median

Definition

Plain language

The middle value in a list of numbers, with half above it and half below.

Why it matters: It provides a stable neutral reference point that isn't thrown off by a few unusually large or small values the way an average would be.

For example, in the list 2, 5, 100, the median is 5 because it sits in the middle, unswayed by the extreme value.

Heard on the show

“With the refusal pin alone, the marker gets buried on the covered prompts — median first-token rank of 217 — but stays near the surface everywhere else.”

Mentioned in 24 episodes

1. 245
   Fifteen Models Ran Football Clubs for Twenty Years, and Size Didn't Decide It
2. 244
   The Open-Weight Defense That Feeds Attackers Confident, Falsified Answers
3. 243
   How a Hundred Meaningless Word Choices Add Up to Flip a Model's Answer
4. 232
   Coding Models Can Find the Bad Line, They Just Won't Delete It
5. 222
   The Bias Isn't in Your Prompt — It's Inside the Model
6. 209
   How 2.6 Billion Doodles Exposed the Culture Words Quietly Delete
7. 204
   The Length Estimate Hiding Inside a Word-by-Word Model
8. 200
   The One Mechanism That Turns Twenty AI Clones Into an Actual Team
9. 157
   When an AI Coding Agent Drives a Phone Through the Terminal, No Screen Needed
10. 156
    Why More Human Demonstrations Made a Computer-Use Agent Worse
11. 149
    When Cornering a Chatbot Makes It Lie: J.P. Morgan's Case for 'Playing Dead'
12. 140
    When a Reasoning Model Says "Let Me Double-Check" After It's Already Decided
13. 139
    When Optimizing One GPU Kernel Quietly Breaks the Whole System
14. 125
    AI Coding Agents Run a Marathon, and Fewer Than One in Three Finish
15. 078
    Training a Markdown File: When LLM Self-Improvement Borrows the Discipline of Neural Net Training
16. 069
    When Smarter Models Forecast Worse: The Hidden Failure Mode in LLM Predictions
17. 068
    The OS Trick That Makes Tree Search Practical for Coding Agents
18. 053
    An AI Agent Swapped In Focal Loss And Beat A Human-Tuned Training Script
19. 048
    How a 30B Open Model Reached Olympiad Gold With the Right Recipe
20. 040
    Two Frozen Models Learn to Whisper: Coupling Through Hidden States
21. 037
    Why Hallucination Detectors Miss Stale Facts: A Geometric Story About What Models Know But Don't Say
22. 032
    A Sticky-Note for Every Layer: Letting Transformers Remember What They Were Just Thinking
23. 027
    When AI Agents Build the Serving Stack: A Bet on Bespoke Infrastructure
24. 004
    The Sycophancy Circuit That Survives Alignment Training