game-design
Pages tagged game-design — mechanism analysis and structural evolution of games (turn structure,
verb viability, capital allocation) read as design specimens.
- L’Empereur — the two-tier turn (NES→SNES design hinge)
- Breaking Down an SNES Cart — the teardown method — the platform layer under future SNES design specimens; the architecture that shapes what games could do
- KOEI’s portable VM (SNES ROTK2 reversal) — first KOEI SNES title reversed; byte-for-byte portable bytecode VM, faithful port + facelift (NES↔SNES comparison)
- Gemfire (SNES) fully decompiled
- Nobunaga’s Ambition (SNES) — the VM exception — same game, two compile targets (NES bytecode VM vs SNES native); the execution layer got simpler on the later console
- NA1 NES↔SNES — grading two blind reverse-engineerings — SNES-derived vs NES-derived, all 5 sections converge (record bytes, Grow formula, event cadence, weakest-neighbour AI, 8-stat combat table); create-then-check method
- Battlezone (1980) — 3D Without a Multiply Instruction — hardware constraint as design: cull-before-transform pipeline reordering, a loose 90° frustum test that is correct because it must be conservative, and errors budgeted into the art rather than fixed in code
- Arithmetic Scarcity and the 3D Problem — Doom’s constraint list reads as a design document, not an optimization: what the level editor refuses to build is what makes the renderer tractable; the strategy choice as a dominance frontier over silicon / cycles / design freedom / art budget
- Stellar 7 (1983) — The Same Game Without the Coprocessor — the mesh format is the optimization: polar vertices make rotation an
ADC, and the shared-vertical-edge shortcut biases every shape in the game toward boxes — data co-designed with code, visible in the art
- Risk — The Attrition Constant — a Risk battle solved exactly as an absorbing Markov chain; the 3-vs-2 dice cap fixes the engagement frontage regardless of stack size, so attrition is Lanchester-LINEAR and concentration of force buys nothing
- Hangman — Solving Both Sides — solving both sides yields a rules verdict: under near-optimal play the fair miss allowance is 1–8 by word length — every number except 6, the one everybody plays