AI Now Spots Bloated Websites — And Why It Matters

Website redundancy doesn’t have one fixed meaning—the same repeated element can distract in one task and serve as backup in another. A new auditing procedure called CORA separates three dimensions: repetition load, normal-use tax, and failure-domain recovery reserve. Each run keeps screenshots, stable element identities, and task traces, while a versioned vision-language model proposes the annotations. A typed validator and release gates decide whether a dimension can be reported; failed or malformed outputs stay in the fixed denominator. On a controlled mechanistic testbed, CORA’s factorized representation separated recovery reserve from normal-use tax and predicted perturbed success better than scalar-load baselines. Two small local vision-language models produced recurring outputs, but neither met all release requirements, so CORA withheld automated scores from both while retaining raw responses and failure records. The authors position CORA as an auditable candidate procedure for the controlled benchmark studied—not a general standard. Human agreement, AI-versus-human accuracy, and validation on independent production sites remain open empirical questions.