Consequential Actions Podcast

← Consequential Actions Podcast10 sept. · 1 h 08 min

David Evans of Sentiero Ventures on Artificial Intelligence, Lyapunov Time, and the Knowledge Problem | Self-Evident Week 12

David Evans of Sentiero Ventures on Artificial Intelligence, Lyapunov Time, and the Knowledge Problem | Self-Evident Week 1210 sept.1 h 08 min

<p>Self-Evident: The Road to 1776 — Episode 12i</p><p><br></p><p>I Asked an AI Investor Whether a Machine Could Set Every Rent in New York. He Said It Absolutely Could.</p><p><br></p><p>David Evans is the founder and managing partner of Sentiero Ventures, an early-stage venture fund outside Dallas that invests in business software where artificial intelligence is the core of the product. He has been writing software since the early 1990s, built and sold a national digital signage network, and founded EasySeat, an online ticket resale business that reached the Inc. 500 and sold in 2015, and teaches graduate courses on innovation and entrepreneurship at the University of Texas at Dallas. He and Jeff met as seventeen-year-olds sharing a dorm room in Baltimore in August 1995.</p><p><br></p><p>The question was direct: could a system read every rent-stabilized building file in New York and produce a number the board could not produce? He said it absolutely could — and then he scoped it, and the scope turned out to be the whole answer. It could read every unit, tenant, price and alternative in the city at once, at a fixed point. It cannot carry that forward, because every decision compounds against every other one until the thing diverges past recovery. He has a name for that horizon, borrowed from mathematics rather than economics, and he ran into it twenty years ago building ticket prices, where he spent years assuming it was a data problem before concluding there was no data to find, because the events had not happened yet. We also cover why large language models are not on the road to general intelligence, what a dog and a bell reveal about the difference between having the words and having the world, data poisoning as an attack on the record itself, the arithmetic of federal equity stakes, and the moment an aggregator becomes a market maker and the number on the screen stops being a price.</p><p><br></p><p>Recorded 6 September 2026; the Rent Guidelines Board litigation (argued 2 September, decision reserved) and the Senate FY2027 defense authorization may have moved since.</p><p><br></p><p>Did this leave us more free, or more governed?</p><p><br></p><p>⏱️ CHAPTERS</p><p>0:00:00 — Introduction to the Episode</p><p>0:01:15 — Guest Introduction</p><p>0:05:11 — Interview Commences</p><p>0:05:47 — Machine Learning in Ticketing</p><p>0:09:09 — The Business Value of AI</p><p>0:13:10 — AI and the Job Market</p><p>0:17:27 — The Scottish Enlightenment and LLMs</p><p>0