I am an assistant professor in the Computer Science Department of Columbia University. My research is in human-computer interaction (HCI), crowdsourcing, and computational design.

I build tools enhance people's problem solving ability. The three main approaches are to:

  • organize massive amounts of information and make sense of it
  • design better solutions through brainstorming, synthesis, and iteration
  • communicate complex ideas more easily with visual symbols and illustrative metaphors

Lydia Chilton

Lydia Chilton
Columbia University
Computer Science Department
Google Scholar Page

Computational Design is a grand challenge in HCI

Design is an the process for solving problems. It is used in disiplines ranging from architecture to software to graphic design. We know that experts follow the design process, we don't fully understand how to do it, how to teach it, or why it is successful. Let's break down and rebuild the design process from the ground up.

Start with the assumption that the design process can be viewed computationally: that it has modules that can be isolated and then computationally composed. We can then see design as a search process for transforming an input into an output that meets a desired specification.

Such a model enables:

  • collaboration among large groups,
  • optimization through systematic exploration of the space,
  • avoiding pitfalls such as fixation on one solution to the problem

Two current project I have in the computational design space are:

Constructing visual metaphors for creative ads


A message your ad will convey
"Smoking kills you."
An image which blends two visual symbols.

One represents the subject ("smoking") and one symbol represents the predicate ("kills you").

HumorTools: Writing news satire like The Onion.


A real news headline
"McDonald’s Testing Customizable Burgers to Compete with Chipotle"
A funny man-on-the-street style reaction to the headline.

The reaction should reuse an aspect of the headline ("McDonald's") in an unexpected way (insult McDonald's for always getting your order wrong) and blending that back into another element of the headline (customizable burgers).

Key Insights of Computational Design

Design is a highly creative domain with many excellent solutions for a single design challenge. It seems highly unlikely that there is no simple formula for good design.

However, centuries of thinkers and practitioners have come up with structures they use to describe their process, either abstractly or anecdotally. We can borrow structures and define them more rigorously.

The central model of design according to the the philosophies of design thinking is the notion of flare and focus.

  • The designer consider many aspects of the problem environment before defining the problem.
  • The designer should ideate many ideas before embarking on pursuing a solution.

We borrow the flare and focus model and apply it to both visual blends and humor, using “brainstorming tools” to expand the input into many different ideas, and using "synthesis tools” to combine the pieces we need together.

Constructing visual metaphors for creative ads

HumorTools: Writing news satire like The Onion.

Job Materials

If you are on the job market and looking for examples of application materials, you are welcome to mine.

Job Talk Slides (download)
Research Statement
Teaching Statement
List of References

Past Projects

Cascade: Crowdsourcing Taxonomy Creation
Taxonomies are essential for getting a big picture view on large datasets. Often human insight is needed to find the connections in data, but people find large organization tasks overwhelming. Cascade is an algorithm that crowdsources taxonomy creation by distributing the task into hundreds of easy subtasks. Each worker makes local judgements about data items without needing a global view of the data.

Frenzy: Collaborative Data Organization
Frenzy is a communitysourcing tool that builds on the ideas of Cascade, but which affords more transparency and communication that community members require. We deployed a production version Frenzy to a group of 60 domain experts to categorize conference papers, then group them into sessions. This organizational task involves both crowdsourcing a cohesive picture of a large dataset as well as collectively meeting a global constraint.

Frenzy was deployed at the CSCW 2013 and CHI 2014 program committee meeting to organize the accepted papers into conference sessions.

TurKit: Human Computation Algorithms on Mechanical Turk
TurKit was the first demonstration of crowd algorithms on MTurk. It could solve hard problems like handwriting recognition by allowing workers to build on the insights of others. TurKit inspired and was used to implement Michael Bernstein's Soylent and and Jeff Bigham's VizWiz.


HumorTools: A Microtask Workflow for Generating News Satire
Lydia B. Chilton, Daniel S. Weld, James A. Landay.
In submission. 2016

Frenzy: Collaborative Data Organization for Creating Conference Sessions
Lydia B. Chilton, Juho Kim, Paul Andre, Felicia Cordeiro, James A. Landay , Daniel S. Weld, Steven P. Dow, Robert C. Miller, Haoqi Zhang.
Full Paper at CHI 2014. Honorable Mention for Best Paper

Cascade: Crowdsourcing Taxonomy Creation
Lydia B. Chilton, Greg Little, Darren Edge, Daniel S. Weld, James A. Landay.
Full Paper at CHI 2013.

Community clustering: Leveraging an academic crowd to form coherent conference sessions.
Paul Andre, Haoqi Zhang, Juho Kim, Lydia B. Chilton, Steven P. Dow, and Robert C. Miller.
HCOMP 2013. Honorable Mention for Best Paper

Cobi: A community-informed conference scheduling tool.
Juho Kim, Haoqi Zhang, Paul André, Lydia B. Chilton, Wendy Mackay, Michel Beaudouin-Lafon, Robert C. Miller, and Steven P. Dow.
Full paper at UIST 2013.

