10 reproducible studies · open data · open code

Can Courts Actually Detect Lies?

The "deception cues" used in courts, police interrogations, and jury rooms are scientifically inverted. They flag truthful people as liars. Here's the proof — run it yourself.

10
Studies
73.9%
Cues inverted
1.8B
Person-years
75
Countries

If you're autistic, you look guilty by default. If you're not white, you look guilty by default. If English isn't your first language, you look guilty by default. This isn't opinion. Run the numbers yourself.

The 10 Studies

01

Trial Testimony

Trained a computer on the same "deception cues" courts use. It couldn't tell truth from lies. Humans: 54%.

SVM / ML N=121
02

Confession Linguistics

False confessors say "you" 7.59x more and "that's" 11.67x more. They're echoing interrogators, not confessing.

Mann-Whitney U N=135
03

Belief vs Reality

74% of things people think indicate lying actually indicate truth-telling. Surveyed across 75 countries.

Binomial test N=11,227
04

Convergence

Meta-analysis across all studies. Same direction, same conclusion. Not a fluke.

Meta-analysis
05

Four Pillars

Four completely independent lines of evidence. Detection at chance, memory distortion, false confessions, detention effects.

Multi-method
06

Autism Compounding

Autistic people's natural behaviour maps perfectly onto what courts call "deceptive." 3.5x more false alarms.

Kruskal-Wallis d=3.18
07

Cross-Cultural

A Mexican person telling the truth looks "deceptive" to a white American jury. Same for Romanian, Indian speakers.

Cross-classifier 4 cultures
08

Language & Birthplace

Where you're born determines what you speak. Not genes. Census data, N=1.8 billion person-years.

Census regression N=1.8B
09

Environmental Determination

Your postcode predicts your outcomes better than anything about you personally. Language predicts college better than ancestry.

Logistic regression ~390K
10

The Framework

Mathematical proof of what real consensus looks like. Borrows from Bitcoin. Courts produce verdicts, not truth.

Mathematical proof

Run It Yourself

You don't need a science degree. You need Python and 5 minutes.

1

Get the code

Download the ZIP from GitHub, or:

git clone https://github.com/webmasterproT/signal-inversion-research.git
cd signal-inversion-research
2

Set up (one time)

bash setup.sh          # Mac / Linux
python3 setup_windows.py   # Windows
3

Run all studies

bash run_all.sh        # Mac / Linux
python3 run_windows.py     # Windows

Or run just one: bash run_one.sh 7 (cross-cultural)

4

See the results

Each study's results/ folder has the output. results/figures/ has the charts. WHAT_THIS_PROVES.md explains it in plain language.