Free the Digital Natives

James Geller raised important questions in his letter to the editor "Beware BYOD" (Sept. 2013) but mixed learning and assessment with the practicalities of supporting multiple devices. He highlighted possible outcomes—"distraction, cheating, more cheating, still more cheating"—along with the possibility that incompatibility between devices could result in poor grades, but failed to propose ways to address them.

In today’s digital environment, where the simplest online search can turn up more information than will ever be found in a physical classroom, educators must confront the always-present interconnected nature of these devices by changing the way they teach. The traditional model of didactic lecturing from a podium before a classroom of students should complement collaborative activities, where students learn to research, filter, judge, and create their own learning. Educators must provide an "on-ramp" for them to acquire these skills in the context of the subject being covered and ensure assessment practices are an opportunity to demonstrate individual achievement.

Software and hardware fragmentation has always been an issue for educators, but with declining educational budgets in public institutions across Europe and the U.S., we must acknowledge that many students arrive with better technology in their pockets than is likely to be provided by the academy. Free/open-source software and virtualization are potential solutions, either standardizing on technologies students access and install themselves or providing applications not available through a virtualized server platform.

---

Educators must confront the always-present interconnected nature of these devices by changing the way they teach.

---

The related potential solutions are neither straightforward nor cheap, but educators must at least be willing to acquire the skills needed to address the changing worldview and technical competence of the students in their classrooms, who, as digital natives, will have always had access to these technologies.

Barry Avery, London, U.K.

Back to Top

Idempotence More Than Ever

Pat Helland’s article "Idempotence Is Not a Medical Condition" (May 2012) addressed a serious reliability topic—messaging in a service-oriented world—but in an irreverent way. His axioms (such as "Every application is allowed to get bored and abandon its participation in the work.") are generally obvious when being read but often ignored by software developers in the real-world press of development.

Along with the proliferation of tools supporting distributed Web apps, it is easier than ever for software developers to (mostly) ignore the pitfalls of underlying distributed messaging, at least until the software is stressed in a production environment. Helland pointed us toward the dragons lurking in our assumptions concerning the robustness of messaging in a networked environment, concluding with "basic principles" cast as "four insidious illuminations":

• Because retries happen, all messages must be indempotent;
• Messages can be reordered;
• Hidden effects can cause one’s dialogue partner to miss part(s) of a conversation; and
• Guaranteed delivery of the last message is impossible.

…ignorance of which can result in latent problems to surface only when repair is most costly.

Kudos to Helland for illuminating a significant source of bugs, amusing us in the process.

Steven Pothier, Tucson, AZ

Back to Top

Innovation Vs. Pride in Disruption

Moshe Y. Vardi’s editorial "Has the Innovation Cup Run Dry?" (Sept. 2013) offered two divergent conceptions of innovation: In one, drawn from a 2013 Bard College commencement address by outgoing U.S. Federal Reserve chair Ben Bernanke (http://www.federalreserve.gov/newsevents/speech/bernanke20130518a.htm), innovations (1913–1963) were described as having produced "…dramatic improvement in the quality of daily life…" In the other, a report from the McKinsey Global Institute (http://www.mckinsey.com/insights/business_technology/disruptive_technologies) predicted (for 2013–2025) the emergence of "…technologies [having] the potential to disrupt the status quo, alter the way people live and work, and rearrange value pools."

A good way to view the divergence is to focus on what is changing in the first (quality of daily life) and the direction of that change (dramatic improvement). In the second, how people live and work is changing. The direction of that change (disrupt, alter, and rearrange) is far from unambiguously positive and likely to be viewed negatively by many.

I was thus prompted to explore ideas introduced by the influential Austrian-American economist Joseph Schumpeter (1883–1950), who used the term "creative destruction" in describing his theory of innovation-driven economic growth. In it, innovation operates within the system of production to cause old inventories, ideas, technologies, skills, and equipment to become obsolete. Replacement benefits consumers who experience improvement in daily life through increased capability, variety, and affordability of products and services. Part of the cost of this improvement is disruption in the world of work, as experienced by workers, managers, business owners, and investors, some benefiting and some forced to endure loss and painful adaptation. Schumpeter’s theory thereby contributes to resolving the divergence of conceptions by showing how the effects of innovation once fell unevenly on separate aspects of life—outside work and within work.

