Cod(e)ifying The Law
summary
An experimental study recruits mixed teams to build web-based legal advice tools from bankruptcy statutes, using JavaScript and HTML and starter front-end code; researchers record 100 hours recorded of sessions and code history. Fifteen final tools mostly implement an incorrect all-or-nothing logic: three of fifteen tools treat each debt type separately as the law requires. Ambiguity (Brunner vs totality tests) is often pushed to the user or resolved by whatever cases LexisNexis surfaces, and 15 of 19 programmers say their tool could decide court outcomes.
AI-written summary · about
chapters (15)
- 0:00Intro
- 0:46Online Legal Advice Tools
- 1:18Online Legal Advice Tool Design
- 1:43Errors in Legal Advice Software
- 2:38Research Question
- 4:02Study Approach
- 4:49Study Design
- 5:42Study Tasks
- 6:55Starter Code Design We provide starter code for both tasks informed by the design of existing legal advice tools
- 8:59Errors in Tool Logic
- 9:40Potential Sources of Error in Tool Logic
- 11:17Judicial interpretation of ambiguous language
- 12:58Interpretation of Ambiguity in CS+Law teams
- 14:35Encoding Indeterminate Legal Provisions
- 17:31Thank you!
related talks
26mJavaScript Tools at Scale Using Type Information
20mCode Mode: Let the Code do the Talking
45mConquering complexity: refactoring JavaScript projects
40mJavaScript ... for Science!
40mRefactoring large JavaScript codebases
17mState isn't all you need, but It helps: building better LLM apps in the browser
53m!important
32mBreak all the rules