Wednesday, January 30, 2013

Yoav Shapira, SDM '05: From SDM, Around the World, to Happier.com

By Tom Kadala

Yoav Shapira
SDM alum Yoav Shapira is the type of person everyone would like as a friend, advisor, and colleague—especially when deciphering, debugging, and deploying a complex software development project. He is an engineer by training and a manager of talent at heart.

Shapira's pursuit for a graduate degree began after working at a biotech firm in Boston. At the time his goal was to complement his management experiences with proven academic theories.

In choosing between a masters program from Harvard or MIT, Shapira rationalized his decision in favor of MIT's SDM program as the one offering the best of both worlds, engineering and management. The SDM core course requirements were relatively few, which enabled him to customize his program by giving him unprecedented access to MIT's School of Engineering and Sloan School of Management, as well as to courses at Harvard. He could take the classes he had always dreamed of, with professors he had often read about, and with a cohort of early-to-mid-career professionals that matched his brilliance and his passion for learning.

Shapira's all-time favorite systems thinking course at SDM was 'The Human-Side of Technology' taught by Senior Lecturer Ralph Katz. "I simply loved his class and enjoyed every minute of it. The professor was awesome and my peers were amazing," he said. Initially Yoav thought that after receiving his master's degree he would return to his former life as a software engineer. To his surprise, however, SDM unleashed his inner curiosity as well as a hidden, burning 'entrepreneurial bug'.

"I was supposed to return to my company after the SDM program but instead wound up joining a tiny startup called HubSpot with a group of Sloan students I had met in one of my classes," he explained.

Shapira grew HubSpot's software development team from a staff of four in 2006 to 40 by 2010. The team is now closer to 70. "There was no easy street here. Every month was a flog until 2008, when the firm reached a critical threshold of about 1,000 qualified leads." From then on, signup numbers began to grow exponentially and the number of company employees overall grew from a handful to hundreds.

Shapira's SDM education and his instinctive management abilities aided him in the unduly task of hiring and managing multiple teams of software engineers. "I hired engineers who were open-minded and could work in an agile environment", said Shapira. His management style earned him high marks among his peers for always mentoring, challenging, and nurturing with contagious enthusiasm and passion. His secret was simple. He would offer suggestions then step away to give his talented engineers the space, time, and confidence they needed to come up with their own solutions.

In 2012, Shapira left HubSpot to fulfill a lifetime dream—a trip around the world. During his one-year tour, he visited Japan, Hong Kong, Korea, Thailand, India, Israel, Turkey, Russia, Sweden, Denmark, Norway, Germany, Czech Republic, Italy, Spain, Iceland, and England, meeting with MIT alumni and other like-minded individuals who shared a passion for making the earth a better place.

Shortly after returning from his world tour, he was offered the CTO position for another startup Happier.com. With the HubSpot experience fresh on his mind, Shapira felt confident and ready to take on this next entrepreneurial challenge. "This venture will be different from HubSpot," he said. "Happier.com is a purpose-driven startup based on a lot of scientific research into human psychology and related behaviors. We all have things that make us happier, and most of us don't do enough of them. Can this product help? Can we put more smiles on more people's faces? I'd love to try. It is a life-long dream that truly defines me."


Tuesday, January 29, 2013

Dan Braha: Applying Complex Systems Theory to Real World Data

By Lynne Weiss

Dan Braha
Dan Braha
Dr. Dan Braha, visiting professor in MIT's Engineering Systems Division, has spent his career applying complex systems theory to engineering systems, biological systems, financial systems, and product development systems. He will offer his insights in a February 11 SDM Systems Thinking Webinar titled "From Politics and Finance to Power Grids and Products: Addressing Complexity in the Interconnected World."

Braha, a co-faculty of the New England Complex Systems Institute (NECSI) and a full professor at the University of Massachusetts, Dartmouth, began his career in engineering design, operations research, and supply chain management. He has also contributed research to semiconductor manufacturing, data mining, and artificial intelligence. His transition to complex systems research has been continuous and gradual, he said, dating his interest to 1993, when he started exploring statistical physics in the context of large-scale engineering design. The move to complexity theory has shifted the focus of his research from only looking for "the best solution," to trying to "understand how systems behave—whether engineering systems, product development systems, or social networks."

