In Chapter 7 of the book Weapons of Math Destruction, we get an idea about mathematical models that create deadly feedback loops for workers in the job field. Nowadays, in order to optimize profit and save costs, employers model customer traffic data and based on the model’s predictions, employers schedule staffing. As a result, the employees are usually always on the edge with their schedules. Their logistics, growth and life in general becomes very chaotic because of such just in time scheduling. They end up in a poisonous feedback loop where due to the nature of their clopening and other unreasonable work shifts, they are unable to study or provision time for personal growth. So, they stay stuck in a minimum-wage job for most of their lives.
However, if companies are a little thoughtful of the employee’s life standards, they can easily make their workers’ lives more stable and convenient. A simple step like informing the employees about their schedule one week ahead of time can bring a positive change in the workers’ lifestyles. If the employers are using the right data and right model, they would not have to compromise a lot of profit for early scheduling. In fact, a convenient lifestyle for the employees can actually ensure a productive and profitable workplace. Sometimes, even if company executives participate in pledges of early scheduling, store level managers see their own benefit in just in time scheduling. So, the pledge is not executed at the ground level. Companies can monitor these store level operations to enforce the managers to follow early scheduling. State and federal regulations will also definitely be helpful.
Apart from employee scheduling, such value-added models in workplaces can wreak havoc in teachers’ careers as well. At some point, teachers were being evaluated and laid off based on the test results of the students. Student results are influenced by a multitude of factors beyond the scope of a teacher. In the absence of the such influential data, WMDs use proxy data to model teacher success. In the case of a tenured teacher named Tim Clifford, we can see ridiculous randomness in how he was scored in his evaluation. At first, he got a score of 6 and the next year he got 96. In the first year, he had a very skewed distribution of students who were either needy or excellent. However, in the second year, the distribution of the students centered around the middle. As a result, the model detected this as a measure of success and he got a higher score the next year. However, his teaching method had not changed at all from when he got a 6. Such value-added models inaccurately score social circumstances. In the case of the teacher model, an effective approach was to eliminate the requirement of standardized tests. So, teachers would no longer be evaluated based on student performance. But, the capitalist and meritocratic world of the present will by no means allow the removal of such tests. Inevitably, we are stuck with these models for better or for worse. So, in order to make the most of it and to make them fairly effective, we can focus on the data they are trained on. At the same time, these models should be less opaque so that people can justify their predictions. Also, the model’s designer team should have diversity so that the model caters to all classes of the society.
With that said, some Universities are indeed starting to make test scores optional. I think that this is a very good idea. In order to compensate for these ridiculous tests, institutions can focus on a student’s aspects that matter. More weight can be placed on the student’s academic performance and extra-curricular activities. Good letters of recommendation are indispensable. Therefore, such letters and other achievements should get more spotlight when deciding regarding a student’s admission. Since GPAs, ECAs, Awards and letters are unbiased and competitive platforms of evaluation, using these effectively and equitably can ensure a transition away from biased standardized tests.
