Friday, April 20, 2018

Are Risk Algorithms the New Leaders of the 21st Century American University?: The Riskless University and the Veiling of Discretion

Mockery is the sincerest form of criticism. And there was mockery aplenty directed toward the leaders of Penn State University.  And that leadership now appears to have moved from the office of its titular administrative heads to another place, that is from the individuals designated as the lead officers of the university to the risk and compliance algorithms of the university (and the individuals who tend them), which now appear to have assume the highest authority at the university. That, at least, is what we appear to be told in the way in which some decisions are now made at American universities.  

This should come as no surprise; nor is the issue unique to any particular university.  As universities shift to a corporate model, and as the authority over education and student programs shifts from faculties and departments to central administration and non-teaching administrators, it should come as no surprise that cultures of risk aversion in internal governance should move to the center of the educational mission of the university. What is more interesting, though is that this shift is not merely about moving from one set of individuals to another set--distinguished by experience and their relationship to the actual work of education (rather than in the management of human resources in institutional settings).  Rather it is about the shift in governance from people driven cultures of administration to data driven cultures in which administrative discretion is not exercised (and abused) by individuals, but rather built into the algorithms that now serve to mask the exercise of discretion within the more neutral sounding relationships of data organized into relational.  

The recent decisions by the leadership algorithms of Penn State (not of course by those administrators who serve those decision makers) to disband three ancient student clubs because their activities were too dangerous  provide a nice but generalized example of the trend that affects all universities. In the process, the necessity and character of shared governance, of consultation, and of the role of human beings in the administration of the university  has changed substantially as well. This ought to be a matter of general concern to those who think about the trajectories of university (and more generally of institutional governance).

The story appeared recently, and as reported, seemed odd though innocuous enough:

Penn State is disbanding three student clubs at the end of the semester because the university considers them too risky. Penn State Outing Club, Nittany Grotto Caving Club and Nittany Divers Scuba Club are "losing recognition due to an unacceptable amount of risk to student members that is associated with their activities," university spokeswoman Lisa Powers said in an email. This spring, Campus Recreation, a division of Student Affairs, reviewed all of its supported organizations, including 76 sport and the three outdoor recreation student orgs, she said. Changes were recommended to 20 registered student organizations, including the three disbanded. Among the factors considered in the evaluation, Powers said, were environment and location of organization activities; access to emergency facilities and distance to medical care, risk associated with various types of impact likely in an activity; and impact the equipment used in an activity has on the risk of an activity. ( "3 Penn State clubs lose recognition, university says they're too risky," Centre Daily Times 20 April 2018).
more here:
The Pittsburgh Post Gazette was more direct:
The Outing Club isn’t allowed to go outside anymore. According to an announcement posted by the club on its Webpage last week, the university will not allow the club to organize and run outdoor, student-led trips starting next semester. “This is a result,” the announcement said, “of an assessment of risk management by the university that determined that the types of activities in which PSOC engages are above the university’s threshold of acceptable risk for recognized student organizations.” After a two-month review that did not include consultation with student Outing Club leaders, the university’s offices of Student Affairs and Risk Management, made the determination that the hiking, canoeing, kayaking, trail building and camping activities the student-led club has long engaged in are too risky. The club is one of the oldest entirely student run organizations at Penn State. “Student safety in any activity is our primary focus,” Lisa Powers, a Penn State University spokeswoman, said in an email response to questions about the school’s assessment. (Don Hopey,  "It's risky out there! That's why PSU's Outing Club can no longer go outside," Pittsburgh Post Gazette, 18 April 2018).
In an earlier essay I considered the manifestation of this trend on another context, one affecting faculty rather than students.  These centered around the move toward greater control of the travel activities (and thus the research and knowledge production-dissemination) of faculty.  These policies mean to shift authority over that aspect of faculty activities from the individual researcher to the university, and to substitute the mechanical risk management strategies of the institution for the risk-reward balancing required of front line research in a globalized educational and research environment. (The Riskless University and the Bureaucratization of Knowledge: From "Indiana Jones" to Central Planning 19 Nov. 2014).

