Every figure is simulated from synthetic testers. The pilot has not run. Rewards for missions, the puzzle and Chronons are proposed settings, not production-approved values.
Pilot at a glance
Bottlenecks and inflation risks
Signals the model raises under the current assumptions. They describe the simulation, not observed behaviour.
Journey progress across the cohort
Show data
Wallets over the pilot
Sponsored-mission funnel
Testers by tiles revealed
Show data
Wheel V2: where the Energy goes
Show data
Sources, sinks and balances
Mean over simulated cohorts. Per active divides by testers with at least one active day. The range is the 10th to 90th percentile of cohort totals.
Rules, settings and assumptions
What the model takes as given, what the pilot proposes, and what is simply assumed.
Model-verified Game rules
Proposed Pilot settings
Hypothetical Behavioural assumptions by scenario
Measured pilot data
Nothing on this page has been observed. When the pilot runs, measured values for activity, the mission funnel, tile reveals and wallet balances belong here beside the model, so each assumption can be replaced by evidence.
How the model works and where it stops
- Synthetic testers. Each run simulates the enrolled cohort day by day, many times over with a fixed seed. Activity, steps, discovery, acceptance, completion and stake choice are assumptions, not measurements.
- The wheel is modelled by its expectations. Each attempt earns 1.045 × stake expected Progress Points and is refunded with probability 0.2%, giving 0.998 × stake net Energy. No reward table is assumed, so individual testers' Progress Points vary more than shown and no outcome is guaranteed.
- Energy is not all spent at once. On an active day a tester wagers only the assumed share of the balance, at stakes the remaining budget covers. Unspent Energy carries forward.
- One journey unless priced. After Jerusalem's 1,500 PP the model makes no further attempts unless an additional journey has both a Chronon price and a length. Progress Points past a journey's requirement are not carried into the next one.
- Missions. Each business contributes a fixed number of missions, each paying a tester at most once. One mission is in progress at a time; it resolves on the tester's next active day. Declined missions are not offered again. Only the share of the route assumed to be extra walking is added to the day's steps, so no step is counted twice.
- Puzzle. Tiles are revealed whole as soon as pieces cover one; which tile a tester chooses does not change the economy. The 100 Energy bonus is paid once, only when all tiles are revealed. Only the first puzzle is modelled; later pieces stay in the wallet.
- Rules not in the pilot brief. The bank cap and daily step bonuses of the existing case study are off unless switched on under Model settings. If the pilot build applies them, Energy figures here are wrong in both directions.
- Not modelled. Revenue, merchant performance, retention as a decision, fraud, travel modes other than walking, and Coins. No conversion or retention rate on this page has been observed.
Configuration
Download the rules, proposed settings and current assumptions as JSON. The link in the address bar also carries the current scenario and edits.