“How We Can Eliminate Crime” | Ben Horowitz and Garrett Langley
“How We Can Eliminate Crime” | Ben Horowitz and Garrett Langley
Summary
- The episode’s central call is to replace punishment severity with apprehension certainty. Horowitz argues that reliable detection changes incentives before crimes occur, whereas sporadic enforcement followed by long sentences merely “throws people away.” His governing maxim: “Certain punishment means no punishment.”
- Langley’s national playbook combines people, products, and policy: rebuild police staffing, integrate sensors through AI, and consistently prosecute declared crimes. Horowitz proposes retiring student debt for people who serve two to four years in patrol, analysis, or civilian roles; Langley calls it an important staffing idea and proposes a national law-enforcement act. The investment thesis is that technology cannot compensate for collapsing institutional capacity, but it can make restored capacity dramatically more productive.
- Murder clearance is the starkest measure of the system’s deterioration. Langley puts the national rate outside Las Vegas at roughly 47%—a “53% chance of getting away with murder”—versus well over 90% in Vegas. He attributes the gap to lost witness cooperation, more randomized gang and drug-related killings, evidence volumes outrunning investigative tools, and veteran detectives retiring faster than replacements mature.
- Las Vegas is presented as proof that a relatively small technology layer can transform a headcount-heavy police budget. Horowitz says cameras and drones coincided with an approximately 75% reduction in police shootings of suspects, while private funding equivalent to less than 1% of the department’s budget enabled capabilities government procurement struggled to initiate. The operating promise is precise identification: “You can commit a crime in Vegas; you can’t get away with it.”
- Flock’s defensibility rests on turning fragmented public-space sensors into an intelligence and orchestration layer, not merely selling license-plate cameras. Langley describes integrating LPRs, gunshot detection, drones, video, criminal-history context, and a potential agentic software layer that coordinates responses. Langley says Flock does about 1 million arrests this year and that it helped return more than 450 missing children—high-stakes proof points for network coverage and cross-agency interoperability.
- The strongest criticism is institutional trust, not privacy in isolation. Langley argues that a federal cell-phone location dump is more revealing than a public-road license-plate record, but concedes that communities distrusting police will resist any additional capability. His answer is configurable retention and sharing—30 days, five years under New Jersey law, a 90-day California maximum, and as little as seven days at some agencies—plus public transparency about use.
- Enforcement and criminal-justice reform are complementary rather than opposing investments. Horowitz separates protecting victims from repairing prisons whose U.S. recidivism rate he puts above 70%, versus below 40% in better-performing systems; Langley favors work-school-home restrictions over jail for many young, nonviolent offenders. Their shared condition is that legitimate career paths must expand while the criminal path becomes reliably unattractive: “Those go together.”
Deep dive
Not yet available upstream; scheduled sync will retry.