Blitz Debate
The Project
Blitz Debate is a Retrieval Augmented Generation (RAG) system that evaluates the performance of Lincoln-Douglas debates. The project was inspired by the consistent struggle for small-school debaters who cannot either cannot hire judging or don’t receive constant unbiased judging. The process of development was tough but rewarding. At first, just figuring out how to structure the retrieval part—so the model could pull relevant evidence, framework analysis, or even judge paradigms—took a lot of trial and error. Then came the generation part, where I had to make sure the model could not only summarize arguments but actually evaluate things like clash, weighing, and impact calculus. It wasn’t just about coding—it was about teaching the system how real debates are judged. It took time, tons of testing, and a solid understanding of both debate theory and NLP, but seeing it work made all the effort worth it.
