This post is the second half of a two-part argument. The first half examined llms.txt advertising and found it fails on delivery — the file almost never gets fetched, so the impression never happens. This half examines an advertising product where delivery is solved. The failure is one layer up. Disclosure: this site builds and operates agent-detection infrastructure. The verification gap described in the close is one this operation would benefit from commercially. The analysis is published because the argument follows from the evidence, not from the incentive.

It is a common confusion in our industry: “GEO” often refers to “Generative Engine Optimization.” But for the scientific community, GEO means Geology. And interestingly, geological data provides one of the best case studies for how to ground Large Language Models in physical reality. The Hallucination of Physical Space Ask an ungrounded LLM “What is the soil composition of the specific plot at [Lat, Long]?” and it will likely hallucinate a generic answer based on the region. “It’s probably clay.” It averages the data.

An analysis of how Large Language Models ingest and utilize structured data during pre-training, moving beyond ’text-only’ ingestion to understanding the semantic backbone of the intelligent web.