My Weird Prompts

Most people know Mossad, Shin Bet, and Aman — but there's a fourth body inside Israel's Defense Ministry that almost no one can name. Malmab, the Director of Security of the Defense Establishment, protects the country's most sensitive nuclear and biological facilities, and its practices reveal how governments actually organize information security when a breach could be catastrophic. From clearanc...

Reasonableness is the load-bearing wall of the legal system — negligence, self-defence, harassment law, and constitutional equality all run on it. But the "reasonable person" is a fiction, and what counts as reasonable is shaped by class, culture, gender, and professional habit. Drawing on Mayo Moran's work on the reasonable person and Victoria Nourse's argument that the standard was never a perso...

Daniel wants to know if you can bolt together a search-and-grounding stack from open-source parts and never rent a search API again. On the table: SearXNG for discovery, Crawl4AI for extraction, and a TypeScript aggregation library for grounding — plus how you'd actually run the whole thing on Modal. The honest answer is yes, with one large asterisk: the bill you save is not the bill you pay. You...

What actually separates a research agent that grounds its claims from one that piles up web searches and writes a confident summary? This episode breaks down the four-stage deep research pipeline — planning, retrieval, verification, synthesis — and the three failure classes that survive every surface check: unsupported claims, silent coverage gaps, and unbounded drift. The fix isn't more search to...

LangChain spent 2026 hardening the parts of the agent stack nobody demos. LangGraph 1.2 cut checkpoint storage by 41x with DeltaChannel, added per-node timeouts, error handlers, and graceful shutdown. Deep Agents v0.7 deleted 65% of its base input tokens — including the entire base system prompt — and moved orchestration out of the tool-call loop and into code the model writes itself. Meanwhile La...

Polymarket markets look like a crowd forecasting the future. The data says otherwise: a tiny cohort of professional wallets generates the overwhelming majority of volume, while millions of retail accounts slowly pay for the forecast. This episode digs into who actually trades, what liquidity and volume on the screen really mean, and why the platform's fee revenue comes almost entirely from crypto...

Humanity's Last Exam launched in January 2025 with 2,500 expert-written questions across 100+ subjects — and the best model in the world scored 2.7%. Eighteen months later, the leaderboard leader sits at 54.8%. Is that a triumph of capability or a confession about the test? This episode digs into the origins and mechanics of HLE: Dan Hendrycks and the Center for AI Safety, the crowdsourced pipelin...

Everyone draws the reward model as a smiley face between "human feedback" and "better model" — but what is it actually doing? This episode takes the reward model apart: how it's built by cutting the unembedding layer off an SFT checkpoint and bolting on a single linear head, why it's trained on preference trios with a Bradley-Terry loss for exactly one epoch, and why a 6B critic was used to steer...

Daniel wrote in with a practical question: he understands RLHF concepts, but now he has to stand up a real pipeline. What does it actually take? This episode walks the full three-phase shape — supervised fine-tuning, reward model training, and the policy optimization loop — and lands on a counterintuitive conclusion: the algorithm is the commodity, and the job lives in the rollout engine. We compa...

When a group of students was caught in an explosion on the Ramim Ridge in the Upper Galilee, there were no commercial tourniquets on scene — so they improvised from belts and socks, in one case coached over a video call by a father simultaneously scrambling rescue services. This episode digs into what the major first aid and military organizations actually recommend about improvised tourniquets, w...

Service workers sit between your page and the network as a programmable proxy, running on their own thread with no DOM access and no window. They can wake up when no tab is open, intercept every request under their scope, and serve content from a cache you control — which is exactly why a deploy can appear to do nothing at all. This episode walks through the full lifecycle: download, install, acti...

Daniel sent in a general speech-to-text cleanup prompt built to fix the failure patterns he actually hits, and asked for a full read, a critique, and improvements. So the prompt is the artifact: we read it first, argue about it second. It does two genuinely unusual things — executing spoken self-corrections and stripping out speech aimed at someone nearby — but it also assumes every error is a wor...

