← back to paper map
ARENA · v5+v6 v5 mechanics + v6: phase isolation · autorun (no stances) · disposition gates · two-layer partnerships

Moatfight Arena v5

Strategies are the only force that migrates share. Outcomes per side are independent · acquired · extinct, each with a $ value. Run 100+ trials to see what patterns emerge.

ATTACKER phase —
Selected for deployment
Budget
click a play below to deploy
Available plays click to add
— pick attacker to preview —
AI-era moat tags
🌀
flywheel
data-driven feedback loop: usage feeds the model, model improves outcomes, better outcomes attract more usage.
🔗
agentic
workflow lock-in via agent memory & calibration — the agent learns the customer's context and switching means losing it.
⚖️
evaluator
institutional standing to certify, refuse, or stand behind an output — backed by licensure, E&O, or accountability surface.
🚩
land-grab
speed of capturing an AI-native category before the window closes; head start converts directly into the other three.
Click Run simulation to fight quarter by quarter.
DEFENDER phase —
Selected for deployment
Budget
click a play below to deploy
Available plays click to add
— pick defender to preview —
Share trajectory
Pre-game position
Other market players
White space
Attacker capital slots —
Defender capital slots —
How v5 differs from v4

v4 let four hand-asserted Loud Claims (Flywheel above-threshold, Switching×Agentic stack, Counter-Pos × incumbent capex, Evaluator beats Generation) move share independent of strategies. They dominated 3:1.

v5 removes the LCs as share-movers. Strategies govern. Loud Claims should EMERGE from running thousands of trials and observing which patterns actually win — not be hand-coded up front.

What's still active in v5:

  • 3×3 compounding-class effectiveness: static / workflow / learning. Each strategy's share migration is multiplied by class-vs-class effectiveness (workflow eats static 1.5×; learning eats static 2.0×; learning eats workflow 1.5×).
  • AI-Native Land-Grab as a moat type: cards with Land-Grab primary/secondary get a per-quarter conversion bonus (1.4× / 1.24×) on flywheel/agentic/evaluator strategy effects through window-close + 2 quarters.
  • Switching×Agentic immunity (preserved as passive): defenders stacking switching+agentic absorb 60% less of incoming share-grabs from opponents.
  • Capital is a runway: 10% quarterly burn. Insurgents can bleed out before their next big play lands.
  • Other market players consolidate at horizon-end proportional to each side's quality scores (the long-tail share fragmented across non-player incumbents migrates to whoever has the stronger moat stack).
  • Acquisitions full-inherit: share + stats (additive cap-10) + in-flight strategies. Exit valuation = 1.5× capital_cost recorded for Monte Carlo aggregation.

Per side, three possible outcomes at horizon end:

  • Independent — organic valuation = share × seg_TAM × multiple
  • Acquired — outcome = acquisition price received (exit valuation)
  • Extinct — share ≤ 0.05% AND not acquired (commercial loss)

Run 100 or 1000 trials to see path-mix and value distributions for each side.