Making good things easier: Takeaways from MS&E’s virtual reunion event

MS&E alums and faculty spent their Saturday morning learning and reminiscing with one another virtually on January 27, 2024.

Over 300 alums of MS&E and its legacy departments registered for the event, which highlighted the recent work of three MS&E faculty: Bob Sutton, Madeleine Udell, and Johan Ugander. The virtual format allowed alums to join from around the world; participants checked in from Vancouver to São Paulo, and from Athens to Honolulu.

Fortinet Founders Chair of MS&E Pamela Hinds kicked off the event by sharing the state of the department.

A few highlights from her talk:

• Strategic planning. Faculty created a refreshed mission statement for MS&E: We create solutions to pressing societal problems by integrating and pushing the frontiers of operations research, economics, and organization science.
• Research alignment. The department has identified four overlapping research themes in which we have impact and expertise. We wrote about the themes in detail in a recent newsletter.
• Where our graduates go. Undergraduates tend to work in financial services and consulting roles after graduation, and most master’s students enter engineering and financial roles. Learn more in our latest employment report.

Faculty lightning talks

Three faculty then shared short talks about their latest research. Written descriptions are below, and videos will be available soon, but only for a short time. If you don’t already receive the MS&E Newsletter, make sure to subscribe so you don’t miss the videos!

Bob Sutton: The Friction Project

Professor Emeritus Bob Sutton began the lightning talks by sharing insights from his latest book, The Friction Project, which is coauthored with professor Hayagreeva "Huggy" Rao of the Stanford Graduate School of Business.

Professor Sutton started by highlighting the impact of bad friction in organizations, drawing on the example of a 1,000+ question form used by a state government to assess residents' needs for various benefits. By changing the form and asking only the essential questions, the state was able to make the form 80% shorter, which saved time not only for the form-fillers but also the employees processing the forms.

Prof. Sutton then pivoted to instances where introducing friction can be a good thing. He gave the example of a tech company that put job applicants through up to 25 interviews before making a hiring decision. To combat this friction-fraught process, a leader at the company introduced a new type of friction: If you want to interview the same person more than 4 times, you need to get written permission from me personally. The number of interviews per applicant dropped dramatically.

Madeleine Udell: AI and the Future of Optimization

Next, Assistant Professor Madeleine Udell shared her latest research that aims to automatically model optimization problems using large language models (LLMs).

The impetus for this work stems from the practicality of optimization work in the real world. Small businesses often don't have the resources to hire an optimization specialist, even though their expertise could be valuable for business’s growth and success. And most optimization models require more expertise to utilize than the average business owner has.

Enter OptiMUS, a new tool being developed by Prof. Udell and her research team. The tool draws on the technology behind tools like ChatGPT, but aims to reduce the errors that general-purpose tools return when attempting to solve optimization problems—things like hallucinations or forgetting to include parameters set in the middle of a long string of instructions.

In testing, OptiMUS has been able to solve medium-complexity problems almost twice as accurately as general-purpose LLMs, and can solve very complex problems around 14 times as accurately. Prof. Udell celebrated the advancements, while cautioning that the technology is not ready for high-stakes situations yet.

Johan Ugander: Auditing Social Algorithms

In the final lightning talk, Associate Professor Johan Ugander shared his research on how to audit the impact of algorithms behind social media platforms.

Professor Ugander started by describing the many ways in which algorithms shape contemporary society, from recommendations of which news to consume, to what products to buy, and even which people to date. He then discussed the difficulties inherent to auditing the complex and constantly-evolving algorithms that issue those recommendations.

One example of the impact of recommendation algorithms deals with the spread of false news. Prof. Ugander shared how false news doesn't necessarily spread in a different manner than accurate news, but it often travels farther and wider. This makes it difficult for algorithms to detect when a story might be false, because the mechanisms of its spread are the same as an accurate story.

Prof. Ugander closed his talk by discussing the ways in which platforms combat the spread of misinformation, including the "community notes" feature on X (formerly Twitter). This feature is part of an emerging generation of "bridging" algorithms that include humans in the loop with AI-powered systems—exactly the type of intersection that MS&E specializes in.

We welcome your feedback

If you attended the virtual reunion event, we would appreciate your feedback on the event! Please take our quick ~2 minute survey available here.