In the first article, Mr. Kieran Healy discusses how to use “metadata” in order to find associations, such as his example of finding terrorists in colonial America. He bases this off information found by David Hackett Fischer and a beta of PRISM (a database of over 260 people from the 18th century). Healy stresses that this data is only groups these people were involved with, as opposed to what they actually did or conversations they had. Keep in mind, these groups are considered terrorist groups in the time period, and what they are trying to achieve.
So, to begin we have a chart, which he calls “A” of 7 groups (columns) and 254 people (the rows). If a person is in a group, a “1” is shown in the cell. If they are not in said group, a “0” is shown. To find more information, Healy then used a technique created by Ron Breiger. We take our chart and flip it, so the groups become the rows and the people become columns, which is called “Aᵀ”. Now we have two charts, chart (which is the original chart, People by Groups), and chart Aᵀ (the new chart, Groups by People).
Now, we create a workable chart by multiplying the chart A by chart Aᵀ, the equation is A(Aᵀ). It is essential to do this equation, and NOT Aᵀ(A), or else we will get different results. The result is a chart of Person by Person with 254 rows and 254 cells. This will show us how many groups a person may be in with another person. So, if I wanted to find how many groups John Adams was in with Samuel Adams, I would go over to Adams, John, and down to Adams, Samuel (or vice versa) and I see the number 2, meaning Samuel Adams and John Adams were in two groups together.
We can also make a picture of the charts, the picture made from the original chart is very simple, and looks much like the email example Professor Kane used in class. The picture made from our chart of people by people (254×254) is much more complex. In the center, one can see Paul Revere, meaning he was involved in many groups. This makes him a person of interest.
Following, Healy does three mathematical equations to find more information. The first is called betweenness centrality, finding how many people one is the fastest link between. Revere has the largest score in this area. He then proceeds to find the eigenvector centrality, or the measure of centrality by connection to other central people. Here, Revere appears 4th. Last, Healy finds bonacich power. In this category, a lower score shows more centrality. Revere has the lowest score.
With all of this data in mind, Revere seems to be a prominent terrorist during this time period. However, Healy again stresses that this may be misleading, because it does not include any context. In the second post, Healy also points out that the data he uses, is indeed metadata. In addition, the chart may be historically inaccurate. Last, he points out that there are many different ways to manipulate metadata. He is using a very early, very basic form.
Discussion questions:
-Healy uses terrorist groups to demonstrate this concept. Where else do you believe this data could be used?
-The data was being criticized for not being metadata. Do you think this is metadata, and why?
-Do you believe there are any more reasons this information could be misleading (besides the lack of conversations as evidence)?