Packetability

December 1, 2017

How does behavioural or attitudinal change propagate through groups? One process - that of seeding social contagion (also known as diffusion or peer influence) - is of increasing interest within organizations looking to leverage the power of social change at scale. Typically influencers are identified, and these influencers are leveraged to spread new beliefs or behaviors. Compared to more traditional HR practices such as training sessions, webinars, etc, attempts to seed social contagion are low cost, efficient, and highly flexible. 

What is often lacking in the planning and execution of strategic social contagion is a basic understanding and acknowledgement of what the “thing” being spread is. More specifically, the degree to which the belief or behavior is easily replicable and communicated across individuals. Recently, we (Gladstone, Rubineau, Taylor, forthcoming in the Oxford Handbook of Social Networks) have coined the term ‘packetability’ to describe the ability for something to be more or less easily transferred between two or more actors. The more fixed the form of the belief or behavior, the more ‘packetable’ it is. For instance, a manager attempting to get employees to login to a new piece of software is very packetable: the behavior is a simple binary “do or do not do” and is easily tracked; the manager in question can easily monitor and track the spread of the contagion she initiated. On the other hand, some beliefs or behaviors are highly ‘non-packetable’. A manager attempting to seed a contagion designed to increase personal responsibility among employees, or decrease unethical behavior, is generally low in packetability. Why? Ethical behavior, or employee personal responsibility, mean different things to different people--as these concepts are transferred from person to person, they change and mutate, they are highly interpretable. As such, this makes it difficult for the manager to identify and track the spread of the contagion she crafted. Further, because ethical beliefs or feelings of personal responsibility don't necessarily take binary or digital forms, knowing that a contagion was successfully diffused is difficult. 

The strategic use of social contagion to enact organizational change is a massively powerful tool. At interstitio, we are experts in Network Analysis--the science of understanding the connections among people, and how those connections can be leveraged to manipulate social contagions, if you’d like to learn more about how you can using data you already have, be sure and drop us a line.

Bavelas Networks

December 1st, 2017

In the 1950’s, social psychological researchers were increasingly concerning themselves with the study of groups, and how groups solve problems and how these processes differ from individuals solving problems. Often, however, these researchers overlooked a fundamental component of group interaction: communication patterns. More specifically, the structure of these communication patterns and how they impact the transfer of information between individuals. If problem solving is a function of information transfer among parties, then it stands to reason that the “flow” of this information between individuals has a significant impact on the solution generated for a given problem. Bavelas hypothesized that certain types of group structures are better (or worse ...) for solving certain types of problems. 

Why does this matter for an organization? What can a manager do with this information? That’s a dandy-ass question. Let interstitio tell you! The reveal (The Prestige?!?!), in short, is that organizations can strategically create certain types of communication structures in their task groups which are ideally matched to the problem at hand. 

Put on your manager hat and consider the following two communication structures implemented by a savvy Executive type who reads this blog. What do you see? The blue circles are individual employees, and the arrows between them represent possible avenues of communication. Where no direct arrow exists, no direct communication is possible between any two given problem solvers. As you may see, the two groups differ in the ways in which members can, and cannot, communicate with one another.

The graph--what we call a communication network in Network Science--on the left is ‘maximally connected’. Every person in the task group has a direct line of communication to everyone else in the group. The graph on the right is more towards ‘minimally connected’ i.e. people can only directly speak to their direct neighbors. 

These types of graphs (arrangements of people) are strikingly common in their appearance in daily life. For instance, the graph on the left, resembles an assemblage of jurors. Each person, in theory, can directly address every other person in the group. While, in contrast, the graph on the right more closely resembles a hierarchical chain of command, such you might find in the military or in a production setting. 

This is an important observation if you pause to consider the problems these two task groups are, well, tasked with. The juror group are making decisions based on inherently subjective, intangible, possibly incomplete, and often, biased information. The juror group, in turn, are full of their own personal, social, and cultural biases which impact how they process information. Because of this murky information, and the biases inherent to each juror, it is likely that certain jurors are going to have more, or less, extreme reactions to the evidence presented. What a maximally connected graph allows for is the averaging of these extreme opinions into something more central, and in theory, less biased (maximally connected graphs are also realllllyyyy good at promoting group-think and mass hysteria. Different story for a different day). 

In contrast, the task of a chain of command in the military is to efficiently pass information from person to person. Perhaps more importantly, the information being passed is not to change or be impacted by personal, social, or cultural bias. In theory, what a minimally connected graph does is to minimize the likelihood that dissenting opinions, or cross table discussion, interfere or alter the original communication. This serves to preserve the original communique, albeit sometimes with disastrous consequences (see for example here and here). Certain communication structures are, on average, better for coping with ambiguous and complex social problems. Other types of communication structures are better at preserving the status quo and enacting standing orders. 

A interstitio, we are experts in assessing the types of problems you might be facing, and helping you think through how to effectively structure your task groups to promote optimal outcomes. If this sounds interesting, if you’d like to learn more, or just have some questions - be sure and drop us a line.