Concerning those whose work contributes to innovation through technology, this analysis highlights challenges and their related questions, including:

Lost distinction. Why and how did the world of technology lose interest in the important distinction between effect on work and effect on daily life?;

Pride in disruption. Why and how did the world of technology begin to feel pride in disruption as such, usually expressed without comment, along with or in place of pride in improvement of daily life?;

Degraded communication. Have the loss of distinction and pride in disruption contributed to degraded communication between the world of technology and the world at large?;

Diminished effectiveness. Have degraded understanding and communication subtracted from the ability of the world of technology to satisfy human needs in the world at large?; and

Friction. Have degraded understanding and communication contributed to friction between the world of technology and the world at large?

Although I have not conducted an extensive review, I can say the world of technology gives them insufficient attention.

Robert E. Levine, Sierra Vista, AZ

Back to Top

More Fair Than Just Envy-Free

Although Ariel D. Procaccia’s article "Cake Cutting: Not Just Child’s Play" (July 2013) offered an interesting overview of recent research on cake division in computer science, it emphasized envy-free solutions that are not sufficiently fair.

Envy-freeness is a property that violates the axiom of monotony, or the requirement of Pareto-efficiency, in which a solution that improves the outcome of an agent by any amount is preferable in terms of fairness to a solution in which all agents have equal smaller outcomes. However, such a solution cannot be envy-free.

Envy-free solutions for distribution of multiple goods involve a subtler theoretical shortcoming. Consider allocating food to animals in a zoo. Though the animals have utilities that can be expressed in terms of the calories they consume, they can consume only certain foods. One might eat only eggs. Another might eat eggs but also other foods. A non-envy solution might therefore be to give the egg eater insufficient eggs for its survival. This theoretical weakness is alleviated through the concept of maxmin fairness.

---

Why and how did the world of technology lose interest in the important distinction between effect on work and effect on daily life?

---

Procaccia mentioned utilitarian and egalitarian fairness but none of the important theoretical advances in the definition of distributional fairness. For example, Ogryczak et al.² formulated "equitable optimality," generalizing both utilitarian and egalitarian approaches by imposing three axioms on the preferences of fair solutions: impartiality (the permutation of outcomes from another solution should be indifferent); monotony; and the Pigou-Dalton principle of transfers (a small amount from a better-off agent to a worse-off agent should be preferred). It can be shown that finding equitably optimal solutions does not increase computational complexity by more than a polynomial factor. Such solutions can be found as Pareto-optimal solutions to a problem involving criteria obtained first from the Lorenz curve of the original distribution problem. Moreover, optimal solutions can be selected with consideration for the trade-off between equality and efficiency, increasing the likelihood of their acceptance by stakeholders.

The concept of envy-free solutions is useful only if agents are fully autonomous and selfish in a non-cooperative setting. However, Procaccia’s example of CPU and RAM allocations to computing jobs does not require such a model. In most practical cases, there is a single administration of the computing resources, either cloud or grid, that can and should impose a better solution than the envy-free solution to the cake-cutting problem. It also illustrates that equitably optimal solutions requiring a cooperative setting can be obtained in practice.

A body of literature has aimed to find equitably optimal solutions in various settings, of which we recommend Luss¹ and Wierzbicki.³

Adam Wierzbicki, Warsaw, Poland
Włodzimierz Ogryczak, Warsaw, Poland

Back to Top

Author’s Response

I welcome these points and am happy to respond. First, many of the solutions I discussed in my article are both envy-free and Pareto-efficient. Second, in most fair-division settings envy-freeness implies proportional shares of the resources; in the zoo example, assuming eggs are allocated, the poor egg eaters would be envious. Third, I interpret the letter’s penultimate paragraph primarily as a criticism of strategy-proofness rather than of envy-freeness. However, it is not the jobs that are autonomous but rather the users running them; the scheduler can impose a solution based only on the available information, which can be manipulated.

Ariel D. Procaccia, Pittsburgh, PA

Back to Top

Back to Top