Blumenstock leads his audience in with a strong usage of logos, a hopeful view of the age of information. Then, he strings his audience together in the one way they can all be connected–technology.
He continues on with a theme of hope, depicting how large bodies such as the UN and World Band are seeking to hire data scientists. He talks about the leaps and bounds made in the technological sphere: Over the course of nearly two decades, the number of individuals owning a mobile device worldwide has increased exponentially.
With a pocket-sized device constantly harvesting data, the resulting graphs and/or maps have the potential to change how the world, primarily developed countries, remedies global issues and distributes humanitarian aid.
From there, Blumenstock proceeds to discuss the possible disadvantageous of pursuing such an ambitious humanitarian feat.
With any deed, no matter the intention, comes externalies. Blumenstock states that many of these externalities would originate from the project’s intended parent–big data.
While the majority of humanitarian propositions seem like they can do no harm, many have secret prices to pay. If not refined, many of them have exploitable loopholes or force users into a cycle of dependency.
Alternatively, all data collection is subject to sampling error, meaning data application can be faulty. Specifically, Blumenstock cites maps that chart distribution of wealth. They are strongly susceptible to outliers and skewed means. The algorithms that produce these maps are largely untested, translating to a greater margin of error.
Algorithms are also biased against those who are less financially successful. Blumenstock highlights the barriers for those who live in poverty which includes many developing countries.
The legal portion of commercial data is largely unexplored. In many cases, a misdeed must occur for legislation to be enacted. Developing nations act as a nearly perfect breeding ground for corruption within big tech.
Blumenstock states that initiatives that attempted to achieve this ideal, such as the one-laptop per child initiative, failed. He blames this on the ineptitude of data scientists to be aware of the social/cultural environment surrounding them.
The most obvious need is to perfect the science of the algorithm. While data collection grows more advanced every day through means such as Google or Tiktok, the capitalistic nature of big tech corporations hinders progress. These companies exist in a zone void of collaboration, which is one of the first rules of any science.
Blumenstock emphasizes increased transparency as a solution. In order to achieve this, data, or information in general, must be digestable. Therefore, more educated individuals, those who harbor influence, are incapable of filtering information distributed to the masses.
I find this to be a widespread phenomena throughout the scientific community as a whole. As individuals become more educated, they learn new methods to express themselves. With these methods comes jargon that is only decipherable by the scientific community. For example, consider the field of statistics, where expertise means nothing until one understands the value of a confidence interval, r-squared value, p-value, z-test, t-test, etc.
I would liken this to Gutenburg’s printing press which allowed the Bible to be translated from Latin into common dialects. Up until then, the people were forced to place their trust in corrupt church leaders. The printing press came with its own set of externalities–primarily mistranslations, but society has evolved in such a way that more of the population is able to debate and have civil discourse regarding the text in question.
While the scenario we face is more complex, transparency is key. The common person should be able to interpret information for themselves. This can include simplifying the distribution of the information, educating people to understand it as it is, or meeting somewhere in the middle.