ORIE celebrates its 50th anniversary with talks, panel discussions, reminiscences, a gala dinner, and a quiz show

Event marks the academic year in which Operations Research first became part of the School’s name.

Alumni, faculty, students and guests assembled in April for a day-long commemoration of a time when Operations Research came into its own at Cornell.  The day featured talks, panel discussions, reminiscences, a trivia quiz, and a gala dinner.  Video recordings of the sessions are available on the ORIE YouTube channel and are individually  linked from this news article.  Articles and additional photos from the event are included in the Spring 2016 ORIE Magazine. 

The celebration was kicked off by the 15th Bangs Lecture, delivered by Theresa Wise ’92 Ph.D. ’95, retired Chief Information Officer of Delta Airlines, who traced the evolution of Operations Research in the airline industry.  The Bangs Lecture was established in 1990, at the midpoint of the past 50 years, to honor John R. Bangs, Jr., ME 21, who pioneered the administrative engineering program that led to today’s ORIE.   Wise provided a personal perspective on the ways in which crew scheduling, revenue management, technology integration in merged airlines, and data analytics are understood, solved, and put into action at Delta and elsewhere in the industry. 

Greeting the assemblage, Cornell Engineering Dean Lance Collins noted that the mid-1960’s were a time of change, marked by the establishment of Computer Science, Materials Science and ORIE.  The celebration comes at an interesting time marked by “a real expansion of opportunities in data analytics, with ORIE  in the center of it all.” Collins reported that in Ithaca, ORIE has had “tremendous hiring in a fiercely competitive world for OR faculty, bringing about a new generation” replacing faculty from the 60’s and 70’s who retire.  At the new Cornell Tech in New York, “ORIE is a key element, with the first Master of Engineering program on the New York campus, which one year from now moves to Roosevelt Island.”

Talks by Brenda Dietrich PhD ’86, Retsev Levi PhD ’05 and Peter Cramton ’80, respectively, took stock of the role of Operations Research in successive waves of computer technology, the risk associated with food imports, and the successes, failures, and future of markets he is involved in designing.  Talks by three retired ORIE professors, Bill “Max” Maxwell, Lee Schruben, and Narihari “Uma”Prabhu, captured key elements of the past half-century from a personal perspective.  Two panel discussions, one comparing the experiences of different generations of students in the same families and the other looking back and forward at the impact of OR and analytics on industry, positioned the 50th anniversary as a point in time.  A trivia quiz, moderated by Professor Emeritus Jack Muckstadt (with Director Shmoys as his Vanna White) provided levity as he tested  attendees’ ability to recognize faculty and students in old photographs and to answer questions such as “who wrote the thesis with the shortest title” (Ans: Chris Jones ’80).   At the gala dinner, ORIE Director David Shmoys announced a major gift, by Helen and Arthur Geoffrion '59 MIE '61,  to endow a Professor of Practice as part of a new structure to increase student interaction with industry. 

Reacting to the event, Sam Mallette ’80 said “in some ways I sort of felt like an ant walking in a redwood forest when I see and listen to all of the ’giants‘ from our field and hear about their various accomplishments!”   “The network of people associated with Cornell ORIE is pretty awesome” said Eric Laub ’81 M.Eng. ’83.  “It was a pleasure to revisit Cornell for ORIE’s 50th anniversary, especially to hear so many stories of OR careers and OR work.”

Talks

Theresa Wise PhD ’96, who just stepped down this year as Chief Information Officer and Senior Vice President of Delta Airlines, gave the 15th Bangs Lecture, which served as a “kick-off” event for the 50th anniversary celebration. Wise described industry trends for the airlines, while interspersing these more general comments with stories from her own personal trajectory. She stressed that the most elegant OR models might not always win, but “they can win, particularly if we structure them, implement them, and most importantly communicate about them effectively.” She credited the use of OR models as being crucial for the success of the airline, which today has 80,000 employees, 330 destinations, 11 hub airports, and over 15,000 flights per day. Furthermore, she viewed the OR model-based approach as being instrumental in the smooth merging of Delta and Northwest Airlines (in which she played the critical role of integrating the information systems of the two airlines).

