Response to Joshua Blumenstock’s article

Question: Describe the promise, pitfalls and ways forward Blumenstock uses as the foundation for his thesis.

Answer:

Many scientists look to data science and geographic information science as a way to bring accessible aid and global interconnectivity to developing nations. Though Blumenstock acknowledges that machine learning techniques allow developing nations to begin self-advancement, he mainly focuses on the potential downfalls and gaps associated with these forms of data analysis.

Bloomenstock suggests that there are many ways to connect developing countries to the rest of the world while helping the other country’s economic advancement, but for each technique, he draws to one or more associated fall backs. For example, Bloomenstock points out how people who made multiple international calls or who have many facebook friends were more likely to pay off their debts then not. Therefore, machine learning techniques were applied to the citizens, and the citizens with those characteristics simply received a higher credit rating, leading to an easier time for them to receive loans for any reason. Though the advancement has the potential to create a strong economy because of an increased money flow in the country, Bloomenstock brings in strong rebuttals to display that the situation is not as good as everyone thinks. For instance, he asserts the example with the credit and loans have short term benefits, but we cannot know if it will produce long term harms or not. Also, he points to how even though the study might hold true during the holiday seasons, data surrounding international phone calls change given the time of the year. Therefore, he suggests that data should be taken with a grain of salt.

Bloomenstock takes these potential downfalls and pairs them with possible solutions. For instance, he suggests that if scientists pair data from satellite imagery and cellphones with census data, they would have a solid base from which they could draw conclusions from, acting as sort of a check on the new type of data. Also, he anticipates that having data scientists work with a hand full of other types of workers - cultural awareness agents, engineers, government employees - would greatly validate the results of the data by providing deep collaboration, because the deep collaboration would include multiple people from varying perspectives which could add to the methodology.