My Weird Prompts

The handshake feels eternal, but it isn't. The oldest image of two people clasping hands is a ninth-century BCE limestone relief of two kings renewing a treaty — not a greeting at all. So how did a rare gesture between monarchs become something you do at the school gate? This episode traces the handshake from Assyrian and Roman treaty rituals through the Quakers' radical insistence on greeting everyone the same way, to a Russian Empire that banned handshakes as indecent and prescribed mitten-clad hand-striking instead. Along the way: the gongshou, namaste, the wai, the Japanese bow, and the salaam — gestures with lineages older than the handshake itself. The answer to whether the handshake is a human universal is no. It's a Western export that got very, very lucky.

Key sources:

• The Conversation, From ancient kings to Trump and Xi — https://theconversation.com/from-ancient-kings-to-trump-and-xi-when-did-humans-start-shaking-hands-and-why-283078

• HISTORY, Evan Andrews, orig. 2016-08-09, updated 2026-05-07 — https://www.history.com/articles/what-is-the-origin-of-the-handshake

• Smithsonian Center for Folklife, 2021-12-05 — https://folklife.si.edu/magazine/handshake-history-folklore

Episode 5979 — myweirdprompts.com/n/5979 (permanent short link: myweirdprompts.com/489526)

A friend came back from Ireland changed by two words from a stranger. That set off a question: why does a phrase like "God bless" survive in a country that has largely stopped believing — while Israel's "be'ezrat Hashem" is moving into pop lyrics and the Muslim world's "Inshallah" quietly changes the subject in conversation? This episode traces the linguistics of pragmaticalisation: how prayers wear down into punctuation, how "goodbye" used to be "God be with you," and how a 1910 account of Irish farmyards priced the rudeness of skipping a blessing at one seventh of a workman's feeding. Three societies, one process, three directions — and a hard edge at the end, because for people who don't believe, hearing God invoked at you all day can feel less like a blessing and more like a hand on the back of your neck.

Key sources:

• RTÉ Brainstorm, Dia dhuit: what's behind Irish language's religious roots? by Peter Weakliam (DCU), 2024-08-22 — https://www.rte.ie/brainstorm/2024/0822/1406358-irish-language-gaeilge-religious-roots-dia-dhuit-diabhail/

• The Irish Times, A blessing in disguise (An Irishman's Diary), Frank McNally, 2013-11-29 — https://www.irishtimes.com/culture/heritage/a-blessing-in-disguise-1.1610750

• Shlomi Rosenfeld, HaShem: Why Israelis Don't Say God's Name & What They Say Instead (*The Israeli Story 1948 — https://www.shlomirosenfeld.co.il/en/hashemisgod

Episode 5978 — myweirdprompts.com/n/5978 (permanent short link: myweirdprompts.com/669548)

What did Russian illegals actually say to each other? In June 2010, the DOJ unsealed a complaint that put real recognition phrases into the public record — "Excuse me, but haven't we met in California last summer?" — and the answers came back with corrections, as if two colleagues were arguing about a shared memory. This episode sorts four different machines that get constantly confused: agent codenames, operation names, recognition phrases, and physical signals like a folded newspaper in an Arlington park. It traces the CIA's two-letter digraph system (AMTHUG, AMQUACK), the Stasi's strange tradition of letting informants name themselves (Figaro, Ramses, Mikrobe), and Markus Wolf's three-card-index compartmentalisation that made the names almost irrelevant. And it confronts the twist: the New York Times reported that some of the 2010 phrases were supplied by an undercover FBI agent, meaning the best-documented example of spy tradecraft is partly counterintelligence theatre.

