Episode Details

A developer spent three years building AI to turn beach volleyball broadcasts into automated stat sheets for pro teams. Then it all fell apart in one instructive collapse. In this series premiere, we explore the volleyball problem—a task so difficult it might be the first real challenge AI couldn't solve. 00:00 - Introducing the volleyball problem series 01:15 - What does a volleyball stat sheet actually need? 03:45 - Why video analytics fails at this specific problem 07:30 - The collapse: how three years of work unraveled 10:45 - What we learned and what's coming next --- Sources & further reading: • game-senser-fable repo — docs/podcast/LEARNING-CONTEXT.md (the source document for • game-senser-fable repo — GAPS.md (the premise autopsies: detection coverage, the • this series, written 2026-07-27) • 75.8% provenance, the circular ground truth) • game-senser-fable repo — JOURNAL.md + REVIEW-BRIEF-2.md (iteration log; the taste check and • the p79 false ace) • game-senser-fable repo — PROBLEM.md + baseline/BASELINE_SCORES.md (problem definition; the • 25.4% zone floor) • game-senser-services repo — docs/CV_PIPELINE.md (the 2025 experiment history, Exp 1–18) This podcast episode was fully generated by AI — research, script, voices, and production. Built with Claude, Piper TTS, and automated pipeline tooling.