Typo Positive: When Collection Doesn't Get You What You Need
summary
Claims that collection choices, not topic labels, break data: sensors show condensation and calibration cause pressure drift, GPS fails in urban canyons so mount antennas on the roof or add a navigator backup log, printed circuit boards need proper angles and electrical design, fuel and maintenance corrupt sensor readings, web forms produce off by one month errors, surveys see an 8 minutes engagement dropoff, many databases have 75‑column tables with 50 timestamp fields — fix by tagging columns UTC or local and prefer timestamp with time zone, kiosks and offline‑first sync actually work.
AI-written summary · about
chapters (14)
- 0:00Start
- 0:41What Is Data
- 3:06Validity
- 4:33Two Kinds of Data
- 5:48What Can Go Wrong
- 10:20Gps
- 12:03Gps with Cold Starts
- 16:42Contacting Respondents
- 18:03Program Data
- 20:50Social Acceptability Bias
- 21:31Forms
- 22:09Data Entry Delay
- 26:51Access to Technology
- 37:05Things That Programmers Get Wrong about Names
related talks
35mBad data - anecdotes and examples from the real world
44mKeep it Clean: Why Bad Data Ruins Projects and How to Fix it
36mEventually, time will kill your data processing
05mWhy Your Data Sucks
46mExacting Standards: Considerations for Precision and Accuracy ...
19mLet me show you why data types still matter
28mWhat Data Analysts Wish Application Developers Knew- 19mLet me show you why data types still matter