Most personality quizzes are a lookup table wearing a costume: five questions, add up the letters, here is your result. This one is built differently, and since the difference is the entire product, it seems fair to show it.
Six axes, not one score
Every answer you give nudges six hidden values. None of them is good or bad; each is a position between two poles.
- Chaos: beautiful chaos ↔ load-bearing order
- Heart: runs on feeling ↔ runs on logic
- Drive: perpetual motion ↔ professional resting
- Dreaming: head in the nebula ↔ feet in the concrete
- Luck: suspiciously fortunate ↔ charmingly cursed
- Orbit: gravitational centre ↔ distant moon
A single answer moves an axis by a small amount. Nothing you pick can decide the ending on its own, which is deliberate: an oracle that can be steered by one dramatic answer is a toy.
The first five questions are the same for everyone
They are calibration. Between them they touch all six axes, so that after five answers there is a rough shape to work with. This also means two people can compare their openings, which turns out to be half the fun.
After that, the tree branches
From question six onward, the next question is chosen from a pool of gated candidates. A gate is a condition on your running profile. Some questions only exist for people leaning strongly one way, and two questions in the bank require a specific combination of two axes before they will appear at all.
The selector alternates between two intentions. On even steps it digs into the axis that is already shouting, so the ending is decisive rather than lukewarm. On odd steps it probes whichever axis it knows least about, so you do not finish as one spike surrounded by blanks. Gated questions get a scoring bonus, because a question that only unlocked for you is the most interesting thing the oracle can ask.
Then twelve endings compete
Each archetype is a weighted opinion about which way the axes should point. The Quiet Engine wants low Orbit and high Drive. The Lucky Fool wants high Luck and some Chaos. Your final profile is scored against all twelve and the strongest agreement wins. Because the rule sets overlap, the boundary between any two archetypes sits in a different place for nearly every combination of answers.
And a seed writes the details
Which three prophecies you get, which warning, which lucky object: all drawn by a seeded random number generator whose seed is a hash of your exact answer sequence. Two people who land on the same archetype by different routes get different cards. The same person answering identically gets the identical card, forever, on any device.
That last property is why a shared link works with no server: the fortune is not stored anywhere, it is recomputed from twenty-one characters in the URL.
None of this makes the result true. It makes it consistent, personal and reproducible. That is the most any oracle has ever honestly offered.