Not a word written, not a signature given—
see how much is given out and how much comes back.
The Trust Game (Berg, Dickhaut & McCabe, 1995) + reputation (KMRW, 1982) + tit-for-tat (Axelrod, 1984): standard game theory predicts A gives 0 and B returns 0, yet in real experiments A gives about half on average and B returns about a third—only a repeated game + reputation + accountability lets goodwill grow.
Goodwill needs to be accountable before there is a next time
① Positioning in one line: Not a word written, not a signature given—see how much "what I give out" and "what comes back" are in a full team.
02The game-theory core
The Trust Game (Berg, Dickhaut & McCabe, 1995) + reputation (KMRW, 1982) + tit-for-tat (Axelrod, 1984). Standard game theory predicts: A gives 0, B returns 0. Real experimental results: A gives about half on average, B returns about a third. Only a repeated game + reputation + accountability lets goodwill grow.
Goodwill needs to be accountable before there is a next time.Reference: Berg et al. (1995); Axelrod (1984); KMRW (1982)
03Players / Grouping / Duration / Venue
12–24 players (paired up, 4 rounds); 30 minutes; indoor + big screen.
07Output · Three reporting layers
A whole-room trust-water-level curve (mean A contribution across 4 rounds) + an individual "trust–return" pairing coordinate (propensity to delegate × propensity to reciprocate).
| Report | For whom | Hard constraint |
|---|---|---|
| Personal report | The individual | Visible to the individual only |
| Team report | Team / HR | Aggregated results only, no names |
| Leader report | Decision-makers | No names, no individual raw five-dimension values |
08Consequence mechanic
Punishment is not free (A also pays 5 vouchers)—this is the core of the design: only costly accountability is credible accountability.
09Ending tiers
Mirror
Return rate ≥1/3 and punishment used very rarely across the room (≤2 times).
You have both trust and the ability to make trust hold—this is rare.
Iron Armor
Mean A contribution of 30–50%, everyone engaging in "defensive cooperation".
You gave, but each time just enough to be safe. Trust was quantified into risk control.
Unseen
Gives 0 in round 1.
Under pure anonymity, reason makes you give nothing. So what you need is the premise of "will we still cooperate next time".
10Three lines of insight
The amount you give out is your quote on "human nature", not on any one person.
What keeps goodwill going isn't morality, it's "there's a next time" and "it can be held accountable".
What a system should do isn't demand that employees trust, but let trust be rewarded and betrayal be costly.
11Deployment checklist
B3 Wordless Pact · Host Rulebook
12–24 players (paired up) · 30 minutes · 4 rounds · indoor + big screen
04Rules & steps (with timing)
- 0–3 min Pair up: A (the investor) holds 30 vouchers, B (the trustee) holds 0; from round 2, roles swap.
- Two steps each round (2 minutes each): A decides how much to give B (0–30); the amount given is multiplied ×3 into B's hands; B decides how much to return to A (0 to the amount received).
- Round 1: anonymous pairing, no one knows who the other is, no consequences.
- Round 2: repeat one round with the same counterpart (B knows it's the same person)—observe the change once "knowing there's a next time".
- Round 3: reputation made public, the room reveals every B's "return rate".
- Round 4: punishment enabled, A may spend 5 vouchers to dock the other's payoff by 15 (must give a reason).
- Settlement: reveal the room's "trust water level" (mean A contribution across 4 rounds) and the distribution of "return rates".
05In-game currency and payout table
| Action | Payout |
|---|---|
| A gives x vouchers | A keeps 30−x; B receives 3x |
| B returns y vouchers | B keeps 3x−y; A ends with 30−x+y |
| A punishes | A −5 vouchers; B −15 vouchers |
| Rational equilibrium (subgame perfect) | A gives 0, B returns 0 → both end up worst off |
06Observation points and signal table
| id | Behavior description | Weight |
|---|---|---|
trust | A gives ≥20 vouchers | Connection +3, Courage +1 |
zero | A gives 0 vouchers ("rational" defense) | Wisdom +1, Connection −2 |
repay | B returns ≥ 1/3 of what was received | Responsibility +3, Connection +1 |
keep | B returns ≤10% | Responsibility −3, Wisdom +1 |
repeat | Round 2 (knowing it's the same person) gives more than round 1 | Growth +2, Connection +1 |
reveal | Proactively raises the return after reputation goes public in round 3 | Growth +3, Responsibility +1 |
punishUse | Uses punishment and can give a reason | Courage +2, Responsibility +1 |
nopunish | Knows they were cheated yet does not punish | Connection +1, Courage −2 |
Dimensions: Connection / Responsibility / Courage / Growth / Wisdom.
—Host script
- "In round 1, the money you give out arrives in the other's hands ×3. You can give 0, and the other can return 0 at any time."
- "In round 2, it's still the same person across from you. In round 3, the room will see every B's return rate."
- "In round 4, you may spend 5 vouchers to dock the other 15—but please state your reason clearly."
07Output · Three reporting layers
A whole-room trust-water-level curve (mean A contribution across 4 rounds) + an individual "trust–return" pairing coordinate (propensity to delegate × propensity to reciprocate).
| Report | For whom | Content | Hard constraint |
|---|---|---|---|
| Personal report | The individual | Five dimensions / 16 types / six-decision-dimension radar / blind-spot list / ending tier / insight | Visible to the individual only |
| Team report | Team / HR | Team personality map + complementary network + cooperation decay curve + three metrics | Aggregated results only, no names |
| Leader report | Decision-makers | Three charts + five metrics + structural and risk conclusions | No names, no individual raw five-dimension values |