What you actually get back

Six things this assessment does that a personality quiz usually does not. Each one follows from the same starting assumption: that the person answering has a stake in the answer.

Feature one

It adjusts for the answer you wanted to give

Five of the fifty-five statements are never scored as traits. Their job is to catch the moments where your answers describe someone slightly better than anyone actually is.

Each of those five is compared against the five trait scores. A high score paired with a cross-check that should have moved with it, and didn't, is a contradiction. Contradictions are counted, and the count feeds into how much correction is applied.

How a contradiction between a trait answer and a cross-check is spotted Two answer rows. The first, a scored agreeableness statement about sympathising with others, is rated five out of five. The second, a cross-check asking whether the person can recall acting selfishly, is rated one out of five. A bracket joins them to a note saying the pair rarely holds together and raises the adjustment. SCORED ยท AGREEABLENESS "I sympathise with others' feelings." 5 CROSS-CHECK "I can recall a specific time I acted selfishly." 1 Rarely true of one person The pair raises the adjustment. Every statement is answered on the same 1 to 5 scale, so the two are directly comparable.
Neither answer is wrong on its own. It is the combination that carries the information, which is why the cross-checks only work as a set.
Feature two

You are shown both sets of numbers, not the tidier one

The correction is an argument, not a fact, so hiding the originals would be dishonest. Your results keep the scores as you gave them, the scores after adjustment, the size of the adjustment, and which traits it moved.

That means you can disagree with the method and still use the output. If you think the correction has read you wrong, the uncorrected profile is right there to read instead.

The anatomy of a results page A simplified results screen on the left with five labelled parts called out on the right: the type card with its letter and name, a confidence bar, a written interpretation, six area score bars, and deeper analysis panels beneath them. Your nearest pattern The Organiser Confidence 76% What this means Your five traits Your nearest pattern one of six found in the data How clearly it fits low means you sit between two A written reading generated from your numbers The five trait scores the part that carries the detail Deeper panels bias, coherence, response quality
A simplified view of the results screen. The headline loads first and the deeper panels fill in behind it, so a slow panel never holds up the rest.
Feature three

A profile, not just a label

The pattern name on your results is a summary of five numbers, and a summary always throws information away. The five numbers underneath are the part worth reading.

Two people can share a pattern and differ sharply in the traits that produced it. That is why it comes with a confidence figure rather than as a category you belong to, and why the trait scores get more room on the page than the name does.

Five trait scores summarised into one pattern name On the left, five labelled bars showing scores for openness, conscientiousness, extraversion, agreeableness and neuroticism. An arrow points right to a single pattern name, illustrating that the name is a compression of the five numbers. ALL OF THIS Openness 63 Conscientiousness 81 Extraversion 42 Agreeableness 71 Neuroticism 35 summarised as BECOMES THIS The Organiser one of six patterns at 76% confidence
Read left to right. The name is convenient shorthand, but almost everything that distinguishes you from someone else matched to it lives in the bars.
Feature four

Computed in one pass, then kept

Everything is worked out the moment you submit: the averages, the cross-check comparison, the adjustment, the pattern match, and the write-up. There is no queue and nothing to wait for afterwards.

All of it is stored against your assessment, so the link to your results reopens the same page with the same numbers rather than recomputing and drifting. The arithmetic steps are deterministic, which means the same answers always produce the same scores.

What happens between pressing submit and seeing results A timeline with four points: you submit, the answers are scored and matched, the summary is written, and the results appear on screen. A note underneath says everything is saved so the link reopens the same result. You submit Scored and matched averages, adjustment, nearest pattern Summary written On your screen headline first, panels behind it Everything below is saved, so returning to your link reopens the same result rather than recalculating it.
The one step that can vary between runs is the written summary, because it is produced by a language model. The numbers do not vary.
Feature five

It asks for almost nothing about you

There is no account, no sign-up, and no password. The assessment needs your age and your sex because both feed the scoring, and after that it only needs your answers.

Your record cannot be looked up by name or email, because neither is ever collected. What identifies an assessment is the link you are given at the end.

What is stored against an assessment and what is never asked for Two columns. The left lists what is stored: your age, your sex, your 35 answers, when you took it, and a technical record of the request. The right lists what is never asked for: name, email address, phone number, password or account, location, and anything from your contacts or social accounts. What is stored Your age Your sex Your 35 answers When you took it Your scores and write-up A technical record of the request What is never asked for Your name An email address A phone number A password or account Your location Contacts or social accounts
The technical record is the ordinary web-server kind: the address the request came from and which browser sent it.

Being straight about what this is not. Collecting little is not the same as being secure, and this is a prototype rather than a hardened product. Treat your results link as semi-public: anyone who has it can open your results. Do not put anything in this assessment you would mind a stranger reading.

Feature six

The categories come from an argument, not a hunch

The five traits are not this project's invention, and that is the point: they are the Big Five, measured with a public-domain instrument that has been in use for decades. What is argued for here is the correction sitting on top of them, and that argument has a traceable lineage.

It starts with fieldwork on birds, becomes a general principle about signalling, gets applied to human self-knowledge in a book published by Oxford University Press, and ends in five awkward questions and a million-response dataset used to train the model that places your profile.

Where the method comes from Four connected boxes tracing the idea: Zahavi's Arabian babbler fieldwork, the handicap principle, The Elephant in the Brain by Simler and Hanson, and this assessment's fifty public-domain IPIP statements with five cross-checks. FIELDWORK Arabian babblers Amotz Zahavi, in the Negev, from the 1970s PRINCIPLE The handicap principle a signal is believable in proportion to its cost APPLIED TO PEOPLE The Elephant in the Brain Simler and Hanson, Oxford Univ. Press, 2018 THIS ASSESSMENT 55 statements 50 public-domain IPIP, 5 cross-checks Each step is public and checkable. You can read the source and decide whether the reasoning holds. The lineage justifies the correction. It does not validate it: that has not been tested against any outside measure of behaviour.
A good pedigree for an idea is not evidence that an implementation of it works. The two are worth keeping separate.
Side by side

Where this differs from an ordinary quiz

None of the differences below are about being cleverer. They follow from one decision: treating your answers as evidence about a person with an interest in the outcome.

A typical personality quiz This assessment
What it does with your answers Takes them at face value and averages them. Estimates how much they were shaded, then adjusts each area by that estimate.
What the questions ask Whether you have a trait, and stops there. The same trait questions, plus five that ask how you are answering them.
What you are shown A label, and a paragraph about the label. Both score sets, the size of the correction, your nearest pattern, and your percentile in the reference sample.
Where the categories come from Convention, or a bespoke set with no reference data behind it. The Big Five, on a public instrument, with a million-response reference sample behind the model.
What it says about itself That it is accurate. That it is a prototype, and exactly where it stops being trustworthy.

The honest caveat, kept in view. This is a research prototype, not a validated psychological test and not a clinical instrument. It should not inform decisions about hiring, admissions, care, or anyone's standing. The full limits are here.

Try it on yourself

Thirty-five statements, roughly eight to ten minutes, no account needed.

Start Assessment See How It Works