Wise reflected back on her first summer internship at Northwest Airlines.  She was in the middle of her doctoral work at Cornell in Applied Mathematics under the supervision of Professor Leslie Trotter, in the School of Operations Research and (then) Industrial Engineering.  She arrived at Northwest thinking that she would be spending her summer working on models for revenue management -- the idea of creating different prices for seats within the same class of service, based on expected demand and remaining capacity. Although this idea was just taking off in the airline industry, the project at Northwestern was scrapped, and so Wise needed to find a new project on which to work. She settled on thinking about the problem of scheduling the assignment of crews to flights. Most airlines then (and indeed still today) first construct a schedule for all of their planes, which leaves  each leg that a plane flies in need of  a crew to staff that leg. This “crew scheduling” problem is solved on a monthly basis, with each crew member specifying a detailed set of preferences, subject to an intricate set of regulations governing allowed patterns of flights (with constraints such as the maximum number of consecutive hours a crew member can work  as well as a minimum length of time off between tours of duty). Wise showed how to solve these problems effectively via a so-called set partitioning model and developed integer programming methods for its effective solution.

Wise moved from heading operations research at Northwestern, to taking a lead role in their IT department, eventually becoming the Chief Information Officer (CIO) of Northwestern, before becoming CIO for the merged Delta Airlines. Wise talked extensively about being the CIO while leading the IT merger of the two airlines. She highlighted a “wall walk” that helped unify the process where hundreds of different sub-projects were required to integrate the two IT systems. This wall walk was based on a long wall (at right) completely filled with “post-it” notes delineating the state of different aspects of the overall project. Wise also reflected on the changing technologies that drove many of those changes, particularly noting the way that cell phones  enable direct communication with their customer base both for routine operations, as well as for providing information to the passenger in a time of service “irregularities.” Wise talked about the increasing role of “big data” in the operations at Delta, citing that, for example, the annual growth in data stored for analytics is over 40%,  and concluded by stating that OR has been an invaluable force for business change and results at Delta.

Brenda Dietrich, PhD ’86, who is Vice President of Data Science in IBM’s Business Analytics organization, “has led IBM’s thrust in analytics for years,” said ORIE Director David Shmoys in his introduction.  Dietrich described “How OR fits into the larger information technology (IT) picture, and things that we as practitioners can either take advantage of the trends or get run over by them.” 

Dietrich traced the evolution of IT from transaction processing, through personal computing, to mobile, social, cloud, and ‘big data’ computing, noting that at each stage, computing activities leave behind an ‘exhaust’ of data including transaction records, search trajectories, and transmissions from apps that broadcast cellphone locations.  In the 80’s and 90’s, data like this was used by OR analysts primarily for forecasting.  The PC brought ease of access, but the data was left on the desk and disks of the individual analyst.  The internet provided a trail of data associated with searches and orders.  Now, new devices and apps are generating vast quantities of data, which use of the cloud makes economically feasible to collect and analyze.

She reported that, as ways are sought to “monetize data sets,” there are now markets for them as well as established supply chains.  However “we can have all the data in the world but if we don’t implement the decisions that are made from it, it is worthless,” she asserted.   OR analysts can contribute to improving the world through influencing the choices people make by the way we present them with information.  The potential lies in bringing together data from multiple sources to create “systems of insight.” Using examples from her work at IBM, Dietrich discussed various ways in which data, for example from the “internet of things” or the trajectory of user interactions during a search, can be used to move “from data to insight to action,” and showed how this process offers a wealth of opportunities for operations research. 

Looking back to the last century, computerized inventory systems led to the ability for companies to hold “less stuff,” a significant transformation, Dietrich recalled.   By analogy, the exploitation of sensor data in the internet of things leads to ways to transform business through analyzing data rather than just presenting it.  Moreover, exogenous data, such as weather data, can be exploited to great benefit when tied together with internal operational and customer data. 

Compared with the limited capability of a search engine, cognitive computing, as illustrated by IBM’s Watson system of “Jeopardy” fame, can interact with the decision maker to understand the question, produce possible answers and evidence, analyze the evidence, computes the level of confidence, and deliver the response, evidence and confidence. 

Dietrich described opportunities for operations research in cognitive computing, for example in creating models, locating and navigating evidence, and capturing and incorporating user actions and outcomes in decision support tools in order to learn from the choices that are made.   She placed particular emphasis on the ways in which people interact, using words.  Most current operations research communication involves numbers and equations.  “With ubiquitous computing capacity and data storage available, will this change?” she asked.  She used an example that combines classical operations research techniques with a natural language interface and the potential to learn from the interaction of user and system to illustrate the kind of change that is becoming possible.  

