Alex Berke, D Calacci, Robert Mahari, Takahiro Yabe, Kent Larson, Sandy Pentland

The result of our crowdsourcing experiment, this dataset details the purchase histories of over 5,000 users over 5 years. We show a few ways the data can be used and validate its' representativeness.

Varun Rao, Eesha Agarwal, Samantha Dalal, D Calacci, Andrés Monroy-Hernandez

This work experiments with using LLMs to analyze and extract specific insights from large unstructured text corpora (such as all the posts on the rideshare subreddits).

Alex Berke, Robert Mahari, Alex (Sandy) Pentland, Kent Larson

We investigate how different incentive structures impact people's willingness to contribute their Amazon purchase history data to an academic research project. One big takeaway: data transparency matters for crowdsourced data participants.

D Calacci

I was invited to speak at this invite-only event that convened investigative journalists, non-profit researchers, academics, and activists to discuss the promises and challenges of doing data-driven investigatory work in tech accountability.

D Calacci & Jake Stein

How does current data protection law work for workers? In this invited article for a special issue of the European Labor Law Journal, my colleague Jake Stein and I argue that worker co-determination should be seriously considered as a way to regulate AI and data use in the workplace.

D Calacci

Contribution to ACM Interactions' Tech + Labor Forum on automation and worker agency. I argue that workers—and researcher+advocate allies—can & should collectively help direct automation at work.

D Calacci, Deb Raji, Abeba Birhane, Brandi Geurkink, Becca Ricks, Claire Pershan, Mehan Jayasuriya, Victor Ojewale, Marc Faddoul

I participated in an invited panel discussion on the state of algorithmic auditing tools! This was a great discussion that touched on the new Digital Services Act, community control of data and auditing technologies, and the taxonomy of audit tools.

D Calacci, Jeffrey Shen, Alex (Sandy) Pentland

We use spatial regression models, structured topic models, and an experimental survey to understand how users of Amazon's Ring Neighbors network racialize and criminalize their subjects, and document what kinds of communities nationwide use the network most.

Alex Berke, D Calacci

In 2020, Google introduced FLoC, a way to facilitate interest-based individual advertising without 3rd-party cookies. This paper shows that the FLoC proposal included serious privacy risks and explores FLoC's risk of leaking sensitive demographic information about it's users.

D Calacci, Alex (Sandy) Pentland

This paper introduces the Shipt Calculator, a tool used in a 2020 worker-led campaign that revealed that Shipt's new black-box payment algorithm cut the pay of over 40% of studied workers.

I was an invited panelist at FTC's PrivacyCon in November 2022, speaking about commercial and worker surveillance. In my talk, I argued that data regulation for workers should be more about worker power and agency than privacy rights.

Lam Thuy Vo

Our work on Amazon Ring was featured heavily in an investigative piece published by the Guardian and Type Investigations examining the rise of new laws in the US turning package theft into a felony.

Mozilla

I was featured in the 2022 Mozilla Internet Health Report. This year, it took the form of a podcast where I was featured in an episode about workers, organizers, and researchers thinking about how technology and data impact modern work.

D Calacci

In July 2022, I gave an invited talk at the Hackers on Planet Earth conference outlining some existing projects related to using data for worker organizing, and discussing more generally how a hacker ethos can fit within the modern labor movement.

WNYC's Radiolab

I appeared on Radiolab's August 26, 2022 episode, sharing my experience working with worker-organizers to audit Shipt's black-box pay algorithm and to discuss the condition of gig workers more generally.

Willy Solis, Vanessa Bain, D Calacci, Drew Ambrogi, Danny Spitzberg

A real-world community audit of a black-box algorithmic system, the Shipt Calculator impacted workers, organizers and researchers and demonstrates how community-led research can be part of the FAccT community.

D Calacci

Position paper in CHIWORK '22 arguing that a new "Digital Workerism" in the CHI and CSCW communities is needed to bolster the labor movement and balance information asymmetries.

Gizmodo

Google's FLoC was a proposal that would change the way the web fundamentally worked for millions of people. Why was studying it so inaccessible? In this Op-Ed, I argue that centralized gatekeeping of future web technologies is dangerous for the future of the web. I call for Google and other major companies to publish toolkits that let researchers study new technologies that will fundamentally change the web.

Esteban Moro, D Calacci, Xiaowen Dong, Alex (Sandy) Pentland

Is your local coffee shop more segregated by income than your favorite movie theater? We use a massive data set of mobile phone mobility to answer this question and model how individual segregation is related to people's tendency to explore new places and interact with those different than themselves.

D Calacci

Invited talk at Data & Society on mapping and analyzing Amazon Ring's network.

D Calacci

Remote presentation at HOPE (Hackers On Planet Earth) 2020 on hacking Amazon Ring's Neighbors app to reveal and measure the extent of neighborhood surveillance captured in the Ring Doorbell camera network.

Dhaval Adjodah, D Calacci, Abhimanyu Dubey, Anirudh Goyal, P.M. Krafft, Esteban Moro, Alex Pentland

Can network structures inspired by human social networks improve distributed reinforcement learning algorithms? This paper proves that arranging agents in different network topologies can massively improve evolutionary deep reinforcement learning algorithms.

Gizmodo

At the beginning of COVID, many states and universities were experimenting with using location data to track covid spread. In this op-ed I argued that location data is a dangerous technology to break out for state-level disease surveillance and is a poor technical choice for tracking airborne illness.

D Calacci, Alex Berke, Kent Larson, Alex (Sandy) Pentland

Location data collected from mobile phones and aggregated in massive databases poses enormous risks to individual and collective privacy. It also poses clear utility for research, marketing, and policymaking. This paper explores and conceptually models the risks that large-scale location datasets introduce, and speculates on ways that location data can be regulated or protected while offering significant utility.