Every question's options
carry a cost.
Revealed preference. A question without cost can only measure rhetoric; a question with cost can measure your true order of preference — this is measurement, not a questionnaire.
No optimal answer, only preference
Five acts: Setting Out / Iron Armor / Mirror / Court / Return to Truth, 20 questions each. Each question is timed at 12 seconds; timeouts are recorded as "avoidance" and counted into the decision-speed dimension.
02Game-Theory Core
Revealed preference. A question without cost can only measure rhetoric; a question with cost can measure your true order of preference. Literature anchor: Samuelson (1938).
One-line insight: you aren't uncertain — you keep choosing the least painful option.
03Headcount / Duration / Venue
1 person; 20–45 minutes (5 acts × 20 questions, pausable by act); mostly online (ce.mulanxu.com).
07Result Output · Three Reporting Standards
| Report | For whom | Content standard |
|---|---|---|
| Personal report | The person | Five dimensions / 16 types / six-dimension decision radar / confidence |
| Team report | Team / HR | The whole room's aggregated distribution on the same question set (five dimensions + six roles), no names named |
| Leader Report | Decision-maker | The team profile presented as "decision preference," a team-level number, no names named |
Confidence bands: 12 choices ≈45%｜40 ≈70%｜100 ≈88%｜151 ≈95%.
08Consequence Mechanism
After finishing, no "right or wrong" is shown, only "your order of preference." A timeout = avoidance is recorded, but not punished.
09Ending Tiers (by confidence)
Mirror
Confidence ≥88%, and consistent on a retest of the same scenario.
Your choices are already answering "who you are" on your behalf.
Iron Armor
Confidence 70–88%.
Go another 60 questions and you'll see the outline close in; right now it's only a sketch.
Unseen
Confidence <70%, or a beautifying tendency detected.
The inaccuracy this time isn't the system's fault — you were just answering "who you want to become."
10Three-Part Insight
You aren't uncertain — you keep choosing the least painful option.
Do the same question set across the whole room, and the team owns a unified language for the first time (five dimensions + six roles).
Swap "personality" for "decision preference" and HR discussions finally have something to stand on.
11Play Kit
Hundred Costs · Host-Side Rulebook
1 person · 20–45 min · mostly online (ce.mulanxu.com) · 5 acts × 20 questions, pausable by act
04Rule Steps (with timing)
- Five acts: Setting Out / Iron Armor / Mirror / Court / Return to Truth, 20 questions each.
- Each question is a choice of two, and both options carry a cost (e.g., A: tell the truth but lose this deal / B: keep the deal but make the team work two extra weeks).
- Each question is timed at 12 seconds; a timeout is recorded as "avoidance" (counted into the decision-speed dimension).
- You can stop at the end of each act and get that act's summary; finish all five acts to get the full report.
- Triple anti-faking detection runs in the background; once detected, it states openly that "this report's confidence has been lowered."
05In-Game Currency and Payout Table
| Item | Standard |
|---|---|
| In-game currency | None. |
| Cost presentation | Presented as "in-story losses" (losing a client, the team working overtime, a relationship damaged). |
| Scoring | No total score across options — there is no optimal answer, only preference. |
06Observation Points and Tag Table
| id | Behavior description | Weight |
|---|---|---|
costA | Repeatedly choosing "costly to self but beneficial to others" | Connection +2, Ownership +1 |
costB | Repeatedly choosing "beneficial to self but costly to others" | Wisdom +1, Connection −2 |
timeout | Timeout avoidance ≥8 questions | Decision speed −2 |
speeder | Average response <3 seconds with low internal consistency | Wisdom −1, Growth −1 |
straight | Straight-line answering (same position ≥12 times in a row) | Confidence −15% |
socialgood | Choosing the "respectable answer" on all social-desirability items | Confidence −10% |
switcher | Changing the choice when the same scenario is reworded | Growth +2 |
consistent | Keeping the same choice when the same scenario is reworded | Ownership +1, Wisdom +1 |
Dimensions: all five dimensions covered + the six decision dimensions. Tag format follows the signals mechanism: id / t (behavior description) / vec (dimension weights); repeated triggers within one session accumulate per occurrence, capped at ±6 per session.
AHost Script (state only, don't judge)
- Opening: "There's no correct answer; choose the one you'd really choose right now — both options carry a cost."
- Timing: "Each question is 12 seconds; a timeout counts as avoidance, not a punishment, only a record."
- Acts: "You can stop at the end of an act and look at the summary first; if you want to continue, go on to the next act."
- Anti-faking: when a beautifying tendency is detected, state one line: "This report's confidence has been lowered; this time it's only a sketch."
- Closing: Present "your order of preference" only; give no right or wrong, and comment on no personality.
BThe Three Reporting Standards
- Personal report (visible only to the person): five dimensions / 16 types / six-dimension decision radar / confidence (12 ≈45%｜40 ≈70%｜100 ≈88%｜151 ≈95%).
- Team report (team / HR): the whole room's aggregated distribution on the same question set (five dimensions + six roles), no names named.
- Leader Report (decision-maker): the team profile presented as "decision preference," a team-level number, no names named.