Key sources:

• FBI — https://www.fbi.gov/news/stories/operation-ghost-stories-inside-the-russian-spy-case

• BBC News — https://www.bbc.com/news/10442869

• Reuters / The Moscow Times — https://www.themoscowtimes.com/archive/excerpts-from-us-justice-department-papers-on-spy-ring

Episode 5977 — myweirdprompts.com/n/5977 (permanent short link: myweirdprompts.com/455797)

North of North drops you into Ice Cove and never stops to explain anything — which is exactly the point. So when a listener wrote in admitting that "Eskimo" was basically the whole extent of her Arctic vocabulary, we decided to do the work properly. This episode covers who the Inuit actually are: roughly 155,000 people across four countries speaking dialects of one language family, descended from the Thule who spread across the top of the world a thousand years ago. We get into the word "Eskimo" itself — why it's not the raw-meat slur most people think, why Canada and Greenland dropped it in 1977, and why Alaska never did. Then the traditional half: a diet that got 75% of its energy from fat, vitamin C from raw seal liver and muktuk, the amauti as both baby carrier and social document, and gender roles that bent whenever survival required it. Finally, the modern half — a fully contemporary people who are also fully Inuk, and a show made by Inuit hands in Iqaluit that refuses to translate itself for outsiders.

Key sources:

• UAF Alaska Native Language Center, Inuit or Eskimo: Which name to use? (Lawrence Kaplan) — https://www.uaf.edu/anlc/research-and-resources/resources/archives/inuit_or_eskimo.php

• UAF Alaska Native Language Archive, Comparative Eskimo-Aleut — https://www.uaf.edu/anla/collections/cea/

• CBC Media Centre, North of North program page — https://mediacentre.cbc.ca/program/north-of-north

Episode 5976 — myweirdprompts.com/n/5976 (permanent short link: myweirdprompts.com/779723)

When a speech recognizer and a correction model do one job, can they be folded into a single set of weights? The answer depends entirely on a distinction most coverage skips: model merging and distillation are not the same operation. Merging combines parameters from models that share a pretrained ancestor — no training, no extra inference cost, same parameter count. Distillation trains a new model to imitate others, and it's the only route that works across different architectures. This episode walks through task arithmetic, weight averaging, re-basin permutations, and LoRA merging to find exactly where each one breaks, then follows the distillation route from FuseLLM to cross-architecture transfers between CNNs, transformers, and Mamba. Along the way: why an elaborate feature-alignment loss made a student model worse than plain output matching, and why you can't average a lesson into a model that never learned it.

Key sources:

• Model Merging in LLMs, MLLMs, and Beyond survey (Yang et al.) — https://arxiv.org/abs/2408.07666

• Donato Crisostomi, Model Merging — https://crisostomi.github.io/blog/2025/model_merging/

• Uncertainty-Based Gradient Matching (ICLR 2024) — https://arxiv.org/abs/2310.12808

Episode 5975 — myweirdprompts.com/n/5975 (permanent short link: myweirdprompts.com/586126)

Whisper bakes punctuation into its weights. Parakeet-TDT emits it natively from a transducer. And yet a 484-megabyte text model called S1-mini exists purely to clean up after speech recognizers — and it never hears a single second of audio. So where does recognition actually end and cleanup begin? This episode works through the architecture question from the ground up: CTC, RNN-T, TDT, encoder-decoder transformers, and the newest audio-language foundation models, and what each one can and cannot emit. The answer turns out to hinge less on topology than on training targets — and on a metric that didn't exist until recently.

Key sources:

• Superwhisper blog, Introducing the S1 family of models, 2026-08-19 — https://superwhisper.com/blog/s1

• Hugging Face, superwhisper/s1-mini model card (release v1) — https://huggingface.co/superwhisper/s1-mini

• arXiv:2212.04356, Radford et al., Robust Speech Recognition via Large-Scale Weak Supervision (Whisper) — https://arxiv.org/abs/2212.04356

Episode 5974 — myweirdprompts.com/n/5974 (permanent short link: myweirdprompts.com/558256)

Model Context Protocol servers are multiplying fast — 97 million monthly SDK downloads, over 10,000 public servers — but the protocol assumes every server is trustworthy. Tool descriptions load into the model's reasoning context with system-prompt authority, and there's no cryptographic attestation or native injection detection to catch tampering. This episode works through what MCP-specific observability actually requires: tool-call logging rather than HTTP logging, agent identity as a first-class principal, client-side hooks for the 86% of servers running locally, and write-time signed audit trails. Then the attacks that break conventional monitoring — tool poisoning, rug pulls like CVE-2025-54136, tool shadowing, and the Microsoft-described pattern where every individual action is authorized but the combination isn't. Plus the exposure numbers: 492 unauthenticated MCP servers found exposed to the open internet, a count that nearly tripled on rescan. Anchored by three products with three different postures — Mintive, Obot, and Docker MCP Gateway.

