Wednesday, 14 October 2026
39 episodes · 6596 in all
#5980: Do Spy Agencies Have Distinctive Tradecraft Styles?
Dead drops, illegals, and the Cambridge Five — do spy agencies share one discipline, or does each have a fingerprint?
#5979: Why We Shake Hands: A 2,800-Year History
The handshake wasn't inevitable — it won. From Assyrian treaty reliefs to Quaker radicalism, how one gesture conquered the world.
#5978: God Bless, Inshallah, Be'ezrat Hashem: Three Ways to Say Maybe
Same God, same syntax, three completely different social lives — how "God bless," "Inshallah," and "be'ezrat Hashem" do totally different jobs.
#5977: Spy Codenames, Dead Drops, and the Phrases That Fooled Everyone
Real recognition phrases from the 2010 Russian illegals case — and the twist that some were written by the FBI.
#5976: What North of North Assumes You Know About Inuit
A crash course in Inuit history, language, and modern life — the context the show assumes you already have.
#5975: Merging vs Distillation: Can Two Models Become One?
Merging and distillation get lumped together, but they're opposites. One composes existing capability; the other has to teach it.
#5974: One Model or Two? ASR Cleanup Explained
If speech models can punctuate and capitalize on their own, why does a separate 484MB cleanup model exist at all?
#5973: MCP Security: Tool Poisoning and Agent Identity
MCP servers trust their own tool descriptions. That design choice creates a whole new class of attack — and a new observability problem.
#5972: AI Meeting Agents Are Becoming Stand-Ins
Bots now book follow-ups, update CRMs, and attend on your behalf. Here's what's real, what's speculative, and where the etiquette is heading.
#5971: Personal CRM in 2026: Dex, Mesh, folk, Monica, Twenty
What actually makes a good personal CRM — and which ones can an AI assistant talk to?
#5970: Swapping Your Linux Kernel: What Actually Changes
Your kernel is swappable without reinstalling anything. But do Liquorix and XanMod actually beat the default? The benchmarks say less than the mark...
#5969: Ubuntu 26.04 LTS: Is Staying on LTS Still a Trade-Off?
Ubuntu 26.04 LTS shipped in April. Point releases, HWE kernels, and SRUs have quietly closed most of the freshness gap — but not everywhere.
#5968: Gentoo in 2026: Still Worth Compiling?
Twenty years on Linux and he's never tried Gentoo. In 2026, does the compile-everything reputation still hold?
#5967: How UEFI, Secure Boot, and the Boot Chain Actually Work
Twenty years of Linux and nobody explains the firmware. Here's the boot chain, the keys, and what's changing in 2026.
#5966: How Much of Your Linux Desktop Is Actually Misconfigured?
Most "Linux tweaks" are folklore. A few are real — and one toggle most people never open is worth up to 42%.
#5965: Compiling From Source: What You Actually Get
Source builds aren't faster — they're smaller. What USE flags, compiler flags, and the build feedback loop actually buy you.
#5964: Containers vs VMs for GPU Inference: What You Actually Pay
Containers cost 0.03%. Passthrough VMs cost 1%... or 36%. Same hardware, different systems — here's how to tell which you'll get.
#5963: Running One Model Across Mismatched GPUs
Tensor, pipeline, or offload? What actually happens when you run one model across several GPUs — including mismatched ones.
#5962: How Word Error Rate Lies to Us
WER is the number everyone reports for speech recognition — and the one that agrees with human judgment least. Here's what it actually measures.
#5961: What Your GPU Is Actually Waiting On
Your GPU says 100% utilized. It's multiplying zeros. Here's what the counters actually measure during inference.
#5960: There's No Such Thing as Tokens Per Second
Why one tokens-per-second number can't describe an LLM server — and which metrics actually tell you if it's working.
#5959: What RTF Actually Tells You About Speech Models
Accuracy gets the headlines, but speed decides what ships. How RTF, latency percentiles, and memory shape real ASR deployments.
#5958: Freedom of Information: Law vs. Reality
140 countries have freedom of information laws. A 2025 test found 38% of requests got no substantive reply at all.
#5957: The Scientific Plugins Hiding in Plain Sight
ChatGPT's plugin directory has 5,723 entries — but the real scientific AI work is happening in arXiv papers, not plugins.
#5956: Will ChatGPT Eat Your Grocery App?
MCP servers, ChatGPT apps, and a prediction that AI vendors will absorb your grocery list one micro-app at a time.
#5955: Work Lights: Lumens, Lux, and the Specs That Matter
Lumens is the easiest spec to print and the least useful on its own. Here's what to read instead when buying a work light.
#5954: Layered Safeguards and the Math Behind Them
Three safeguards at 90% each seems bulletproof—until shared failure modes collapse the math.
#5953: How Militaries Decide What to Buy
Twelve Apaches, a billion shekels, and a question nobody answered: where do weapons requirements actually come from?
#5952: Inside Malmab: Israel's Most Secretive Security Body
Israel's classified defense security body and what its extreme information practices teach about protecting secrets.
#5951: The Reasonable Person Is a Legal Fiction
Courts say they judge you by a neutral "reasonable person." Who decides what that person would do — and whose average is it?
#5950: Self-Hosting Your Search Stack: SearXNG, Crawl4AI, and the Real Bill
Can SearXNG, Crawl4AI, and an open-source aggregator really replace a commercial search API? Yes — with a very large asterisk.
#5949: Inside Deep Research Agents: Claims, Citations, and the 84.7% Orchestrator Problem
Why 84.7% of citation errors originate at the orchestrator — and how a claim ledger fixes it.
#5948: LangGraph, LangChain, and Deep Agents in 2026
Runtime, framework, harness — how three LangChain layers actually fit together, and what changed in 2026.
#5947: Who Actually Trades on Polymarket?
Three percent of wallets drive 90% of Polymarket's volume — and most retail accounts quietly lose money. Here's who's really on the other side.
#5946: Inside Humanity's Last Exam: 2,500 Questions AI Can't Ace
2,500 questions, ~1,000 experts, and a top score that jumped from 2.7% to 54.8% in 18 months. What is HLE actually measuring?
#5945: How a Reward Model Actually Trains an LLM
One number, no gradients, four models in the room. What a reward model actually does inside the RLHF loop.
#5944: Keeping the Rollout Engine Fed: Inside RLHF Pipelines
Most of an RLHF run is spent waiting on generation, not gradients. Here's how TRL, OpenRLHF, and verl actually differ.
#5943: Tourniquets From Belts and Socks: What Actually Works
No commercial tourniquet on the ridge — so students built them from belts and socks. What the evidence actually says about improvised tourniquets.
#5942: Service Workers: The Proxy That Never Sleeps
A JavaScript file with no tab, no DOM, and the power to intercept every request your site makes — and why refreshing never updates it.