This quiz runs a topic tournament. First it compares all broad fields until
every top-level field has appeared enough times. Then it drills only into the fields and
branches you ranked highest, pruning the rest. Finally it makes cross-field comparisons among
the surviving high-rated areas.
About thirty broad-field comparisons, followed by focused drill-down and
cross-field topic comparisons. For each pair, pick whichever research direction you find
more interesting — not more important, just the one you'd rather think about.
Honest disclaimer: this is an exploratory profile, not psychometrics —
keyword-derived noisy proxy axes over a hep-ph corpus, no reliability or validity claims.
It's a conversation piece.
PHASE 2 · DRILL INTO AREAS YOU LIKED
You ranked the broad fields.
Now compare sibling areas inside the fields you ranked highest. Fields below the cut are pruned
from the rest of the quiz.
Fields kept for drilling:
Step 1
PHASE 1 · RANK BROAD FIELDS
Which research direction is more interesting?
A
B
Your six-axis profile
Your six axes
Each row is a bipolar axis: the dot marks
your position between two equally-valid poles, the faint band is the 95% confidence interval,
and the center tick is balanced. This is a position, not a score — neither
pole is "more". Faded rows are borderline (the band straddles center).
Where the quiz focused
Caveats (ship these with the result)
This is an exploratory profile, not psychometrics — no test–retest reliability, no construct validity.
Every axis is built on keyword-derived noisy proxy labels, not clean semantic truth; the exact axis directions are not unique.
D/T (Data ↔ Theory) overlaps a confounded anchor — it may track subfield more than temperament. Least reliable.
H/Z (Phenomenon ↔ Formalism) is single-family-dominant; treat it as the next least reliable.
The adaptive policy concentrates on your high-interest branches, so the axis estimate is not an unbiased sweep — that's the point, but it biases the estimate toward the areas you kept choosing.
This measures self-reported taste, not revealed behavior.