Privacy as Contextual Integrity, Explained | Helen Nissenbaum (2004)

Why do some uses of information feel like a violation, even when it was never secret? In this 2004 paper, philosopher Helen Nissenbaum argues that privacy isn’t about secrecy. It’s about whether information flows appropriately for its context. This video walks through the argument: the three cases that broke the old public/private model, the two kinds of norms behind contextual integrity, its roots in Michael Walzer’s Spheres of Justice, and how the framework judges new practices. 📄 Paper: Helen Nissenbaum, “Privacy as Contextual Integrity,” 79 Washington Law Review 119 (2004). Open access: https://digitalcommons.law.uw.edu/wlr/vol7... Chapters 0:00 Introduction 0:22 The puzzle: “public surveillance” 1:14 Why the old framework says “no problem” 1:58 “No expectation of privacy in public” 2:22 The flaw: one dividing line 2:46 The central tenet 3:04 Life is lived across many contexts 3:33 Two kinds of norms: appropriateness and flow 4:24 Contextual integrity, pictured 4:59 The root: Walzer’s Spheres of Justice 5:49 “Isn’t this just the status quo?” 6:45 The three cases, revisited 7:47 Questions for judging any information flow 8:24 Takeaway: privacy is appropriate flow How this video was made This explainer was made with AI tools, and the narration is a synthetic voice. 🤖 Research, script and production: Claude Opus 5.5 (Anthropic), running in the Claude Code desktop app. Claude: pulled the paper from my Zotero library through a Zotero MCP connector extracted the full text with pdftotext and read all 41 pages wrote the narration as a paraphrase of the paper’s argument and checked quotes and page cites against the PDF wrote the Python code that draws the slides and assembles the video 🎙️ Narration: ElevenLabs text-to-speech, model eleven_v4, voice “George”. Each of the 36 lines is generated separately but sent along with the lines before and after it (“request stitching”), so the delivery flows across cuts. Lines are cached, so editing the script only re-generates what changed. 🎨 Slides: drawn frame by frame in Python with Pillow, set in Avenir Next. 🎬 Assembly: ffmpeg combines each slide with its audio clip, adds fades between sections, and joins everything into a 1080p MP4 with built-in English captions. The whole thing is about 300 lines of Python driven by a single script file, so it can be re-rendered or re-voiced in minutes. ⚠️ This is a summary, not a substitute for the paper. Any errors of interpretation are mine. Please read the original. #privacy #contextualintegrity #philosophy #dataethics #surveillance #Nissenbaum #AIgenerated