I distinctly remember a former professor telling me that she provided false information when filling out her Kroger card. She wanted the rewards, but she didn't want someone to be able to track purchases to her. This type of data is so widespread. When you shop on Amazon, your purchases are fed into an algorithm that suggests and predicts other purchases you might like. Meijer now offers money off future purchases when you reach a limit and log in with MPerks.
I recently began reading Weapons of Math Destruction because a colleague referred me to it. I'm convinced that this book, among others, should be a mandatory read for every educator, young and old. We live in the age of "big data," data that promises to be helpful but is really quite terrifying.
Like Cathy O'Neil points out in her book, an algorithm is really "an opinion formalized in code" (53). But try as we might to make a mathematical formula objective, it will be subjective and flawed because a human created it.
Sometimes we hear promising stories of big data, like this New York Times piece about success in Georgia State's nursing school. As a teacher, this kind of predictive information is helpful. We really do want all students to succeed, and if there is the hope that we can catch even one student before he or she fails, then we are being successful.
But this book also raises questions about the "feedback loops" that WMDs unintentionally create.
O'Neil gives many great examples of these loops, like D.C.'s teacher evaluation system that was so flawed that people using the system could not even explain how it worked.
And she goes really in depth into the ridiculousness that the U.S. News & World Report's college and high school ranking systems really are. Creating an "arms race" between schools, colleges have been forced to spend money in ways that don't really matter because the American public has bought into a system that isn't necessarily based on research.
Take, for example, the fact that 25% of the rankings is based off of other college presidents' opinions of that college. I don't even need to begin to explain that subjectivity. Then you have the other 75% that includes metrics like acceptance rates. O'Neil wrote that "Winning athletic programs, it turns out, are the most effective promotions for some applications" (57). And then schools can game the acceptance rate score by rejecting even excellent applicants that they know would probably not attend the school in the first place.
As teachers, we will continue to encounter big data. I suggest we use O'Neil's questions to guide us as we encounter them:
I recently began reading Weapons of Math Destruction because a colleague referred me to it. I'm convinced that this book, among others, should be a mandatory read for every educator, young and old. We live in the age of "big data," data that promises to be helpful but is really quite terrifying.
Like Cathy O'Neil points out in her book, an algorithm is really "an opinion formalized in code" (53). But try as we might to make a mathematical formula objective, it will be subjective and flawed because a human created it.
Sometimes we hear promising stories of big data, like this New York Times piece about success in Georgia State's nursing school. As a teacher, this kind of predictive information is helpful. We really do want all students to succeed, and if there is the hope that we can catch even one student before he or she fails, then we are being successful.
But this book also raises questions about the "feedback loops" that WMDs unintentionally create.
O'Neil gives many great examples of these loops, like D.C.'s teacher evaluation system that was so flawed that people using the system could not even explain how it worked.
And she goes really in depth into the ridiculousness that the U.S. News & World Report's college and high school ranking systems really are. Creating an "arms race" between schools, colleges have been forced to spend money in ways that don't really matter because the American public has bought into a system that isn't necessarily based on research.
Take, for example, the fact that 25% of the rankings is based off of other college presidents' opinions of that college. I don't even need to begin to explain that subjectivity. Then you have the other 75% that includes metrics like acceptance rates. O'Neil wrote that "Winning athletic programs, it turns out, are the most effective promotions for some applications" (57). And then schools can game the acceptance rate score by rejecting even excellent applicants that they know would probably not attend the school in the first place.
As teachers, we will continue to encounter big data. I suggest we use O'Neil's questions to guide us as we encounter them:
- Is the model opaque or transparent? Transparency is important.
- Does the model work against the subject's best interest? Can it damage or destroy lives?
- Can it scale and become a "tsunami force"?
We need to stop shopping at Amazon. They do not treat workers well, and as educators, we should not be supporting their inhumane labor practices. There are plenty of local stores to buy from in you're in an urban area. I have zero new book stores in Adrian, so I use Powell's when I have to shop online.
ReplyDelete