Mar 20, 2014 · 22m · mad
Chris Volinsky, AT&T // Data Driven #25 // March 2014 (Hosted by FirstMark Capital)
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this Data Driven NYC keynote, Chris Volinsky from AT&T Labs-Research presents how anonymized telecom mobility data can be ethically leveraged for smart city initiatives, urban planning, and commuter traffic analysis, highlighting a detailed case study in Morristown, New Jersey.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 3.1% of the talking time here. How this is scored →
Matt as informed peer 0.0 Guest teaching 3.6 Guest disagreement 0.0 Matt pushing back 0.0
speaking balance: gold is Matt, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 20:21 Clarifying that AT&T does not sell user data
When asked by an audience member which businesses AT&T sells cell data to, Chris directly rejects the premise by clarifying that AT&T does not sell the data at all.
Hardest push from Matt ▶ 21:23 Matt Turck cutting off Q&A to wrap up the eventHost Matt Turck decisively ends the Q&A session and wraps up the evening, maintaining tight control over the event timing.
Biggest teaching moment ▶ 16:20 Explaining Earth Mover's Distance for noisy dataChris educates the audience on applying the Earth Mover's Distance metric to resolve spatial noise in cellular handoff sequences.
Matt holds his own ▶ 0:00 Matt Turck highlighting Chris's Netflix Prize credentialHost Matt Turck showcases technical awareness by introducing Chris specifically as a former Netflix Prize winner.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Matt as informed peer | Guest teaching | Guest disagreement | Matt pushing back | Why |
|---|---|---|---|---|---|---|
| AT&T's Tribeca Headquarters and Recruiting Pitch | 0 | 2 | 0 | 0 | Chris gives a solo presentation introducing AT&T Research Labs and pitching their windowless Tribeca headquarters to potential data science hires. Because this is a monologue presentation without host participation, host-side scores remain zero. | |
| Examples of Open Data Initiatives in Chicago, Boston, and NYC | 0 | 3 | 0 | 0 | Chris reviews smart city open data projects across Chicago, Boston, and NYC, highlighting real-time snowplow tracking and pothole reporting. The host does not speak during this presentation segment. | |
| How Cell Tower Location Data Works | 0 | 3 | 0 | 0 | Chris details how AT&T collects cell tower transaction logs instead of individual GPS tracking to preserve user privacy. The segment is a monologue presentation. | |
| High School and Downtown Nightlife Behavioral Patterns | 0 | 4 | 0 | 0 | Chris demonstrates how temporal spikes in voice calls and text messages reveal high school dismissal times and weekend bar closing habits in Morristown. The monologue continues without host commentary. | |
| Traffic Challenges and Origin-Destination Commuter Mapping | 0 | 4 | 0 | 0 | Chris presents origin-destination commuter mapping derived from aggregate cellular traffic to help urban planners solve congestion. As a solo presentation monologue, host-side engagement is zero. | |
| Route Handoff Modeling and Earth Mover's Distance | 0 | 5 | 0 | 0 | Chris delivers a technical explanation of route handoff modeling and introducing the Earth Mover's Distance algorithm to handle cell tower noise. The segment is an uninterrupted technical presentation. | |
| Synthetic Data Generation and Presentation Conclusion | 0 | 4 | 0 | 0 | Chris concludes his presentation by describing synthetic data generation techniques that enable open crowdsourcing while protecting customer privacy. Host-side scores remain zero for this presentation section. |
Statements from this episode (14)
Volinsky: AT&T Labs recently moved into a 30-story windowless building in Tribeca
“We just kind of recently moved into a floor in an old AT&T building here down in Tribeca, and this, you may have seen this if you're down there, it's down on Thomas Street, a thirty-story building with no windows.”
Volinsky: Cell phone data reflects identity, browsing habits, and physical location
“Your cell phone Is a real reflection of who you are, where you surf, where you are.”
Volinsky: Chicago was an early pioneer in open crime data
“Chicago is one of the first players. They have online a whole bunch of crime statistics. You can go and download the data. They have APIs to it.”
Volinsky: Boston uses Street Bump to improve pothole repair efficiency
“Boston has this thing called Street Bump where you can report potholes so that when they send out pothole crews, they can be more efficient in, in filling those potholes.”
Volinsky: AT&T does not monitor or retain call or text content
“We never have any idea of what you're talking about or who you're texting, or what you're texting, the content of your text. All we have is a transactional record that this phone was at this location and made this particular transaction, but we don't know any …”
Volinsky: AT&T always processes mobile network data anonymously
“We always look at the data in anonymous form. We aggregate whenever possible when the application allows it.”
Volinsky: Cell tower location data pegs users roughly within a square kilometer
“And so that might peg you down to roughly within a square kilometer less if you're in New York, more if you're in a suburban or rural area.”
Volinsky: Effective data visualization often requires large physical paper printouts
“Sometimes to do really good data visualization, You need to kill some trees, unfortunately. There's not, there's very little substitute for, like, actually printing something out huge and putting it on the wall.”
Volinsky: AT&T cell data shows Morristown High School activity spikes
“This was pointed at Morristown High School. So the three spikes on the bottom are texting spikes at the beginning of the day, before school started, during the lunch hour, and when they get out of school. And then you see when they get out of school at two 30,…”
AT&T Labs Drove 12 Morristown Routes 10 Times for Cell Training Data
“We went out and we drove each of these 12 routes 10 times each with two phones in the car, making calls, and GPS running on each one, and to get our training set to understand if you're on this route, what does the cell tower handoff sequence look like?”
Volinsky: Earth Mover's Distance Is a Brilliant Metric for Data Science
“If you've, if you're working in a, if you're doing data science and you need a good distance metric, this, look at the earth mover distance. I had never heard of it before, and it's a brilliant, brilliant piece of science that's, of course, decades old.”
Volinsky: AT&T is developing synthetic datasets to share mobility data safely
“So a project we're working on this year is to create what we call synthetic data, it's just a fancy word for simulated data. What we want to do is create a simulated data set That looks very much like the data that we have, but isn't.”