Conway's Game of Life is normally single-color, with no notion of competing factions. Naïvely adding color by parent-inheritance (the classic Immigration / QuadLife variants) gives dynamics bit-for-bit identical to monochrome Life, the color is cosmetic and nothing actually competes. MultiLife simulates genuinely competitive multi-team automata where territory is contested and flips, lets you compare it against plain single-color Conway, and reports a quantitative endgame verdict. It runs fully offline, no dependencies, no network, no persistence.

• Faithfully simulate two rule modes: single-color Conway (Classic) and a competitive multi-team variant (Conquest), on a toroidal grid.
• Next-state is deterministic and order-independent via double-buffering; Classic is exactly standard Conway; in Conquest contested cells visibly flip allegiance along fronts.
• On stabilization, auto-halt and present a correct verdict: still lifes, oscillators, and extinction detected and labeled, with cycle-start generation, cycle length, winner, and per-team ranking.

• As a user studying competitive automata, I want to watch 2-4 colored teams contest a grid where cells defect to stronger neighbors, so I can see fronts form, advance, and collapse.
• I want a live per-team readout (count, % territory, peak) and a territory-over-time graph, so I can track the battle and watch a team get wiped out.
• I want an automatic endgame summary when the board stabilizes, so I learn the winner, stabilization generation, cycle length, and full ranking.
• I want a Classic single-color mode, so I can study plain Conway with the same analytics.

• Single self-contained index.html: no dependencies, no network, no build step, no persistence; vanilla JS + Canvas 2D.
• 120×80 toroidal grid; cell holds a team id (0=dead, 1-4=team); 8 Moore neighbors; double-buffered step() reads a snapshot then swaps.
• Shared rules: survival at 2-3 neighbors; birth at exactly 3 (plurality team); QuadLife missing-color fallback for a 1/1/1 tie at teams=4.
• Conquest: a surviving cell defects to a single enemy team that strictly outnumbers its own; ties never convert. Classic forces teams=1.
• Render to a 480×320 pixelated canvas; fresh (born/converted) cells at alpha 1.0, survivors at ≈0.72.
• Controls: Run/Pause (spacebar mirrors), Reset, Mode, Teams 2/3/4, Speed 1-60 (default 18); reseed ~16% live, uniform random team.
• Per-team readout bars + generation counter; territory line graph (~940-gen ring buffer, extinct teams as flat-zero lines).
• Stability via full-board double-hash; extinction a distinct stable state; endgame card with verdict, winner, summary, ranked table, Run-again.
• Ranking: generations survived (desc), then final territory (desc), then peak (desc).

• All teams reach 0, treated as extinction (stable); rank by generations survived, top team "lasted longest".
• 3 distinct-team parents at teams=4 apply the QuadLife missing-color rule.
• Chaotic / long-period runs never stabilize: keep running, hash history bounded (~20,000 states) to cap memory.
• Mode or team-count changed mid-run reseeds and resets all stats.

• Performance: sustain up to 60 generations/second on 120×80 using typed arrays, capping ≤4 sub-steps per animation frame.
• Reliability: hash history cleared past ~20,000 states, detection restarts; non-stabilizing runs never error.
• Offline / privacy: no network request, no external resource, no logging or persistence.
• Aesthetic: tactical-console palette; fixed team colors (1 Amber, 2 Teal, 3 Rose, 4 Violet).