Story

CIS Dean Greg Morrisett on challenges and opportunities in CIS

by Bill Steele

Greg Morrisett returned to Cornell July 1 as dean of Computing and Information Science (CIS). He had served on Cornell’s computer science faculty from 1996 to 2004 before becoming the Allen B. Cutting Professor of Computer Science at Harvard University, where he was director of the Center for Research on Computation and Society and associate dean for computer science and engineering. He helped establish Harvard’s new School of Engineering and Applied Sciences. Chronicle reporter Bill Steele sat down with Morrisett in his Gates Hall office to discuss the future of CIS.

Coming back to Cornell, what do you notice that has changed?

Ithaca hasn’t changed that much, but Cornell and CIS have changed tremendously. CIS now includes the Department of Statistical Science, which didn’t exist 20 years ago, and the Cornell Tech campus is an exciting opportunity for everyone, but for CIS in particular. Another big change is that CIS has broadened its engagement with so many other disciplines beyond the physical sciences and engineering. We now connect with Johnson, biological sciences, social sciences and even the humanities.

I’d say that we are helping to build and define what 21st-century computing and information science is all about. Programming computers is a small but important piece of this. Just as important is the range of mathematical and computational models, such as network science, that we bring to any field we touch. And the interaction of technology with society is a key focus for CIS. Modern “computation” involves coordinating people as much as computers or networks. For example, we now recruit humans, through online platforms such as the Lab of Ornithology’s eBird, where they help train computers to classify pictures or sounds of birds.

What are your impressions of Gates Hall? 

Gates Hall is a great new space that brings together our computer science and information science faculty and students. It’s symbolic that we are in the center of the campus – both geographically and academically.

One of my favorite features in the building is the Gimme! Coffee shop, which brings lots of outsiders to the building and provides an opportunity for students, faculty and staff to have impromptu conversations. Those conversations are crucial for sparking new ideas and creating a strong sense of community.

Are there things you might want to change?

Because statistics, computer science and information science interact with every field, students from all over the campus are taking our courses. As a result, our classes are huge. I would love for students to have some smaller classes, particularly for the upper levels. But of course, this means adding more faculty.

One of the questions I want to ask is, “What’s the best way to teach a given course, such as Introduction to Programming, in a residential campus setting?” If you sit in a modern classroom, you’ll notice that students have their laptops or phones out and are constantly distracted by the very technology we put in their hands. Furthermore, solutions to just about any homework or project can be easily found online.

We have to rethink the old models for teaching and adapt to the new world where information is at our fingertips. For some courses, that might mean that we “flip” the classroom so that students watch lectures via video in their dorms, and come to class prepared to do hands-on, development work together. But I have yet to find a way to flip a project-oriented class, such as introduction to programming, much less an upper-level class such as operating systems, when there are 300 students. To do a good job requires the right teaching space and enough teaching assistants that we can give students immediate feedback on their designs and code. Regardless, we should see this as an opportunity to combine teaching and research so that we can give the students the best education possible.

Tell us about your own research.

I’m mostly interested in software security: How do we build software that can’t easily be hacked? There are many different pieces that can help, from better programming languages to better cryptography. Ideally, most of the security should happen automatically, without programmers having to worry about it. But today, it’s just the opposite: One little coding mistake in a program can lead to a catastrophic failure.

One approach on which I’m particularly focused is the application of formal, machine-checked proofs of correctness for safety- and security-critical software. The idea is that if I give you a piece of software and claim that it doesn’t do anything “bad,” then you should have an easy way to verify my claim. One way to achieve this is to construct a mathematically rigorous proof that the software doesn’t do anything “bad” when executed, and then you could just check that the proof is valid. Ideally, the proof checker should be really small and simple, and we should even build a proof that this proof checker is correct.

The idea of formal, machine-checked proofs goes back to early work of Bob Constable, who was the first CIS dean. So I’m following in his footsteps in more than one way.

What are your interests outside of work?

I like to hike, which is another reason I’m glad to be back in Ithaca. I love old movies. And I enjoy collecting these cows. [On the shelves in his office are several figurines and mugs with cow images; in one corner of the room is a huge, floppy stuffed cow.]

I don’t buy them. People give them to me.

You started here as a junior faculty member, and now you’ve become dean. How do you feel about that transition?

One of the things I liked about Cornell was that I got lots of support from senior faculty. I think my job now is to act as a general troubleshooter to eliminate a lot of the frustrations for students and faculty, and I want to make sure everyone has the opportunity to be successful.