Spring Research Rundown

March 2024

Alan Andrew Scheller-Wolf, Richard M. Cyert Professor of Operations Management and Sridhar Tayur, Ford Distinguished Research Chair; University Professor of Operations Management

Publicly traded firms are reducing carbon emissions, at least partly in response to pressure from investors keen on combating climate change. Investor pressure can have positive impacts (fostering sustainable business practices) but can also backfire (spurring firms to sell their carbon-intensive assets to private companies, reducing transparency and eliminating investor oversight). Key mechanisms for investor pressure include environmental metrics, and different metrics focus firms’ actions differently. In a new working paper, researchers explored the effects of two common environmental metrics for carbon emissions—an absolute-based target for absolute emissions and an intensity-based target for emission intensity. They developed a sequential model to analyze the metrics’ impact on a producer’s operational strategy (i.e., their exiting, investment in emission reduction technology, production amount, and disclosure decisions). In so doing, they identified a possible explanation for the phenomena of “greenhushing” (in which firms make substantive emission-reduction investments while not taking public credit): Disclosure costs are so high that they would rather be thought of as not green enough. They also illustrated settings in which the right metric, paired with the right amount of investor pressure, can harmonize divergent public interests. They conclude that using an intensity-based target as the environmental assessment metric may channel investor pressure into more environmentally responsible outcomes.

The working paper, Greenness and Its Discontents: Operational Implications of Investor Pressure, is authored by Uzunlar, N, Scheller-Wolf, A, and Tayur, S, all of Carnegie Mellon University. Copyright 2024. All rights reserved.

March 2024

Peter Boatwright, Allan D. Shocker Professor of Marketing and New Product Development; Director of the Integrated Innovation Institute

Businesspeople today, especially those who are more senior, have seen a significant number of new products and services in their work (e.g., smartphones, Airbnb, artificial intelligence). This is partly because so many products have been digitized and partly because more companies recognize the value of innovation. In a new book, two researchers with a combined 50 years of experience across a wide range of industries and target markets, offer 13 tips on how to manage innovation teams to optimize success rates. Because these teams work in complex territory, going where none have ventured before, managers need predictable, reliable approaches to managing teams that operate in the realm of the unpredictable, the unplannable, and the fundamentally unmanageable. The tips are written from a lens of real-world application and accompanied by stories that illustrate key points. Among the 13 tips are chapters on managing the process rather than the solution, choosing the right team, clearly defining criteria for success, empowering team members to lead, managing technology trends, and using AI.

The book, Managing the Unmanageable: 13 Tips for Building and Leading a Successful Innovation Team, is authored by Cagan, J, and Boatwright, P, both of Carnegie Mellon University, and published by Rivertowns Books. Copyright 2024. All rights reserved.

February 2024

Sridhar Tayur, Ford Distinguished Research Chair; University Professor of Operations Management 

After a sudden catastrophe (e.g., an earthquake, a hurricane), first responders try to reach and rescue immobile victims promptly. At the same time, drivers of cars and other vehicles take roads to evacuate the affected region, access medical facilities or shelters, or reunite with their relatives. The escalated traffic congestion significantly hinders critical first responders’ operations if they share some of the same roads. In a new study on disaster preparedness, researchers evaluated a proposal from the Turkish Ministry of Transportation and Infrastructure to allocate a subset of roads—clearly marked and communicated to residents—to be used only by first responders. The study developed a non-linear objective binary constrained optimization (NOBCO) model and developed a novel bi-level algorithm that is inspired by earlier work at the Tepper Quantum Group. The authors deemed the approach promising when compared with the state-of-the-art method currently being used, by testing the bi-level quantum inspired algorithms on a variety of benchmark problems, including three scenarios based on data from predicted Istanbul earthquake. They encouraged applying such quantum-inspired methods to other disaster preparedness problems as well to other applications such as managing large scale sporting events (Asian Games, Olympics) where some road segments of the city are pre-determined and reserved.

The article, A Quantum Inspired Bi-level Optimization Algorithm for the First Responder Network Design Problem, appears on arXiv and is authored by Karahalios, A (Carnegie Mellon University), Tayur, S (Carnegie Mellon University), Tenneti, A (Carnegie Mellon Univesity), Pashapour, A (Koc University), Salman, S (Koc University), and Yildiz, B (Koc University). Copyright 2024. All rights reserved.

January 2024

Benjamin Moseley, Carnegie Bosch Associate Professor of Operations Research

The modern era is witnessing a revolution in the ability to scale computations to massively large data sets. A key breakthrough in scalability was the introduction of fast and easy-to-use distributed programming models such as MapReduce and Spark. These distributed computing frameworks are regularly used to mine information from large data sources (e.g., Internet searches, shopping online, and social media). In this article, researchers cover the Massively Parallel Programming Model of Computation (MPC), a model used to develop algorithms that can run on frameworks such as MaprReduce and Spartk. The topics such as partitioning, coreset construction, sample and prune techniques, dynamic programming, round compression, and the establishment of lower bounds within the MPC model. In so doing, they showcase how these algorithmic tools can serve as building blocks for developing new algorithms in massively distributed computing environments, highlighting the practical relevance and theoretical implications of MPC in solving large-scale computational problems efficiently.

The article, Massively Parallel Computation: Algorithms and Applications, appears in "Foundations and Trends® in Optimization" and is authored by Im, S (University of California, Merced), Kumar, R (Google, Mountain View), Lattanzi, S (Google, Barcelona), Moseley, B (Carnegie Mellon University), and Vassilvitskii, S (Google, New York). Copyright 2024. All rights reserved.