Relationships as "baggage", or "baggage" isn't always a bad thing

One of the basis of social network analysis is the idea that people have connections between them. These connections can be anything from frequency of interaction or asking advice, to liking, friendship or joint membership in groups or organizations. Whereas these observed links have been thoroughly scrutinized, little attention has been paid to the reason those links exist in the first place. Why they are there, how long they are likely to persist and what forces act on these links, forcing them to die or to come into being. This project proposes a generative model of relationships - these relationships being friendships, collaborations or other interactions that depend on factors inherent to actors and on context within which they exist. This model also considers the fact that context is likely to change and such changes can have a strong effect on existing relationships and probability of new ones. This is an iterative model of the evolution of ties - their birth, their maintenance and their death, conditional on geographical proximity and other factors of similarity.

What if for each iteration, you can have one of several events happen. Some events happen with high probability - i.e. reinforcement of ties while others happen with lower probability - new ties with geographically proximal partners that are rated highly similar on the similarity metric. Others happen with really low probability - new ties with non-geographically proximal partners, or proximal partners that are low on similarity.

Then we need to add shocks to the system - that is, at each iteration you have a check - geographical position of actor compared to prior iteration. You have low probability of change but if change exists, then your probabilities on that set of potential events changes at first, then comes back to normal over time (harmonic frequencies?).

So for each iteration, you can have one of several events happen. Some events happen with high probability - i.e. reinforcement of ties while others happen with lower probability - new ties with geographically proximal partners that are rated highly similar on the similarity metric. Others happen with really low probability - new ties with non-geographically proximal partners, or proximal partners that are low on similarity.

The idea is relatively simple - in each geographical location people meet other people and develop sets of personal relationships. As they move, there is a selection process at work. Movers keep some relationships and loose others at various rates of decay. As movers arrive in the new location, the process of engaging in new social relationships (and work relationships) takes time and effort. Considering that many researchers are not of the most extroverted kind, the process of meeting new people and initiating new collaborations can take a while. So for a while after a move, most collaborations happen long distance with pre-established ties. However, as geographically proximal relationships develop, the frequency of interaction or collaboration with long distance contacts slowly decreases until most of those relationships become dormant and only very few persist.

As people move, the story repeats itself. Over time, one can observe researchers with a huge range of active and dormant ties. Like a spiderweb, their network covers large distances, but only a few of the strands are ever really active professionaly (probably more so personally though). Many of these ties get renewed or at least maintained during conferences or workshop meetings, so that some dormant ties can come into a more active state of collaboration at times. As researchers mature, their range of interests branches out and so does their range of influence. They develop more relationships with their students than with other faculty and continue those collaborations for longer periods of time. Yet even those relationships decay and return to a dormant stage.

One way to observe these processes is to consider a citation database of a particular field, and to match a researcher's geographical location with each citation - this would be a way to monitor geographic movement. Computer Science databases would be better for this purpose because their publication turn around is much quicker and the volume is larger than in social sciences or humanities. The data would have researchers as nodes, their geographical location as a weight on each "co-authorship" - that is, geographically proximate links will have a different value than those that are geographically distant (because proximate links would be easier to maintain). Another attribute of the links would be the number of articles co-authored together to date (regardless of geographical proximity). These networks would be a dynamic system with time as a defining variable, with network snapshots taken in 1-year intervals.

There are two things to discern in such a network -
1. to see whether the model described above fits the data
2. to identify the "connectors" in the network - people who may not publish very much themselves, but who foster a lot of collaborations between others through putting them together on the same paper once or twice.

How:
1. data collection - crawl the ACM database
2. analysis with ORA (although the generative model can be implemented in R first)

On collective individualists: Social relationships govern the world

Tell me your friends,
and I'll tell you who you are.
Assyrian Proverb

Personal relationships are the basic building blocks of human existence. Regardless of our cultural backgrounds or personal biases, we depend on each other for both physical and emotonal functioning. A brief look at psychological and social research shows that social contact is important for psychological well-being. People with more social contact are physically and mentally healthier and happier (Kraut et. al.). Love and belongingness take a prominent place in Maslow’s motivational hierarchy, and the need to belong is supported by later psychological research (Baumeister & Leary). This same need to belong motivates healthy people to form strong interpersonal relationships and put effort into their maintenance. People need the ability to maintain both strong and weak ties in order to function efficiently in the society (Granovetter). One of the reasons for this social need is that people who have strong social relationships are better capable of dealing with life stress and negative events (Bolger, Wellman).

