Skip to content
All findings

Receipts

Raw artifacts behind published findings. Prompts, outputs, scoring, and analysis. Open any kit to verify a number or replicate an experiment.

Why 'Don't Be Generic' Doesn't Work

Jul 12, 2026

Telling a model 'don't be generic' does nothing on its own; giving it specific anchors does. The gain is verifiability, not quality.

5 files

Three Questions Before You Prompt AI

Jul 5, 2026

Three questions before you type structure most of the prompt. Specificity is the strongest single lever (Hedges g=1.34); 'be exceptional' alone does almost nothing.

6 files

Three AIs, No Source, the Same Answer

May 25, 2026

Same model, same prompt. The source you paste, not the prompt you write, decides whether the numbers are real.

4 files

AI Amplifies What You Bring

May 6, 2026

Same model, same task, two paragraphs of operator context, dramatically different output. The kit, the design history, and the principle for adapting it to your own situation.

4 files

Frame Check

May 5, 2026

Drop any document in. See which analytical perspectives it covers, which it skips, the voice, what evidence backs each numerical claim. Free, open source, useful from the first paste.

11 files

Stop Calling It Hallucination

Apr 25, 2026

Hallucination is six or more distinct failure modes. Different mechanisms. Different solutions. Name the type first.

4 files

Most AI Numbers Are Fabricated

Mar 23, 2026

77 to 100 percent of AI-generated numbers are temporally unstable. Source material fixes it. Prompts don't.

9 files

How to Stop AI from Making Up Numbers

Mar 23, 2026

Source material drops unsourced numbers from roughly half to single digits. Three steps.

13 files

The Most Trustworthy AI Output Is the Least Reliable

Mar 23, 2026

The signals you use to judge AI trustworthiness are the same signals fabrication produces.

5 files