In his webinar, Braha plans to discuss what he said are four basic characteristics of complex systems:
  1. Universality. "You can describe many systems across domains and you will find universal properties. The same underlying principles can describe the evolution of language, the evolution of species, and the evolution of companies."
  2. Coupling and Connectivity. "Systems can be loosely coupled or they can be tightly coupled. In the context of complex engineering systems, you can change the characteristics of highly connected 'nodes' which could serve as leverage points for drastically improving the performance of the system. For example, making highly connected components of a piece of software less dependent on each other could dramatically decrease the number of defects in open source software development. In the context of financial networks, regulatory requirements could be set higher for banks that carry the highest risk to the system."
  3. Phased Transition. "The complex systems community looks for signals that a system is about to go into a phase transition. For example, they want to find signals that the economy is on the edge of transition. Think about product development. We can have a state where everything is stable, on time, on budget, but if one element becomes unstable, the whole system goes out of control." Braha said that too much stability is not necessarily ideal. "To increase innovation, you want to be on the border. At the edge of chaos, innovation goes up, but you also expose yourself to vulnerabilities."
  4. Out-of-Equilibrium Dynamics. Braha suggested that a mayor who wants to lower a city's crime rate has to think about internal influences—among the people committing the crimes—and external influences—for example, the strength of the police force or the health of the educational system. "We have developed mathematical models to describe these dynamics in finance and other systems."
The value of complexity research to industry lies in the potential for prediction, but the accuracy of prediction depends on understanding the underlying principles. People focus on what is going on in their immediate environment, but "If you step back," Braha said, "you see more connections. My goal is to understand the big picture—how all of us are connected and how one person's behavior can affect someone else down the road."


Thursday, January 17, 2013

Evaluating Complex Dynamic System Architectures: NASCAR Chassis Setup and Development

By Scott Ahlman, SDM '01

Editor's note: Scott Ahlman will be delivering a webinar on a related topic on January 28, 2013.
Free and open to all. Details/registration.

Scott Ahlman
For several years, I was an independent contractor for Ford Racing. I helped enhance the performance of the five drivers on Ford's two leading NASCAR Sprint Cup Series teams, Roush Fenway Racing and Richard Petty Motorsports. As a chassis systems engineer, I juggled a host of variables that affect a race car's performance, balance, drivability, and tire life. It's a complex system that requires in-depth preparation and split-second decisions. It also varies from track to track and driver to driver.

Over the past five years I've been part of 20 wins, numerous pole positions, and two second place finishes in championships. My studies at the SDM program as a Ford-sponsored fellow contributed significantly to my success. The program taught me to view complex challenges through technical, business, and socio-political lenses and to integrate these perspectives using systems thinking.

The complex system encompassing a high-performance race car, driver, and track involves literally thousands of parameters and variables — primarily car design and setup parameters, vehicle dynamic behavior variables and weather related variables including temperature, barometric pressure, track temperature, and resultant air density. These must be prioritized, analyzed, understood, and adjusted. The system also involves a team of engineers and mechanics.

While in SDM, I developed a hierarchical approach to combining tools and methods for evaluating complex, dynamic, system architectures. My thesis research was conducted in the context of a very high-performance passenger car.

My most recent work focused on chassis and vehicle dynamics. I led the development and use of predictive tools to determine the best chassis setup for a fast, drivable, and consistent race car. I also contributed to chassis development, including target setting, analysis, design, manufacture, and refinement on and off the track.
Figure 1: This concept map of high-performance
passenger car handling shows some of the many parameters
and subsystems that chassis engineers work with.
The optimal chassis differs from race to race. Even at the same track, driver tendencies, vehicle components, and weather conditions differ over time. Preparing a car's chassis for a race involves many adjustments, but given tight constraints the team's goal is to minimize the number and degree of adjustments.

Race weekend is really, really fast. We've done extensive analysis beforehand, and created a short book of recommendations on chassis setup and prioritized adjustments, depending on various handling issues. Planning is essential. When a driver came in from a run and gave us feedback on the car and how it needs to behave differently, we had 15 to 30 seconds to make a judgment call. It was far better to be in predictive or refinement mode than to conduct analyses while the driver was on the track.

Adding to the challenge, there was very limited practice time at a specific track on race weekend — typically 1.5 hours of practice before qualifying and then two one-hour practices on the day before the race. We therefore used system engineering to predict the best race car setup for the different tracks and the infinite number of weather possibilities in a racing season. NASCAR rules preclude testing or practice at any track or facility that sanctions a NASCAR race, so our ability to predict the race car's behavior and make chassis setup recommendations was critical.

The challenges in understanding the variables and race conditions went beyond just car, driver, track layout and track surface. Here are a few of the other variables:
  • Annual tire changes by Goodyear. We were provided lab test data for new tires that we used for analysis and modeling to understand how the changes would affect the car's behavior. However, we were only allowed to participate in Goodyear field tests on a limited basis — about five tracks we raced on a year. We had to be purely predictive for the remaining tracks.
  • Constantly changing conditions. Ambient temperature, sun load on the track, tire rubber buildup on the track, tire wear, and changing fuel load add numerous levels of unpredictability.
  • Limited data acquisition during races. Only driver feedback on the car's behavior is allowed. Therefore, the quality of his feedback and our ability to prompt him for information is crucial.
Parsing all of this and more requires multi-variable and response optimization, which is both a science and an art. It involves applying weighting factors — i.e. determining the relative importance of the elements of a system — via various metrics and models.
Figure 2: This chart shows part of an optimization sheet.
It helps engineers focus on the right things in the right
order and to complete difficult optimization of many
parameters and variables at once, as a system.
As statistician George Edward Pelham Box said, "All models are false, but some models are useful". Knowing which parts of a model are "false" and which are useful is key. To achieve this, the team needed to determine:
  • Which metrics to use and when to use them, since vehicle dynamics models and analyses rarely output holistically accurate values for speed, balance, balance consistency, drivability and consistency.
  • The strengths and weaknesses of various models, when to use which, and which could be relied upon at face value, and which require human judgment.
Balance and balance consistency are central to the challenge of developing and setting up a race car's chassis. A car's balance relates to which end loses its grip first. The ideal car uses all four tires fairly equally (neutral balance), providing the most grip and therefore the highest speed. But a car with neutral balance can be difficult to drive because it's close to the edge of control. In addition, varying track conditions make this ideally neutral balance difficult to achieve and maintain for any length of time, especially over a 500-mile race.