Retsef Levi  PhD ’05. is the J. Spencer Standish (1945) Professor of Operations Management at the MIT Sloan School of Management.  He spoke to the 50th anniversary gathering about the risk of economically motivated adulteration occurring in global food supply chains.  Determining the risk and allocating resources to combat it is a problem that “OR has all of the ingredients needed to analyze: supply chain management, modeling, data, analytics, and making important decisions.”  As part of interdisciplinary work of experts in OR, Biology, and Chemistry focused on China, Levi and his collaborators conjectured that there are two key drivers of adulteration risk: high dispersion of the supply chain, and weak regulatory quality in specific regions in China.

Levi noted that the Chinese farming supply chains are characterized by so-called “dragon-head” companies that consolidate output from a large number of small, high-risk, low-margin farms before passing it along for processing and export.   This dispersed structure is based on small farmers who “have little to lose, and under sufficient pressure are willing to do everything to maintain their income,” Levi pointed out. “Each produces a tiny amount, but for the farmer it is their entire income.”  

By contrast, the Chinese government is encouraging a shift to vertical integration, with large, corporate owned firms that send output directly to processing facilities.   Although such an integrated supply chain structure has a higher risk of disruption, the team used publically available data from U.S. and Chinese sources, a dispersion formula related to entropy, and regression techniques to confirm their hypothesis that the “dragon-head” structure and higher dispersion indeed result in higher risk.  Similarly, they used publically available data on export certification quality inspections, incident reports, and Value Added Tax (VAT) collections to confirm their hypothesis that weaker regulatory quality leads to a higher risk of economically motivated adulteration.   

The project results can be used to prioritize risk at the product, firm and shipment level, and to develop systematic approaches to monitoring and regulations.   

Peter Cramton ’80 is Professor of Economics at the University of Maryland and the European University Institute, and on the International Faculty of the University of Cologne.  A Cornell ORIE major as an undergraduate, he received a PhD in business from Stanford in 1984.   

“Technology has enabled much better markets,” Cramton asserted.  Market designers “try to establish effective rules of market interaction--- and these rules really matter.  What I like to call ‘Economics Engineering’ entails ideas from economics (incentives), operations research (optimization), computer science (algorithms), psychology (behavioral aspects),and general engineering (vectors in communication, transportation, etc.),” he said.  “Good market design -- based on the objectives of efficiency, simplicity, transparency and fairness -- improves allocations and pricing information, and reduces the chance of market failure due to excessive risk or inadequate competition.”  

Cramton discussed the success of designed markets for electricity and telecommunications, and showed how the same principles can improve transportation scheduling, climate policy, and trading of financial securities.

  • Wholesale electricity markets have been transformed by the ideas of open access to transmission capacity and by well-designed products that send the right pricing signals.  Cramton described a successful market in which “every few minutes, computational optimization is used to determine who should be generating how much electricity, who should get it, and how it should be priced.  Underlying market rules translate the preferences of the generators and the preferences of the demand side into an optimization problem with the objective of maximizing social welfare, subject to constraints governed by engineering laws.”   The market optimizes use of existing resources in the short term, and for the longer term, produces pricing that leads to investment in the right quantity and mix of resources, he said.
  • The auctioning of communications spectrum, in which Cramton has also been closely involved, is now well-established, with “current broadcast incentive auctions underway to repurpose TV spectrum to mobile communications – the most complicated auction ever by orders of magnitude, involving combinatorial optimization at every stage.”   As with electricity, telecommunications has evolved from monopoly, through oligopoly (because of the large entry cost), to competition. 
  • In financial markets, Cramton said, “making time discrete can greatly improve market efficiency.”  Currently, financial trades (for example at NASDAQ) are continuously “threaded through” a single computer, with no real computation possible.  This provides “a  prize for whomever is fastest, leading to an arms race with enormous investment and advances in speed, one that does not actually improve capital investment but costs investors.”    With co-   authors, he has argued instead for frequent batch auctions, in which orders are collected over a short interval (say, one-tenth of a second), and the auction can achieve a consensus clearing price.   This “greatly simplifies the market for everybody,” introduces opportunities for operations researchers to use the short interval to compute a clearing price, and transforms competition based on speed to competition based on price. 

Cramton also discussed the potential for market design to improve transportation scheduling and to combat climate change.  Reforms employing improved market designs meet resistance, said Cramton, since there is established money being made from inefficiencies.   Nonetheless, he is “quite confident” that designed markets will eventually come to pass, noting that already several Cornell faculty members are currently involved in market design research.