Addressing Users' Queries directly in the Web Search Results
Lydia B. Chilton, Jaime Teevan
Full paper at WWW 2011.

Task Search in a Human Computation Market
Lydia B. Chilton, John J. Horton, Robert C. Miller, Shiri Azenkot.
Full paper at KDD-HCOMP 2010.

Exploring Iterative and Parallel Human Computation Processes
Greg Little, Lydia B. Chilton, Max Goldman, Robert C. Miller.
Full paper at KDD-HCOMP 2010.

TurKit: Human Computation Algorithms on Mechanical Turk
Greg Little, Lydia B. Chilton, Max Goldman, Robert C. Miller.
Full paper at UIST 2010.

The Labor Economics of Paid Crowdsourcing
John J. Horton, Lydia B. Chilton
Full paper at ACM E-Commerce 2010.

Seaweed: A web application for designing economic games.
Lydia B. Chilton, Clayton T. Sims, Max Goldman, Greg Little, and Robert C. Miller.

Tabulator: Exploring and analyzing linked data on the semantic web.
Tim Berners-Lee, Yuhsin Chen, Lydia B. Chilton, Dan Connolly, Ruth Dhanaraj, James Hollenbach, Adam Lerer, and David Sheets.
Semantic Web User Interaction Workshop, 2006.

Why We Customize the Web
Lydia B. Chilton, Robert C. Miller, Greg Little, and Chen-Hsiang Yu.
Chapter In A. Cypher, M. Dontcheva, T. Lau, and J. Nichols, eds., No Code Required: Giving Users Tools to Transform the Web, Elsevier, 2010.

Rewriting the Web with Chickenfoot
Robert C. Miller, Michael Bolin, Lydia B. Chilton, Greg Little, Matthew Webber, and Chen-Hsiang Yu.
Chapter In A. Cypher, M. Dontcheva, T. Lau, and J. Nichols, eds. No Code Required: Giving Users Tools to Transform the Web, Elsevier, 2010.

Community Leadership

CrowdCamp is hack-a-thon for crowdsourcing researchers. I co-organized the first workshop for crowdsourcing at CHI 2009 with over 100 attendants. Paul Andre lead the first follow-up at CHI 2011. I co-organized CrowdCamp at CSCW 2013, HCOMP 2014 and at a BIRS-CMO Workshop in 2016. CrowdCamp is continues under new leadership at HCOMP 2015.

Follow the Crowd
Follow the Crowd is a blog initiated by myself and the other members of the crowdsourcing community. It is a place to summarize crowdsourcing research across academic conferences.

TurKit and the Deneme Blog
TurKit is a toolkit for running iterative tasks on Mechanical Turk. TurKit was developed by Greg Little and myself. Greg and I and others post to Deneme, our blog about experiments exploring MTurk. If you do MTurk experiments, and want to share, we'd love you to post on Deneme.


I started and led a popular web programming competition as an undergraduate at MIT. As a graduate student, I have been a teaching assistant at Stanford, MIT, and the University of Washington.

6.470 MIT Web Programming Competition
  • I established 6.470, a month-long web programming class and competition for MIT students. I was the chairman in 2008 and 2009, and a staff member and instructor for 2010. The class serves approximately 100 students with a $40,000 annual budget from sponsor contributions.
Stanford University
  • Research Topics in Human-Computer Interaction (CS 376) Spring 2015
University of Washington
  • Artificial Intelligence (CSE473, undergraduate version) Spring 2013
  • Machine Learning (CSE546, graduate version) Winter 2012
  • Artificial Intelligence (CSE473, undergraduate version) Autumn 2011
  • User Interface Design and Implementation (6.831) Spring 2009
  • Introduction to Java for Engineers (1.00) Spring 2008 and Fall 2009
  • Communication for EECS Majors (6.UAT) Fall 2008


Before starting at Columbia, I was a post-doc in the Computer Science Department at Stanford University working with Maneesh Agrawala. My PhD is from the University of Washington where I worked with James Landay and Dan Weld. I was an undergraduate and MEng student at MIT working with Rob Miller. I majored in Computer Science and Economics. However, I also completed the requirements for a major in math, but triple majoring was disallowed starting with my graduating class. Clearly, this was a conspiracy against me personally.

I have lived in Beijing three times. My Chinese name is 高雅丽 (Gao1 Ya3Li4)

I recreated famous paintings on the walls of my undergraduate dorm, the infamous East Campus Dormitory of MIT.

William Shatner photobombed me at a Star Trek convention.