The MarkUp (an investigative publication focusing on Tech) has released an investigation into the Wisconsin’s state algorithm used to predict middle school students’ dropout before they graduate from high school.

Read the whole story here.

The algorithm is called The Dropout Early Warning System (DEWS). Students dropout is an important issue that needs to be addressed. Improving the chances of students staying in school and graduating from high school is a laudable goal. The question is: are algorithms reliable tools to do so? As it happens, it seems that they are not.

DEWS has been used for over a decade. The data used to create scores includes test scores, disciplinary records, and race. Students scoring below 78.5% are marked as High Risk (and a red mark appears next to their name). The MarkUp reports that comparisons between DEWS prediction and state records of actual graduations show the system is wrong three quarter of the time, especially for Black and Hispanic students. In other words, the system invalidates the very purpose for the system to exist.

Even more telling: the Department of Public Instruction (DPI) ran its own investigation into DEWS and concluded that the system was unfair. That was in 2021. In 2023, DEWS is still in use. Does this mean that our over-reliance on algorithmic systems has created a situation where we know they fail us, but we have no credible alternative, so we keep using them?

I am reminded of the seminal book by Neil Postman “Technopoly”. He says that in a Technopoly, the purpose of technology is NOT to serve people or life. It justifies its own existence merely by existing. In a Technopoly, technology is not a tool, “people are the tools of their tools” (p68). More importantly, and more problematically, “once a technology is admitted, it plays out its hand; it does what it is designed to do. Our task is to understand what that design is” (p128). It is safe to say that digital technologies have been admitted, but do we really have any understanding of what their design is?