Exposing Algorithms
summary
Shows concrete investigations: a Google autocomplete probe that maps words blocked by Google and Bing, and an Uber study that samples DC on a dense grid every 3 minutes for four weeks using the Uber API, validates 30,234 valid addresses, merges American Community Survey tracts, and runs regressions in pandas and statsmodels with visualizations in Seaborn. Results highlight longer ETAs and less surge time in higher–percent‑people‑of‑color tracts and point to reproducible code on GitHub for audit.
AI-written summary · about
chapters (16)
- 0:00Computational journalism intro
- 1:05Algorithms in daily life
- 2:29Facebook post ranking experiment
- 3:11Google autocomplete censorship
- 6:40Uber service quality study
- 8:05Measuring Uber wait times
- 10:10DC neighborhood sampling
- 12:14Geographic data collection
- 14:20Address validation methods
- 17:11Wait time analysis in DC
- 19:23Regression on service disparities
- 21:38Uber’s response analysis
- 23:27Research next steps
- 24:53Algorithmic accountability transparency
- 25:58Open data and collaboration
- 26:19Q&A
related talks
34mData Drives Business
24mData science, apps and civic responsibility
01h 03mUnderstanding and Shaping Urban Mobility Using Dollars and Data II
01h 15mSunday Keynotes
15mA Closer Look: Evaluating Location Privacy Empirically
22mExpert Panel | Location Data in Retail & Real Estate: Delivering an ROI Through Analytics
04mYour data is being manipulated.
02h 53mMining digital exhaust