Panel Discussions

Multigenerational ORIE Families  

“One of the nice things about being around for fifty years is that ORIE has been host for quite a large number of families – upwards of a dozen – with  parent-and child pairs who both have OR degrees,” said ORIE Director Shmoys in introducing a panel representing such families.  He asked panel members to comment on differences and similarities among ORIE programs for the different generations, having provided lists of the courses they had taken.   Many of the panelists discussed different individual ways in which they had been able to broaden and diversify their undergraduate ORIE experience, demonstrating how the capabilities to do so have evolved over the years.      

Bill Wiberg ’81 and his daughter Holly, who is in this year’s graduating class, comprise an ORIE pair family.  Bill noted that prior to coming to Cornell he had never heard the term “operations research.”  Today, in his work as a venture capital investor the term “data analytics” comes up every day.  “My guess is that when people come to Cornell now, they are much better equipped to understand what OR is,” he said.   

Comparing her course list with her father’s, Holly Wiberg noted that, for ORIE students in her father’s cohort “there was much more emphasis on traditional engineering fields.”  Not having to fulfill so many underclass engineering requirements “has worked to my advantage,” she said.  It enabled her to do work related to specific applications of computer science and OR, for example building models in such areas as health care, baseball, and bike sharing programs, as well as learning how to handle extremely large and “noisy” data sets.  She expects this experience to be valuable in her work at Athena Health after graduation.     

For Amy Paull ’12, who works at Freddie Mac in McLean VA and like her father participated in the panel via a remote link, the fact that she could follow a more flexible curriculum than her father enabled her to take a semester abroad, at the Hong Kong University of Science and Technology.   However her father Elliot ’Paull 77 noted that even with more required engineering courses than Amy he had been able to diversify by taking courses in five of Cornell’s colleges.  That experience helped put engineering in perspective by showing how important data is to solving real problems, while his OR courses equipped him well in taking data and relating it to business problems, during his career at Microsoft and subsequently as an independent consultant, “working on cost analysis stuff that I learned back as an undergraduate.”

In comparing her ORIE education with her father’s, Amy noted that the foundational courses, such as Math 293 and the core OR courses, have remained in the curriculum, even having the same course numbers.   “I have been able to take courses that were complementary to the engineering and operations research courses,” she said.  I’m glad that the trend of using your operations research understanding in a very multidimensional kind of way continues.”  

Henry Shum ’83 PhD ’89 was a member of the panel together with his two godsons, Adrian Wu and Nathan Ngan, both seniors who will be pursuing the M.Eng. in Financial Engineering next year.  Shum provided some insight into the compartmentalization of theory and practice that occurred in the past, by recalling that in his first job, with the International Paper Company, “I never imagined that in real life there is such a thing as the ‘cutting stock problem,’ despite having been introduced to this famous operations research problem in Professor Leslie Trotter’s class,” he said.  “But that is what I worked on for the next five or six years!”   

Wu noted that, based on comparing course lists, he had been able to take a different approach from Shum, by taking courses in the Johnson Graduate School of Management as a way of combining theory and practice.   “It was interesting that I could apply things that I learned in OR and draw a connection” to topics encountered in his business courses.  “My classes in OR helped me understand the nitty-gritty of how to make these things actually work,” he said.   Ngan (right) decided to take the Financial Engineering program because “it gives me the best of both worlds in terms of academic and also career exposure.” 

Asked by Brooke Schumm ’77 what ORIE can do to add value for a student coming here versus going elsewhere, Shmoys said that a significant differentiator would be getting additional resources to be able to systematically provide hands-on experience beyond what has been available to only about 20% of the students, including some of the panelists.     

OR, Analytics and their Impact on Industry, Looking Back & Looking Forward  

Drawing on a sample of the wealth of experience of ORIE alumni in a range of industries, Director Shmoys convened a panel consisting of Jeff Goldman (Procter & Gamble), Jamie Hintlian (Ernst & Young), Chris Jones (Marchex), Radhika Kulkarni (SAS), Reha Tütuncü (AQR Capital) and Theresa Wise (Delta Airlines) to provide insights into the evolution of the tools and science underlying what OR has brought to their industry and where things are going in the next decades or so.