Records of importance of social relationships can be found in fiction, philosophy, science and just about any other form of record keeping. In fact, there is ample evidence that while isolated genius is rare, communities of brilliance are relatively common (Collins, 1998). Our relationships then, are responsible not only for our health and happiness, but can often be sources of inspiration, creativity, discovery and growth. In fact, many have made the argument that our relationships, our belonging to communities, is of paramount necessity for thinking, advancement and revolutions (Collins, Kuhn, Tournier). This assertion offers an explanation for why advancements in the arts, sciencies, philosophy or technology happen in spurts and bursts. Those unfortunate enough to come after such explosions of creativity often feel like everything "has been done".

Examples of such bursts are easily available. In the arts, one can think of the Renaissance, or the expressionist movement, the Russian Avant-Garde at the dawn of the 20th century or the American abstract expressionists during that centuries' "mature" years. It is easy to think of these as "movements" after the fact. At the time, however, these were probably less movements and more a social vortexes. The feverish creativity of one would infect all others and then re-infect the originator, resulting in achievements of stunning innovation. History is witness to the fact that these vortexes happened as a young generations found ways and reasons to push against the canons of the established, to subvert, to question, to tear the rules into pieces and to build their own. One would be wrong, however, to say that there must be a disconnect between generations then. History is also witness to strong links and inspirations from one generation to another. Collins suggests that while "ideas beget ideas" through social interaction of like minds, intergenerational networks or "chains of eminents teachers and pupils" are also key to understanding abrupt movements of change in thought, art, science or social structure. Kuhn observed a similar pattern in his "Structure of scientific revolutions".

If relationships are so important to our thinking and our ability to create, is it any wonder that the most common use of network technologies has been communication? Yet it has been difficult for companies invested into ideas of "social relationships", to produce sustainable business plans. Somehow, it seems, relationships aren't really for sale. Or are they?

Not too long ago, Nardi and colleagues (2000) suggested that the workplace was changing because personal relationships became easier to access due to computer mediated technologies. These changes, they argued, made it possible for workers to bring their personal networks into play above and beyone, or even instead, of the traditional org chart. Of course, personal networks always played a bit role in success or failure of companies and businesses. Development of computer mediated communication may have made them more obvious and allowed them to exhert and even stronger influence. Yet I would argue that the org chart was never a good description of how a company worked.

What I am trying to say, I think, is that our relationships have always defined how we lived and a large portion of what we thought about and how. The easier it was to communicate though, the easier it has been to maintain a diverse set of relationships, some local, some geographically distant and to maintain a similar level of influence from both kinds.

Here we get to the question of geography by way of long distance relationships. Of course humans have communicated over distance since the invention of smoke signals. Yet as fidelity of communication went up and costs went down, the frequency and volume of communication certainly increased. In her study of scientific collaboration, Liverow (1994) notes that the telephone was a key technolgy that allowed researchers to cluster by interests rather than by geographic location (although the latter also played a big role in cluster structures). With the development of email, ability to exchange files immediately regardless of distance and to communicate quickly and at any point in time may have expanded possibilities for collaboration. I suspect, however, that points of contact, such as discussion spaces at conferences (less frequently but possibly discussion lists online) were and stil are instrumental in fostering collaboration. Bob Kraut once told me that every single one of his collaborations, regardless of whether it was geographically proximal or distant, always began with a shared meal.

There is something to that. A shared meal is an informal space, where people build informal social relationships rather a formal working relationship. It is on the basis of these relationships that good formal collaborations usually work.

So:
1. social relationship (personal relationships) drive our lives and are sources of inspiration that can lead to success (or failure).
2. professional relationships, to be fruitful, may need to be cemented with an informal social interaction
3. mediated communication technologies bring use of social relationships in work contexts to the forefront, they make this practice more obvoius, but this practice is age-old.
4. mediated communication MIGHT allow us to retain more long distant social relationships fostering stronger collaborations within communities of practice (faster growth of revolutions?).

It is this last one, that I would like to test. But first, we need a baseline measure. So... how about that ACM citation index prior to 1995?