About
Purpose
The Brute corpus is a research-grade index of one host's long-form recorded talks. It exists to make the body of work navigable: every episode rendered as a speaker-labeled, lightly-cleaned transcript with anchored timestamps; every recurring theme distilled into citable positions; every distinctive turn of phrase catalogued. It is the substrate for downstream tools that answer questions about the host, map the structure of his thinking over time, and study the rhetoric on its own terms.
Nothing on this site is editorialized. The host's own words are preserved verbatim in the transcripts. The persona files are machine-distilled summaries grounded in direct quotation, with each claim citing the specific show and timestamp it came from. Where the machine summary is wrong, the source quote is one click away to check.
Methodology
The pipeline is straightforward. Audio is downloaded from public sources, transcribed with an automatic speech-recognition service configured for speaker diarization, and the longest-talking speaker in each session is labeled HOST. The remaining speakers become CALLER 1, CALLER 2, and so on, ranked by share of total speaking time. Filler tokens are stripped conservatively from the host's text; no filler removal alters substantive vocabulary.
The persona facets (biography, dossier, philosophy, voice, vocabulary, topics) are produced by passing the cleaned transcript corpus through a long-context language model with a structured prompt for each facet. Output is appended to the corresponding persona file under a dated block; previous distillations are preserved so changes in understanding remain visible over time.
What it is not
This is not an official site, not affiliated with the host, and not commercially associated with any platform. The transcripts are derived from publicly available recordings; original copyright remains with the speaker. If you are the speaker and would like content removed or amended, the contact form below is the right place to start.