DOGFOOD – Autonomous Hackathon Evaluation & Scoring Platform

Описание и применение

DOGFOOD – автономная платформа для оценки хакатонов, устраняющая предвзятость судей, неравномерность оценок и проблемы с подключением.

Задача

Hackathon judging suffers from anchoring bias, judge variance, uneven review distribution, and offline failures.

Результат

Judging bias reduced, review distribution balanced, and offline resilience achieved.

Рабочий процесс

• Deploy Docker appliance via docker compose up --build
• Judges log in to /dashboard/judge and receive blind queue
• Submit scores through /api/v1/judge/scores
• Organizer monitors progress at /dashboard/organizer
• Scores are calibrated with Empirical Bayes and displayed at /results
• Export CSV via /api/export.csv

Доказательства

DOGFOOD was built to solve these systemic problems once and for all. It is an autonomous, zero‑trust, single‑port offline appliance running on port 8080

Источник

Empirical Bayes shrinkage formula: μ̂_i = (n_i · x̄_i + k · μ_0) / (n_i + k)

Источник