The gold-and-blue terrain is an "opportunity landscape": height is potential reward R(x,z) at that spot, a red tint is risk V(x,z) — both are smooth pseudo-random fields, independent of each other, sampled from the founder's current position.
Each attempt samples 8 nearby candidate ventures and picks one with a softmax over expected utility — this is how risk tolerance biases the search:
U(i) = 100·R(i) − (1 − riskTolerance)·90·V(i)
P(i) ∝ exp(U(i) / T), T = temperature = 14
Landing on a venture, success is a Bernoulli trial biased by that spot's risk:
p(success) = clamp(0.92 − 0.75·V, 0.05, 0.95)
After a failure, whether the founder keeps going is a logistic function of resilience versus the current losing streak — this is the self-motivation mechanic:
p(continue) = 1 / (1 + e^-(8·resilience − 2.4·streak))
- Risk tolerance — how much a venture's expected reward is discounted by its risk when choosing where to go next.
- Resilience — raises the odds of pushing on after a failed venture instead of burning out.
- Creativity — widens the search radius for candidate ventures, from staying in a comfort zone to reaching for distant, unfamiliar opportunities.
- If the roll for p(continue) fails, the founder is burned out — auto-run stops. Pressing "Attempt Venture" again represents pushing through on sheer self-motivation, at the cost of half the accumulated streak penalty.
This mirrors real behavioural-economics models of entrepreneurship: bounded-rational search over an uncertain opportunity space (an exploration/exploitation trade-off), prospect-theory-style risk weighting, and grit research showing persistence after setbacks is a trainable trait, not a fixed one.