Key sources:

• MCP spec, Security and Trust & Safety section — https://modelcontextprotocol.io/specification/2025-06-18

• Mintive SIEM export docs (OTLP/Splunk HEC schemas) — https://docs.mintmcp.com/docs/siem-export

• Mint Guard docs — https://www.mintmcp.com/docs/mint-guard

Episode 5973 — myweirdprompts.com/n/5973 (permanent short link: myweirdprompts.com/403507)

Meeting assistants have quietly stopped being notetakers. In 2026, they book follow-ups, update Salesforce and Jira, and in some cases attend calls on your behalf. This episode sorts the shipping capabilities from the speculative ones — pre-meeting retrieval, real-time research agents, behavioral coaching, and true asynchronous stand-ins — and asks the harder question underneath all of it: what are the social rules becoming? Drawing on a three-part taxonomy of recorders, text relays, and genuine stand-ins, plus concrete product rollouts and a telling survey about whether participants want an agent to raise its hand, the conversation lands on the real battleground: not the assistant, but the corpus it queries.

Key sources:

• Zoom, Meet ZoomMate — https://www.zoom.com/en/blog/zoom-ai-companion-zoommate/

• Microsoft Learn, Overview of AI in Microsoft Teams for IT admins — https://learn.microsoft.com/en-us/microsoftteams/copilot-ai-agents-overview

• Microsoft 365 Message Center MC1429018, Meeting memory file generation (rollout Nov 2026) — https://mc.merill.net/message/MC1429018

Episode 5972 — myweirdprompts.com/n/5972 (permanent short link: myweirdprompts.com/296765)

Personal CRM keeps getting declared dead, yet there's now a protocol server with 88 tools for managing your friendships. We break down what separates a real personal CRM from a repurposed sales pipeline, survey the SaaS options (Dex, Mesh, folk) against the self-hosted ones (Monica, Twenty, EspoCRM) on features, pricing, privacy, and portability, and ask the question that actually matters in 2026: which platforms can an AI assistant talk to? Is the best personal CRM a specialized product, a repurposed business tool, or just a well-structured contact database wired to an intelligent assistant?

Key sources:

• Dex MCP Server docs , accessed 2026-10-14 — https://getdex.com/docs/ai/mcp-server

• folk pricing page , accessed 2026-10-14 — https://www.folk.app/pricing

• Monica blog , 2026-08-06 — https://www.monicahq.com/en/blog/new-version-of-monica-in-2026/

Episode 5971 — myweirdprompts.com/n/5971 (permanent short link: myweirdprompts.com/151459)

On Ubuntu you're not locked to the kernel your distribution shipped — you can pull a newer upstream one, stay on an older one, or install a kernel someone rebuilt for a specific kind of work. This episode maps the whole landscape: how Ubuntu decides what to ship (including release candidates under its 2024 policy), what actually changes between kernel versions, why hardware support and performance are two very different reasons to switch, and what Liquorix, XanMod, and the now-mainline PREEMPT_RT kernel each optimize for. The most interesting moment comes when XanMod's own documentation labels the AVX-512 build as having no kernel benefit — the vendor of the CPU-optimized kernel undermining the enthusiast argument. Along the way: EEVDF replacing the completely fair scheduler, the throughput-versus-latency axis every tuning choice sits on, and why desktop and server motivations split apart.

Key sources:

• Liquorix Kernel homepage — https://liquorix.net/

• XanMod Kernel homepage, builds dated 2026-10-11 — https://xanmod.org/

• Canonical Kernel Team policy, 2024-08-09 (updated 2025-08-11) — https://discourse.ubuntu.com/t/kernel-version-selection-for-ubuntu-releases/47007

Episode 5970 — myweirdprompts.com/n/5970 (permanent short link: myweirdprompts.com/310056)