The Visionary Psychometric Scan
An untimed, adaptive reasoning battery built for the high range, in beta calibration since July 2026.
The Scan is 36 items long, with eight options per item. It is administered online, on a computer, with paper and pencil allowed and no clock of any kind. It adapts after the first tier, so most takers see fewer than 36 items. It asks for no outside knowledge, and its report describes the shape of a person's reasoning rather than producing one number to rank them by.
What it measures
- Fluid reasoning Gf
- Finding the rule in material you have never seen before, and applying it.
- Crystallized density Gc
- Precision with concepts you already hold: arbitrating between near meanings and reading structure in passages. Breadth of knowledge is deliberately not what this coordinate counts.
- Visuospatial reasoning Gv
- Holding and transforming spatial structure without a clock running.
- Quantitative reasoning Gq
- Reasoning about quantity and relation, with arithmetic kept out of the way as far as the format allows.
- Integrative intelligence Gi, exploratory
- Reasoning across domains at once. This coordinate is treated as exploratory until the calibration data earns it more than that, and it will be reported as such.
How it is built
The items climb through three tiers. The Gate starts near average difficulty and climbs quickly, so most takers strain from the first items. Clearing it opens The Bridge, and clearing that opens The Zenith. During the Beta, advancement is decided on total score; the intended design advances takers domain by domain, and that is one of the changes the calibration will inform.
Three choices that set it apart
No clock
Timing constraints reward speed, and speed is not the ability the Scan exists to measure. Removing the clock and allowing a scratchpad takes cognitive proficiency out of the equation as far as a reasoning instrument can, which matters most for people with uneven profiles, whose reasoning often needs more time to show up than a timed format allows. The costs are real, and they are stated: fixation on a hard item, the opportunity to consult a model, and fatigue over a long session. The full argument is in the publication.
No outside knowledge
Nothing on the Scan rewards having read the right book. Where a coordinate touches language, it tests inference from the material in front of you rather than recall of material you brought with you.
A fifth coordinate
Integrative intelligence, meaning reasoning across domains, is included as a coordinate because the ability seems real and underexamined. It ships labeled exploratory, and it stays labeled that way until the data says otherwise.
Taking the Beta
The Beta is open through the winter. It is hard from the first items, most people will not clear The Gate, and it takes 45 to 90 minutes on a computer. A quiet hour, a sheet of paper, and a willingness to be stumped are the whole kit.
- You must be 18 or older.
- Work alone, without search engines, AI tools, or calculators.
- Use a computer. Phones do not render the items well.
- Please do not discuss specific item content with anyone afterward.
Beta takers receive a preliminary profile report by email. Because the norms are exactly what this phase is building, the report describes the shape of your reasoning instead of a population-relative score, and every Beta taker receives an updated report once the norms mature.
The floor of the instrument is not the floor of the person. If your session ends early, the report will say exactly that.
What it cannot tell you yet
The Scan is not a clinical instrument and is not designed for selection decisions by employers, schools, or anyone else. Its norms are provisional and will be published as provisional, with the sample described and the error bars visible, until the data earns stronger claims. Resistance to language models is treated as an open problem under test rather than a feature, which is why process-level checks run on every session. A technical report will be published alongside version 1.0.
Calibration notes
As of the September calibration note, 47 people had started the Beta, 8 had finished, and 6 had reached The Zenith. Most people who stop do so on the forms before the first item or at the second item, which is the instrument's floor doing what it was designed to do and also a sign that the on-ramp is too long. Both facts are published, because that is the point.