Hintlian ’82, M.Eng. 85, MBA ’86 leads the Life Science Supply Chain practice at Ernst & Young (E&Y).  He discussed E&Y calls “the digital operations agenda,” which Hintlian said covers three developments: intelligent machines (plant equipment with sensors and products with electronic serial numbers that can be traced in real time), advanced analytics that is “literally at our fingertips,” and “connecting people with work,” with such capabilities as visualization.  He noted that the life science supply chain is much like that in other manufacturing industries, but must operate within a regime of regulation, which now requires tracking and tracing of materials from manufacturer to point of dispensing, as well as a new FDA requirement for reporting of all deviations from the approved “recipe” for the manufacturing supply chain.  Without analytics, these requirements are only useful to management “if you like driving with only the rear view mirror.”  But now being able to bring together datasets from intelligent machines and other systems allows leadership to use predictive analytics to anticipate where issues might arise, and connects line workers with their work, empowering them to understand and make effective use of available information.  

Kulkarni  PhD ’81, Vice President of Advanced Analytics R&D at analytics solutions company SAS, traced the growth at SAS of operations research, which accelerated dramatically some years ago when the company moved from providing statistical software to become a provider of business solutions.  “The CEO recognized that optimization would be an enabler of success in providing solutions,” she said, and now more than 200 people, including more than 20 PhDs, work in OR at SAS.  Looking forward, “Recent developments are making OR even more critical in Data Analytics,” she said.   “Problems that couldn’t be solved before now can be solved” due to the increase in economical computing power. 

Jones ’80, M.Eng. ’81 PhD ’85, who has worked as a software developer for Amazon, a supply chain architect for Aspen Technologies, and an associate professor at Simon Fraser, is now a principal software developer for Marchex, a company that does “call analytics.” He noted that while Moore’s Law may be ending, networked and parallel computing and even quantum computing may take its place as a driver of scale.  Computational tools are better, but it still remains “hard to build a model that is correct, current, and complete,” asking “can we get to a Moore’s Law with respect to modeling?”  As an indication of where analytical computing power is leading he said that Marchex currently joins together records of 1 million telephone calls and 1 billion web clicks to “trace the customer journey leading up to a purchase.” 

Goldman ’97 M.Eng. ’98 is Associate Director of Enterprise Data Science at Procter & Gamble (P&G) after serving as analytics advisor to top P&G executives.   While he agreed with the remarks of others on the role of technology he said that his view of operations research has changed.  When he left Cornell, he saw “the core of operations research was helping organizations make good decisions based on science.” Now he sees the “ability to analyze business data, make unbiased recommendations on the state of the business, what likely competitive activity might be, and how best to respond to different competitive activity, while not directly operations research in themselves” is most likely to come from people with an operations research background, so the challenge for operations research is “to have a seat at the decision table with the senior executive.”  Goldman sees “analytics as the future of industry and of our own field.  The more we can focus the people coming out of school on understanding business aspects and how the combination of analytic mastery, analytic depth and business understanding provide a unique combination that will allow our profession to be differentiated on the world stage.”

Wise PhD ’95, who delivered the Bangs Lecture at the outset of the celebration, noted that in the airline industry “there has been tremendous change but the types of things we look at and the underlying approaches are persistent.”  However “there is more work to be done,” she said, for example in predictive analytics, in coping with messy data, and in seeking customer input when as the airline continuously refines systems that cope  with “ irregular operations,” i.e. “putting things back together after disruptions from weather and operational problems.   Wise credits success at Northwest and Delta to getting OR and IT people to work together:  “Fully leveraging OR involved good IT,” she said. 

In 2005 Tütuncü, PhD ’96, became a “quant” on Wall Street after 10 years as a mathematics professor at Carnegie Mellon University.  He recently moved from Goldman Sachs to AQR Capital Management.   Reviewing his personal history as well as that of OR in finance, he said that when he joined the industry, quants were considered as problem solvers providing a service, and there was a failure to recognize the full value of an OR background, with some people blaming models and software for failures that were actually due to their limited OR knowledge or to inadequate forecasting.  The introduction of new academic Financial Engineering programs, in which ORIE was a pioneer, led to a proliferation of people with OR-based Financial Engineering degrees in the industry, which helped improve the situation.  Later, the introduction of “new financial products in the securitization space” some of which incorporated bad models that led to bad decisions, gave a bad name to the quant culture, he said, leading to an effort by OR quants to educate people about problems in their use of the products. (Even before the financial crisis, academics such as the late David Heath, then at Cornell, had worked to improve the measurement of risk, he said).  Looking forward, he noted that while innovation may have slowed down, there is potential “in the advanced analytics space, for example turni

Other Articles of Interest