Sixteen Adversarial Reviewers
Alexa and Taylor tally the real cost of running agents, then follow that trust boundary into local inference, agentic trading, AI text marks, persistent harnesses, robot training data, open-source supply-chain risk, and the question every loop has to answer: who gets to say stop?
screen segments
The HTML screens used in this episode.
- Nine Hours
An interactive editorial explainer on Gal Ratner's “The End of Open Source” — one compromised maintainer account, nine hours, two billion monthly installs, and the door nobody was standing at.
full view →details → - Headlong
An interactive editorial explainer on Laude Institute's Headlong, the Bash microharness for agents that never stop thinking — and what it costs to keep the meter running while nobody's talking.
full view →details → - Knowing When to Stop
An interactive editorial explainer on a16z's loop-convergence essay — the $1.40 that bought 63 Lighthouse points, the $2.84 that bought zero, and why ‘done’ is a judgment the system has to make.
full view →details → - The Hidden Mark in Your Draft
A sourced visual story about Claude's statistical text mark, authorship, provenance, detection limits, and likeness rights.
full view →details → - The $534 Account
A sourced visual story about a public agentic-trading experiment, the history of automated markets, evaluation, and enforceable controls.
full view →details → - Can Robots Learn Physical Work by Watching People?
A sourced visual story about Dyna-2, human-video pretraining, robot adaptation, embodiment, worker data, and the evidence still needed.
full view →details →