With a "tight" or under-steering race car, the front tires lose grip first, resulting in the car tending to continue straight on a path tangent to the curve when pushed to the limit. A "loose" or over-steering car is one in which the rear tires lose grip first, resulting in the car tending to spin out when pushed to the limit.

In many cases, the three major parts of a corner — entry, middle and exit — can all exhibit very different balance characteristics due to the nature of the part of the corner, the driver's style, the car's inherent behavior, track geometry changes (banking, lateral and normal curvature), and bumps in the track. Therefore, balance consistency throughout the corner is one of our most important — and challenging — criteria for chassis design and setup.

Working with NASCAR chassis was a great example of the need to define, then design. The numerous variables and tight deadlines left the process vulnerable to classic errors that arose in the absence of a systems perspective. Lack of clear goals that can be used to generate system, subsystem and component requirements led to a system that failed to meet the customer need. And the lack of clear goals and requirements led to what I call "engineering on walkabout" where one is on a design journey with no clear destination.

And then there were the human factors. Although the stated objective might be winning races and championships, there are sometimes hidden agendas. Often there were also conflicting goals, including focusing on productivity at the expense of standard goals definition and requirements cascades, i.e. how functional requirements "cascade" down a system hierarchy from system through subsystems to components.

Taking a systems perspective helped me recognize these non-technical, and often non-strategic, influences faster and better, assign them more credence, and cope with or even influence and shape them. This has been instrumental in helping "incentivize" my teammates in the desired direction.

A key part of my process in helping the team tackle this complex system was the use of basic systems engineering tools like chunking, aggregation, and hierarchy. In addition, Design of Experiments and many-parameter and variable optimization were also key parts of this process. They helped achieve high-performance results much more efficiently than standard component-focused and typical one-factor at-a-time strategies.

Ultimately, though, understanding what parameters and variables to focus on and when was critical and required significant judgment. In addition, team members needed to know the strengths and weaknesses of the tools and models they used.

This article shows only a few of the ways that SDM knowledge can be applied in a multi-variable environment. For a more in-depth discussion on, for example, testing concepts vs. track-specific setup; the use of systems engineering tools such as chunking, aggregation, and hierarchy; use of solution neutral requirements; the functional requirements cascade; and mind-mapping, please request an electronic copy of my thesis, Complex Dynamic System Architecture Evaluation Through A Hierarchical Synthesis Of Tools And Methods, from Joan S. Rubin, SDM Industry Co-director, jsrubin@mit.edu.

About the Author
Under his firm Ahlman Engineering, Scott Ahlman, SDM '01, was a chassis/vehicle dynamics engineer for Ford Racing in the NASCAR Sprint Cup Series since 2006. He holds an MS in Engineering and Management from MIT.


Friday, January 11, 2013

Utterback honored as exemplar of excellence


James Utterback
James Utterback, David J. McGrath jr (1959) Professor of Management and Innovation, Professor of Engineering Systems, and one of SDM's most popular professors, was one of seven "exemplars of excellence" honored by KU Leuven in June 2012 at its bi-annual Leuven International Forum. KU Leuven, located in Flanders, Belgium, has been a center of learning since it was founded in 1425. The international forum is primarily a networking event that brings together Belgian, European and international leaders from academia, industry, and government for the advancement of knowledge and service to society.

Utterback — lauded as one of the pioneers of research on innovation at and by spin-offs — spoke to attendees about the confluence of different fields such as biotechnology and nanotechnology. He described this fusion of disciplines as necessary, but cautioned that the benefits of interdisciplinary research must not be left to chance and innovators must purposefully steer the process.

KU Leuven reported that in discussing the problematic issues of innovation, Utterback mentioned the link between automation and unemployment, pointing out that 90 per cent of employees still work in established industries. "It is therefore important to invest in new fields in a well-considered and balanced manner." And what is the secret to 'spin-off sauce', as Utterback himself described it? "Young entrepreneurs are attracted to MIT because they know we support their ideas." (Video | Interview | Laudatio)