Saturday, 24 October 2026
25 episodes · 6596 in all
#6258: Loop Earplugs, Model by Model
Loop Engage 2 carries an SNR of 16 decibels but an NRR of just 8 decibels, showing how regional testing standards diverge.
#6257: Where Your Prompts Actually Go
Selecting US-only inference on Anthropic's direct API incurs a 1.1x multiplier across token categories, while global routing is the standard rate.
#6256: How Surveillance Turns Routine into Identity
NSA's SKYNET analyzed roughly 55 million mobile identifiers in Pakistan to find unknown couriers based purely on travel patterns and call ratios.
#6255: Routing Claude Code Tasks to Local Ollama Models
A feature request to support local subagent endpoints in Claude Code was closed for inactivity after less than two months.
#6254: Where a Synthetic Voice Gets Its Accent
A neural speech interface maps motor cortex signals into phonemes drawn from an eclectic training set of human voice recordings.
#6253: Why Context Windows Stalled at a Million Tokens
A dense 70B model requires about 328 gigabytes of KV cache to serve a single one-million-token prompt, exceeding the memory of an Nvidia B300 chip.
#6252: How the Kilogram Leaves the Basement
A third consensus value for the kilogram took effect on March first, differing by five micrograms from its predecessor.
#6251: How Ancient Markets Fought Fraudulent Measures
An Athenian decree from 120 BC specified container rim widths down to half-fingers to prevent merchants from rigging heaped grain measures.
#6250: Why Satisfaction Is Not Enough for a Good Life
Aristotle argues that human flourishing requires exercising rational capacities excellently across a complete life, not just feeling satisfied.
#6249: Why Caused Choices Can Still Count as Free Will
In the 2020 PhilPapers Survey, over 62 percent of professional philosophers endorsed or leaned toward compatibilism between free will and determinism.
#6248: Why Death Might Harm Us Without Hurting
Thomas Nagel's deprivation account argues death harms us not through bad sensations, but by cutting off future goods we could have experienced.
#6247: Why Meaning Is Neither Pure Feeling nor Plain Duty
Studies show people identifying as givers report higher life meaning than takers, but distinctly less moment-to-moment happiness.
#6246: Why Is There Something Rather Than Nothing?
Peter van Inwagen argued an empty world has probability zero because there are infinitely many populated possible worlds and only one empty one.
#6245: Why Inference Engines Change Benchmark Scores
A study found a 16.6 percentage point spread on GSM8K across different inference engines running identical weights on the same GPU.
#6244: When Your Fine-Tune Isn't Done Learning Yet
Falling loss, 90% word error rate, and a model that still says "Derry gym lag." What undertraining actually looks like — and when more epochs won't...
#6243: Inside the Mixture of Experts Router
What actually decides which expert handles your token? A look inside the router, from sigmoid affinities to load balancing.
#6242: Training a Model on Your Own Taste
What does it actually take to train a small model on your editorial taste using pairwise preferences and DPO?
#6241: Why Your Synthetic Data Isn't as Diverse as You Think
Fifty-nine percent of one GPT-4 dataset started with "Once upon a time." Correct examples can still make a terrible training set.
#6240: Gold vs Silver Training Data: What Makes Labels Trustworthy
Human-checked "gold" labels and machine-made "silver" ones aren't as different as the names suggest — and the gap matters.
#6239: Testing Synthetic Data Generators Before You Scale
A benchmark trains student models on each generator's output to find out which model actually writes useful training data.
#6238: Instruction Models and How We Benchmark Them
Open weights vs closed APIs for synthetic data, and why IFEval scores don't tell you which model will hold up.
#6237: Ranking Instruct Models by Intelligence
Three million fables for pennies — but can a small model tell a good example from a plausible one?
#6236: Synthetic Logs Without Leaking Your Clients
You can rewrite your logs with a cheap model — but can you keep the task boundaries while losing the client names?
#6235: When Your AI Agent Has Too Many Skills
Your agent has 55,000 skills installed — and it's spending the whole task just reading the menu.
#6234: Using Your Phone as a Networked Dictation Mic
Your phone's internal mic may beat a dedicated dictation microphone. Here's how to route it to your desktop without losing what makes it work.