet. The “loose assortative” areas of the network are Digitas and Tribal DDB, and CMU/HCI. Meanwhile the NY Tech Scenesters, Entrepreneurs, and Family/Friends/Misc belong to the “loose disassortative” network class.

These classifications were put forth in a paper titled “Clumpiness” Mixing in Complex Networks by Ernesto Estrada, Naomichi Hatano, and Amauri Gutierrez less than two years ago. It’s amazing to me just how much we have left to learn about the nature of networks in our rapidly-evolving society.

So, given the limited networks of my friends and family, as well as my varied, non-“community” relationship to them, it’s no surprise that their nodes are loosely clustered and non-exclusively connected to one another.

The Digitas/Tribal DDB relationship however is a classic example of high-density, highly-exclusive related networks; they’re defined both by employer and industry, which are both naturally assortative.

What intrigues me the most are the clusters for NY Tech Scenesters and Entrepreneurs. When my map was first created, I was actually confused as to why these networks were broken out into two, but after thinking about it, I realized that I do experience them very differently. Alternate names for these clusters could be social media and startups. Everyone is connected, but it’s a loose disassortative network because these people only share sector (technology) — not employer, function, or industry.

Another loose disassortative area in my larger network: my college friends who work in various industries and functions.

Hmm…come to think of it, UX isn’t clumped assortative, it’s loose assortative — part of the larger blob — because we share function and sector, but not employer or industry. But the lower area really does exhibit significant clumping. Maybe it’s some sort of hybrid. Let me chew on that.


I started to look at some of the more isolated nodes to see who they are. Remembering Barabási’s words, I realized that outliers are connectors. They either cross sectors (finance and technology), cross industry (marketing and internet), cross function (UX and programming), or cross employer.

The outliers that sit between Digitas and Tribal DDB? Of course those are people who have

The outliers connected to Liquidnet and angled towards the center of the map are in fact my fellow former members of the Liquidnet UX team (cross-sector). Meanwhile the outlier connected to Liquidnet and angled upwards was a dev contractor (cross sector and cross-employer).


As I was exploring, I noticed that there are a handful of nodes present in dense clusters that bear the color of another cluster. Interestingly, most of these nodes are of a smaller size, meaning that even though they don’t have large networks of their own, they are social butterflies who integrate themselves in multiple circles. A few of these are recruiters and publishers/editors.

Suddenly a giant infiltrator suddenly popped out at me — a pink dot in the zone between the dark blue cluster and light blue and light orange scatter:

Mr. David Armano. I totally should have guessed that.

David and I met at Digitas, my first job out of college, so that’s why he’s pink. But he has done a remarkable job of straddling the both the marketing and UX worlds, as well as being highly connected in New York because he grew up here and visits often.

When you click on a node in the map, their connections to your contacts are highlighted. So you can clearly see how David is a super connector between clusters:


There are different types of hubs, due to varying degrees of breadth of depth in each hub’s network. I’m still struggling with how to reconcile that an individual’s network map only shows the people to whom they are directly connected, therefore we cannot truly predict the shape of each contact’s own network — we can only represent their network in relation to our own.

Lou Rosenfeld is a localized hub, or in Gladwell’s terms, a “maven”:

So are Ali Khan, Shipra Kayan, and Consuelo Ruybal:

MJ Broadbent is a bilateral hub, or a “salesman”:

So are Eduardo Ortiz, Tony Bacigalupo, and Toby Vann:

Chris Pallé is a generalized hub, or a “connector”:

So are Jessica Valenzuela, See-Ming Lee, and Charlie O’Donnell:

I would guess that in other people’s network maps, I’m seen as a bilateral hub. Is that the case?

Gladwell would say that understanding the role each hub in your network plays can help to clarify its value to you, and identify what areas you still need to develop.

In Closing

There is a lot more exploration that I want to do here. I feel like I’ve barely scratched the surface. In the coming weeks, I’ll dig into other people’s maps, and compare their views to mine — particularly in an effort to begin to understand the relativity of these hub roles.

