David Wolfe — AI-driven revenue systems

TOOL

Stress Test · Scenario Studio

Stress-test the Power thesis

Ask what happens to the modeled advantage when an important assumption changes.

Choose a questionChange a conditionRunSee whether advantage survives

Exploratory scenario model — not a forecast.

Scenario workspace

Experience staleness

Current test

Experience staleness

BaselineWhat happens under the current assumptions?Better foundation modelCan a challenger compress the incumbent's historical experience advantage with higher base model capability?Faster learnerCan a challenger catch up by extracting more learning from each case?Transfer breakdownWhat if accumulated experience transfers poorly across customers or contexts?Feedback delayWhat if real-world outcomes take longer to arrive and grade?Experience stalenessWhat if historical experience loses relevance quickly?Continuous capture advantageWhat if the incumbent continuously generates graded experience faster than challengers?CustomCreate your own competitive scenario.

What if historical experience loses relevance quickly?

Primary changeMonthly staleness

Challenger B monthly staleness 0.012 -> 0.055

ASSUMPTION-DRIVEN SCENARIO

Trajectory reflects scenario assumptions, not measured company performance.

Conditions

What changed

Monthly stalenessChallenger B: 0.055 ↑Incumbent A: 0.015

All other assumptions remain at baseline unless edited below.

Edit all assumptions

From Experience dataset

Current Experience supports 0 starting graded cases and unavailable median feedback delay (n=0).

Only starting graded cases and feedback delay are derived from Experience. Information value, transferability, learning efficiency, staleness, and base capability remain assumptions.

Variable	Incumbent / Company	Challenger / Alternative
Name
Starting graded casesHow much matured experience the system begins with.Observed / derived
Cases / monthHow quickly new potentially learnable experience is generated.Endogenous assumption
Feedback delayHow long before a decision produces an outcome that can be graded.Observed / derived
Learning efficiencyHow effectively the system converts graded experience into improved expertise.Endogenous assumption
Information value / caseHow much non-redundant learning a typical case contributes. This remains a scenario assumption until the Shannon layer exists.Exogenous assumption
TransferabilityHow much learning from prior cases applies to future/customer contexts.Exogenous assumption
Base capabilityCapability available without proprietary accumulated experience, for example from the underlying frontier model.Exogenous assumption
Monthly stalenessHow quickly accumulated experience loses relevance.Exogenous assumption

Experience advantage

Starting graded cases: How much matured experience the system begins with.

Cases / month: How quickly new potentially learnable experience is generated.

Learning dynamics

Feedback delay: How long before a decision produces an outcome that can be graded.

Learning efficiency: How effectively the system converts graded experience into improved expertise.

Information value / case: How much non-redundant learning a typical case contributes. This remains a scenario assumption until the Shannon layer exists.

Transferability: How much learning from prior cases applies to future/customer contexts.

Competitive / environmental conditions

Base capability: Capability available without proprietary accumulated experience, for example from the underlying frontier model.

Monthly staleness: How quickly accumulated experience loses relevance.

Change test

Ready to run

Experience staleness

What if historical experience loses relevance quickly?

This scenario changes Monthly staleness: Challenger B monthly staleness 0.012 -> 0.055. Run the stress test to see whether the modeled advantage survives.

No result is shown until this scenario is run, so stale results from a previous test are not reused.

Secondary tools

Save / duplicate scenario

Compare saved scenarios

Name	Stress-test type	Result	Crossover	Final gap
Challenger B vs Incumbent A	Experience staleness	Advantage Persists	None	+2.59

How the model works

Discretization is monthly over a 36-month horizon. `feedbackDelayDays` is converted to `ceil(feedbackDelayDays / 30)` monthly steps. Cases wait in a pending queue until they mature. `stalenessRate` is monthly decay on accumulated effective graded experience. Expertise uses a logarithmic response to effective experience so historical advantage can be compressed by better base capability, low transferability, slow feedback, or high staleness.

N_eff(t+1) = (1 - staleness_rate) * N_eff(t)
             + cases_whose_feedback_matures_at_t

E(t) = base_capability
     + learning_efficiency * information_value * transferability * ln(1 + N_eff(t))

Exploratory toy model only. A crossover is not a real-world forecast.

Next: Synthesize the investment thesis

Conclusion combines Company Model, Experience, Debates, Power, and Stress Tests into a concise statement of what appears true, what remains uncertain, and what evidence would most change the thesis.

Continue to Conclusion →