YouTube videos:
I found something strange in my childhood IQ tests https://youtu.be/6ylXBtAp95w
STORY
I have a documented psychometric anomaly with a ten-year paper trail from four different testing companies.
Over the past week I've been running blind computational experiments on the data with frontier AI systems.
THE CORE WEIRD FACT
On my 1997 WISC-III IQ test (age 8), three subtests landed at the top of the gifted range while three others landed near the bottom of it.
Using the published correlation structure of a real gifted cohort, a 50-million-profile simulation estimated that six-way configuration at roughly 1 in 36,000 gifted kids.
Not 1 in 36,000 children.
1 in 36,000 of the ones already gifted.
THE MISLEADING NUMBER
My recorded IQ was 135.
Eighteen days earlier, my Woodcock-Johnson Applied Problems score came in at 147 — roughly seventh-grade mathematical reasoning while I was in second grade.
A standard model of the relationship between IQ and achievement predicts around 123 for an IQ of 135. My 147 was roughly a 1-in-30 overshoot.
Mechanically reversing that model, a score of 147 would only look ordinary around an IQ of 167–178. That does not mean my IQ was 167–178. It shows how badly the single score of 135 summarized what was happening.
WE DON'T KNOW WHAT IS CAUSING THE SPLIT
Multiple explanations have been tested against the data — some have died under strict testing, some partially survive, none fully explain the pattern.
The mystery is real and open.
That's what makes this interesting rather than settled.
BLIND AI FINDING
A frontier AI system given only my unlabeled score sequences (with no idea they were test data) independently detected a repeatable fingerprint that survived a roughly 1-in-10,000 permutation test.
That established that the data contain real, nonrandom structure. It did not establish what causes it.
WAIS-5 COMING
Scheduling in progress to sit for the official adult IQ test.
The overall number will probably be uninteresting.
The split is the story.
Archive
Link to the full twitter thread (warning, it's chaotic and massive): https://www.somepsychagogue.quest/psychometric-mystery-archive/
Eventually, I will provide all the data
Eventually, I will have a story to link to
If you study gifted development and this interests you, reach out: amanjm.connect@gmail.com
Updates
July 20 2026
Ok, I'm in the system.
We're going to take an official WAIS-5 IQ test,
talk with a few clinical psychologists,
and see if any of them are interested in studying my profile as a case.
Claude and ChatGPT both agree that my overall IQ result will probably be uninteresting.
I'll probably score ~140 while concealing a much more dramatic internal split. The split IS the story.
The WAIS-5 is capped at ~160.
140 is 99.6%
145 is 99.87%
150 is 99.96%
155 is 99.99%
160 is 99.997%
Based on childhood testing, achievement data, and adult development, my strongest reasoning may fall around 145-155. The other stuff could easily fall below 120.
ChatGPT keeps telling me that sub-120 is not the most accurate prediction, but I'm going to say sub-120.
The story is not about intelligence.
The story is about what is causing this massive internal split in which my abilities are conditionally activated.
We've tested several explanations. Many have failed, and the ones that survive only explain a part of the pattern.
But we managed to find hard data in the old manuals and published research..my particular configuration was 1 in 36,000 gifted kids as the best grounded estimate.
My profile is simultaneously elite and abysmal among the gifted.
So Proteus is sending me to the library to find two specific intercorrelation tables so we can continue testing.