Any thoughts that you have to share on what I’ve written so far are welcome and wanted. And if you have any network theorists in your networks, please send them my way!

Related Posts:


  1. says

    Hey Whitney,

    Nice Writeup. Don’t forget your home grown network visualization business: TouchGraph. We’ve been doing the same thing with Facebook for a few years:

    The TouchGraph Facebook visualization ranks people by who is a connector, which puts emphasis on people that are part of multiple groups. The idea is that if you like people from different circles enough to invite them to a party, and they become friends with each other, they also become connectors between groups.

    The LinkedIn currently does not allow retrieving connections between friends:
    Once the do (hopefully) then TouchGraph can do this for LinkedIn as well.

    • says

      ah…. now that i think about it, I suspect my dot on your graph is probably just in the midst of the thickly intertwingled UX cloud. going now to look for you in my graph :D

  2. Kristine says

    This IS interesting. My dark orange group tends to be people that have been past clients or likely will be future clients. It’s interesting because some of the dark orange dots are people I had not ever considered to be potential clients before. Thanks for the in-depth analysis!

  3. says

    Have you read Ron Burt’s theory of Structural Holes? It’s a really thoughtful analysis of the power that people like the “salesman”, above, have in densely connected networks.

  4. says

    Nice to read your story.

    InMap made a mess out of my personal LinkedIn network though. Many people were interlinked in a too complicated way, so there was hardly any logic in the groups.

  5. Nurit says

    Great analysis. I am just in the process of designing a network map myself so right to the point.
    one comment – I think dot size is about the number of connections in your own map, not the overall number of connections of that person.

  6. says

    Hi Whitney,
    Nice post,
    Are you aware if it is possible to visualize the connections from GROUPS on LinkedIn?

    Although it is interesting to see “my individual network”, it would also be very interesting to map the Network of Groups. There are so many Groups on LinkedIn, and it would be interesting to make comparisons between Groups, across industries, and interest groups.

    Instead of showing, “how am I connected”, it would show, “who is really talking (connecting) to whom” in the groups that I belong to. Do you know any way to do that?

    One thing I find interesting is that many groups that I follow have over a one thousand members, but it’s really only 20-50 people who engage in discussion, asking questions, and leaving comments / answers in a forum style. Most people sit on the sidelines and invisibly consume, some connections lie dormant, and content creation is routinely done by only a small handful of keeners. The Long Tail cast into a network map!

    The power of this would be to represent the true reach of knowledge sharing and connectivity for any given group, and analyze ways to improve or spread relevant information. That success or failure of a group can be hard to see from a stream of disconnected comments and questions in a discussion-forum-like setting.

    Any thoughts on how “discussion group” data can be mapped in the same for as the individual network you have shown in this post? (hopefully without too much pain and gruntwork) :)


  7. says

    Really interesting post!
    I read it only today and I published in my blog my map.
    I think this kind of tools can be very useful to think about our network, both in professional and personal life.
    Thank you

  8. says

    I keep coming back to this post even 18 months later. I’ve spent quite a bit of time looking at people’s LinkedIn maps, and their commentaries about them, and yours is the most thoughtful. Noone else has gone into the analytical depth that you have. Have you thought any more about your network – either on LinkedIn or elsewhere – since then?


  1. […] This post was mentioned on Twitter by Whitney Hess, Chris Avore and UX Feeder, Ingo Bultschnieder. Ingo Bultschnieder said: Hubs and Connectors: Understanding Networks Through Data Visualization: Yesterday a bunch of friends wer… […]

  2. […] friend Matthew Scott sent me this excellent post from Whitney Hess that shows her poking around her LinkedIn network using this fun new tool […]

  3. […] Hubs and Connectors: Understanding Networks Through Data Visualization […]

  4. […] Pleasure and Pain » Hubs and Connectors: Understanding Networks Through Data Visualization […]

  5. […] You might remember the near-novel I wrote a couple months ago on LinkedIn InMaps titled, “Hubs and Connectors: Understanding Networks Through Data Visualization.” […]

Leave a Reply

Your email address will not be published. Required fields are marked *