AI Models degrade after ingesting from “tarpit“ sites. These are sites that are like honeypots for AI knowledge/data scrapers. They are designed to be attractive to scrapers but they are filled with Markov chains of nonsense that are nearly impossible to detect and untangle once they have been ingested. It has caused billions of dollars of lost revenue for model makers who have to re-train models from previous checkpoints, or scrap models entirely. Amalgamated Bossenbrooks provides a service to both help you setup a tarpit, and to help model makers avoid the tarpit sites both from algorithmic assessments and probabilities, as well as because they know the sites that are tarpit sites. Controversial, but they are a service that is in demand.