Daniel has built a 6,000-episode podcast with no admin panel — just Claude Code and a custom admin MCP talking to Neon. But Neon's docs explicitly warn against connecting MCP agents to production databases, and Vercel's 45-minute build ceiling looms over every deploy. This episode works through the architecture question: does the catalog need a bespoke content API, or do Vercel's ISR with tag-base...

Daniel asked us to look at Android notifications from a developer's seat, and the honest answer isn't "native wins." It's about where control actually sits in the stack — and it flips depending on whether you mean the developer's control or the user's. Native Android hands you a rich, opinionated channel architecture with importance levels, categories, and vibration patterns, then deliberately fre...

A listener wrote one text-replacement rule in an Android dictation app and it erased every "um" and "uh" from his transcripts. Then he wrote a second rule about capitalization, and it failed — not because it was fiddly, but because it's provably impossible. This episode uses that gap as a case study in what regular expressions actually are: a 1951 notation from Stephen Cole Kleene, first implement...

Daniel asked a career question, not a technical one: is there a version of fine-tuning work that sits above the human annotation layer — and can it pay properly? The answer is yes, but the role is a bundle rather than a title, spanning Applied Scientist, Research Engineer, AI Data Curator, and forward deployed engineer postings. This episode maps the actual skill stack, the credential gate nobody...

What happens when you need to train an extraction model but the label set doesn't exist yet? This episode works through a listener's parametric-search project — scraping vendor spec sheets to build a filterable home inventory of adapters, cables, and power supplies — where the schema grows during annotation itself. We dig into attribute value extraction, why standard NER assumes a closed world, an...

Most people think mounting tape is peel, stick, press, done — but the bond is actually made in the press, and almost nobody talks about it. This episode digs into what pressure-sensitive adhesive is doing at the microscopic level, why 3M's fifteen psi bond-line target is not the same as fifteen pounds of roll-down force, and how pressure distributes unevenly across a real panel — corners included....

Modiin is one of the few Israeli cities built from a master plan, complete with a form-based code — the exact tool New Urbanists say produces coherent, beautiful cities. It still reads as extremely ugly. This episode uses Modiin as a test case against Paris as a success case, asking what a code actually has to regulate if beauty, not just order, is the goal. The answer sits in the gap between regu...

Hannah wants to build a mid-sized walkable Israeli city with a new Levantine architectural language — and she's asking what the form-based code underneath it would actually have to say. This episode digs into how form-based codes work (frontage types, regulating plans, the shopfront vs. the stoop), why they regulate the shape of buildings but almost never their style, and what Modi'in — Moshe Safd...

When a 37-year-old listener asked why the Twin Towers fell, the honest answer turned out to be narrower and stranger than the popular story. The towers were the first structures outside the military and nuclear industries ever designed to resist jet impact — a voluntary act by the Port Authority, documented in a three-page white paper dated February 3, 1964. That paper concluded a Boeing 707 or DC...

The Council on Tall Buildings and Urban Habitat spent 56 years cataloguing height records — then renamed itself the Council on Vertical Urbanism. Daniel looked up Jerusalem's Marom Tower on its Skyscraper Center database and found a much stranger story underneath: an organisation founded in 1969 by an engineer worried about livability, not records, now handing out trophies that developers put on b...

The spice bag was invented around 2006 as a staff meal at a Dublin takeaway. Poutine lives or dies on cheese curds less than a day old. Chimaek became a word during the 2002 World Cup. These aren't ancient traditions — they're recent inventions that got old fast, because a generation grew up ordering them at two in the morning. This episode tours the world's greatest drunk foods country by country...

Everything happens for a reason" sounds like one belief. It's actually at least three — God orchestrates every event, God permits events for a purpose, or God neither causes nor permits suffering but good can still come of it. This episode traces the vocabulary of divine providence from Maimonides and Nachmanides through the Baal Shem Tov's 18th-century expansion to hashgachah pratit, then compare...

What do you call someone who believes in God but belongs to no religion? Sociologist Grace Davie named it "believing without belonging" in 1990, and the phenomenon has only grown harder to pin down since. This episode digs into the vocabulary (four competing terms, no consensus), the counts (Pew finds 70% of American "nones" believe in God or a higher power, yet the global picture swings tenfold),...