Cloud Trailblazers: 10 for 2013

Mission possible? Jean Yang

Jean Yang seems like she’s on the run, even when she’s merely walking around MIT’s idiosyncratic Stata Center, home of the school’s Computer Science and Artificial Intelligence Lab (CSAIL) where she is a Ph.D candidate in computer sciences.

Her thesis project takes on a very tricky proposition — figuring out how to enforce privacy in the cyberworld. There are guys out there searching for your social security number or any personal details about your life that could help them obtain access to your bank account or other sensitive personal information. Yet in the world of social media, security gets even trickier because the very reason Facebook and Twitter exist is so users can share information. Those users want their information visible on a need-to-know basis to their friends and family, but off-limits to stalkers, exes and assorted creeps.

That poses a huge programming problem that Yang would very much like to solve. She thinks the programming language — actually, a framework —  she’s working on could enforce privacy rules in the workflow of a program. That would mean programmers could focus on cool new features instead of having to constantly review and tweak existing code to enforce or implement privacy settings as they go.


Yang explains: “If you look at Facebook, if you have my GPS location, my mom can see it to make sure I’m safe, and my friends can see it, and maybe everyone else can just see what city I’m in — all based on my settings.”

Ask Jeeves… to lock your social network down

The issue is that folks who are not your friends — maybe your former boss who’s a real tool — can triangulate from stuff your friends post, see pictures of you or even locate you based on your friends’ locations and posts. Even Facebook CEO Mark Zuckerberg was afflicted by this problem when a glitch made his “private” photos public two years ago.

Beneath the covers, a lot of complex stuff is going on. The problem is that all this complex privacy stuff is, in essence, still managed by hand, when it should be pushed into a language and a runtime that can handle it. It’s not unlike the old days when people had to manage memory manually until the systems were mature enough to flush memory automatically.

“If you start doing computations on who is where and who has access to what, the programmer has to track where the data is coming from and going to. It ties their hands. You have all these engineers not rolling out features as fast as they should because they’re dealing with all this,” Yang said.

This challenge is why Yang and her colleagues are building the Jeeves programming language. Their work here is why we named Yang to our Cloud Trailblazers list. Jeeves makes sure that what the user sees about you is based on who that person is to you. A family member could see everything, a person unknown to you would basically see a bunch of stuff about “anonymous.”

Jeeves, according to Yang is:

” … a programming language for automatically enforcing information flow policies for privacy …  Jeeves allows the programmer to implement rules for the flow of values separately from the rest of the functionality. The Jeeves runtime is responsible for producing outputs that adhere to the policies for protecting both reads (confidentiality) and writes (integrity). I have been developing the semantics, theoretical guarantees, and implementation. I plan to demonstrate the feasibility of this approach in web frameworks.”

Running fast in all directions

Yang is also very interested in making computer science and MIT more female-friendly. In 2009, she helped set up Graduate Women of MIT (GWAMIT) to foster a community of women across disciplines as a sort of mentoring and support group, and it now claims 1,200 members.

“I was fed up with this department women’s group where we ate sushi and did breathing exercises. [At GWAMIT]  we hosted a panel where we talked about how hard it is to collaborate with men. They either try to impress you or they ignore you. It was great to be able to talk about this stuff,” she said. Yang is very engaging in the interview, asking nearly as many questions as she’s asked. If the stereotype of a computer geek is someone who lacks empathy or has trouble communicating, she is an exception.

Busy with GWAMIT, industry conferences and her work, Yang still finds time for hobbies like running. Even with her frenetic lifestyle, Margo Seltzer, Yang’s computer science professor at Harvard, is struck by how balanced Yang is.

Seltzer recalled inviting Yang to run in the Tufts 10K race a few years back. “I didn’t expect her to run with me because I’m old and slow … But what I really didn’t expect is that she would finish 7th in her age group.”

It’s enough to make you want to dislike Yang, except it really is impossible to do so.

—Barb Darrow