It is a pity that the university, like other large institutions, has substituted data for thought and has chosen to mask the exercise of administrative discretion through the premises and objectives embedded in algorithms.  There is nothing predetermined about the decisions made--each was a function of the way in which compliance officials chose to model their algorithms and chose the data that would be weighed in the production of the"result" that emerged from the application of these measures. Yet neither is invariably the case.  And that ought to be something that stakeholders in shared governance ought to begin to think about.  But for administrative officials this might serve interests.  First it cloths the exercise of discretion not in individual desire but in the neutral workings of mathematical formula.  And set the algorithm merely expresses the desires of discretion through its premises and assumptions--and through the choices of data that it consumes. Second, it appears to distance the administrator from the decisions themselves--avoiding, to a large extent any effective mechanism for accountability (the "the "algorithm did it" ploy). And third it appears to contribute to the further deprofesisonalization of the faculty by implicitly assuming that the business of data harvesting and algorithm administration is something that ought to be left to proper professionals.

Administrative decision making appears to be moving towards a grounding in the language of algorithms the control of which appears to be more and more embedded deep in the bowls of the non-teaching administrative machinery of the university.  In these dark recesses of the administrative state there is little room for shared governance--and there is certainly almost no room for stakeholders at the only regulatory table that counts--the meetings at which data is selected and algorithmic parameters are set for the construction of risk related governance of the academic and student centered missions of the university. And that is the greatest pity of all.

Yet having noted the foibles of administrative incorporation of data driven algorithms as a substitute fro and as a way of veiling the exercise of institutional discretion, it is also important to note some complicating points.

1.  Data analytics is here to stay, at least for the medium term.  It is bound up intimately with trends toward accountability (or at least the appearance of accountability)--which is itself a complicated issue (see, e.g., here).

2.  Algorithms themselves are also inextricably tied to the trend toward data driven alaytics and accountability. The algorithm is a natural means of marshaling data to either paint a clearer picture of the past or point to the future.

3. The difficulty lies where data, analytics, and algorithms are used not merely to understand, but also to manage behaviors.  It is at this point, and this point only, that the algorithm becomes a managerial and policy tool. It in that guide it is most easily abused where it is rolled out under the guise of accountability or analytics but actually masks critical behavior modification or management objectives that are buried within the premises of the algorithm itself. Policy is embedded in algorithms at a number of key points: (1) the determination of the scope and character of data to be harvested; (2) the assumptions that are used to create and manage thew algorithm itself, and (3) the consequences that then follow from the application of the algorithm to specific acts or situations. The problem here is that at no stage was there any meaningful engagement in any of the processes that produced the algorithms which was then "mechanically applied" to render a judgement. In effect the process from data to analytics to algorithm then substitutes itself for and masks the politics of regulatory engagement and the exercise of administrative discretion, bypassing any conneciton with engagement or shared governance.

4. There is a substantial movement within the legal sphere to extend the liability of institutions with respect to all programs, activities and faculties they maintain to which they invite others. This is a long term trend in tort law that can be lauded or despised but that is in any case real.  Institutions face increasingly larger ambit of liability for conduct and activities for which there was once a much larger space for assumption of risk.   These are societal and political judgments but ones which the institution ought to be able to react to.  The problem here is that where the institution makes adjustments without even the slightest effort to engage its stakeholders--in a governance model that supposedly prides itself on such engagement--then there is a strong oder of avoidance and bad faith that tends to reduce the legitimacy of the actions, the value of the algorithms and the trust of stakeholders in the good intentions of its administrators. 

4. The change in policy masked as mere ministerial application of "formulae" does not leave stakeholders without options.  There is little that prevents individual adults from coming together for whatever purpose amuses them--outside of the paternal em¡brace and control of the institution.  But that requires confronting another trend--the increasing passivity of individuals in their relations with the university. It is indeed much easier to let the university handle activities and to organize (and supervise) them than it is to do it oneself.  And yet this decision by the university is a reminder that there is a price to be paid for dependence. 
Read more here:

Read more here:

No comments:

Post a Comment