Curious, Spurious, and Cyclical: Or why data driven decision making is garbage ...

May 10, 2017

Data driven decision making is garbage. Pure bunk. One foot in the grave, and the other on a banana peel. “Wait”, you say, “don’t you live and breathe science, and data, and problems and answers?” Yep—sure do. Let us elaborate.

Oh, and just to confirm, there are no bears in this post.

It is no secret that companies are attempting to leverage data to make better informed decisions. Proponents of data informed strategy point to the sheer volume and diversity that is accumulated every day from innumerable sources. Managers and executives, it is claimed, can leverage this influx of information to better analyze their firm’s internal or external environments and design accompanying practices. Critics of data driven analytics, who are also in the minority, note that access to diverse data is not in and of itself a novel or new phenomenon. To that, they would say that data driven decision making can be crazy harmful. The 1960’s, for example, saw a landslide of information coming down the mountain in terms of survey and demographic information. It was believed that, no joke, a literal social utopia could be formed by just going through the data. An entire center for survey analysis was formed at the University of Michigan with this very purpose—to create a purely analytically driven society where rationality, and actuality, rule the day. 

Alas, as one may observe, this utopia did not emerge. The researchers at U of M found themselves up data creek with half a paddle. Projects were began and abandoned. Spurious correlations between x and y variables drove actual political and social policy initiatives. In short, the researchers focused too heavily on the data and not the process. They began their work with assumptions as to what the problem or question was. And these untested assumptions drove them down the (wrong) proverbial rabbit hole.

At interstitio we pride ourselves on the process of data driven decision making. The process is a creative, even artistic, curiosity driven episode grounded in rigorous science. The scientific method requires that initial conditions, also known as assumptions, be tested before one can even begin to think about the manifest problems or questions. This is the start of the process. Too often, organizations gather data based on chains of implicit assumptions about what the problem is. From this, they get answers to problems they themselves engineered. It’s curious, and spurious, and cyclical.  

Think about a problem facing your organization, or perhaps even a supposed troubled employee. Now, force yourself to answer this question: do you know—like, fully and without a doubt-you-just-met-the-partner-of-your-dreams, put your life on the line—that your assumption about the problem is as close to accurate as possible? Now, answer one more question: If your prior answer was “no,” how do you go about testing your assumptions? 

A good deal of the work we do begins with those two questions. It’s amazing how often we, as people, cruise by on schemas and assumptions. It’s even more amazing that teams of highly trained professionals guiding massive firms do the same.  

Don’t be those folks. Question your assumptions, and then question those questions. Find ways to gather data on your assumptions, before gathering more data. Do better than U of M did as you look to make your own organizational utopia. We’d love to help, so don’t hesitate to drop us a line.

Conceptual Knowledge Networks

April 14, 2017

The above graphic represent a URL scrape of Wikipedia. In particular, a seed Wikipedia page was chosen (the United States of America). From here, all internal hyperlinks on that page were scraped. And then those hyperlinks were scraped. This process occurred until a terminal point was reached. 

What’s super cool is we can see how “big” concepts act as network hubs. These hubs allow travel to both other hubs, and “smaller” concepts. In the middle, we see the “large” concept of the US. Surrounding it are “smaller” concept such as “Flint, Michigan.” Towards the outside, we see various hubs that connect the conceptual network. For example: in the upper left we see a hub—this is the Wikipedia page on “religion”—an obviously “large” concept. Other “large” concepts that act as hubs are “War,” “Buddhism,” “Jesus,” “English Language,” and “WW2.” One can easily imagine how easy it is to get to a given Wikipedia page by simply referencing any of these “big” concept hubs. The darker red the lines between nodes/concepts, the greater the amount of ways one can conceptually get to a given concept. For example: there are many, many ways one can navigate the hyperlinks in Wikipedia to get from the page on the US to a page on WW2. 

Hyperlinks represent a significant enhancement in how human beings store and represent information. Ted Nelson coined the term in 1965 (or possibly 1964) while a member of Project Xanadu. Mr. Wilson was inspired by a 1945 article by Vannevar Bush in which Bush described a hypothetical manner of linking any 2 given micro-films via a “trail” of related information.

Hyperlinks are a distinctly human way of organizing information in that unlike previous systems (i.e., the Dewey Decimal Classification System, 1876), they are relational. Traditionally, information and knowledge were organized in a categorical or nominal way. A given designation or code was assigned to a given piece of knowledge or information. In this way, knowledge was organized as discrete pieces of information in isolated bins. Hyperlinks, in contrast, organize information in relational manners—once concept links to another concept, and to another, ad infinitum. This, it turns out, is exactly how the human mind organized information—not hierarchical, but in a web or network of interconnected ideas, beliefs, and concepts. Conceptually, it is trivial to think of how one can get from the concept of “Blue Whale” to “My dog Buddy.” Blue whales are mammals, and the larger category of mammals includes canines. My dog Buddy is a canine.