TASK 18 / HTML · V1 · Interactive simulation

Rube Goldberg Machine

鲁布·戈德堡连锁机关

2 runs · 18.A.01 · HTML

Compare works →
TASK 18 · 18.A.01 · HTMLRube Goldberg Machine
Open independent copy ↗
Rube Goldberg Machine

Preview note: 多文件页面,需保留 js/ 与 styles.css 的相对结构

ORIGINAL PROMPT

Original prompt

PROMPT/PHASE2_30_ADVANCED_CREATION_PROMPTS_BILINGUAL.md · The original prompt is kept verbatim

Source ↗
ORIGINAL PROMPT

Original prompt

Create an animated interactive Rube Goldberg machine in the browser.

The machine must contain at least eight distinct stages forming one continuous causal chain.

The user should be able to:
- start the machine
- pause it
- resume it
- reset it
- step through the simulation slowly

Each stage must visibly cause the next stage rather than merely playing on a fixed independent timer.

Make the result visually polished and satisfying to watch.
TRANSLATION

Chinese translation

在浏览器中创建一个可交互的鲁布·戈德堡连锁机关。整个装置至少包含八个彼此不同的阶段,并形成一条连续的因果链。用户应能够启动、暂停、继续、重置,并以慢速或逐步方式观察模拟。每个阶段必须由前一阶段的实际状态触发,而不是依靠相互独立的固定时间定时器播放。整体视觉应精致,并具有令人满意的连续运动感。

Expected outcome

The deliverable should be a state- or event-driven chain-reaction simulation rather than a sequence of unrelated timed animations. Each stage should visibly trigger the next through a causal transition. Pause and step controls should operate on one coherent simulation clock, and reset should fully restore the initial state.

Additional verification guidance

  • Pause at arbitrary stages and confirm later stages do not continue independently; each stage should start only after its preceding trigger condition is satisfied.
  • Use step or slow-motion mode to inspect the chain and verify at least eight distinct, visibly causal stages.
  • Reset and rerun multiple times to confirm repeatability, no stale state, no premature triggers, and no permanent deadlocks.
  • Verify that the machine remains usable, complete, and synchronized across different viewport sizes and frame rates.
THREE INTERPRETATIONS

Same task, three model interpretations

All three results use the same width and the same preview ratio; no model is marked as best.

GPT-5.6 Sol—
Not archived yetThis model has not completed this task in Archive 001.
K3—
Not archived yetThis model has not completed this task in Archive 001.
COMPARISON METRICS

Comparison metrics

Cross-model comparison must note differences in harness, budget and isolation.

MetricDeepSeek-V4.1-Flashrun-deepseek-v4-1-flash-exp-0910-task-18-r1Muse Spark 1.3run-muse-spark-1-3-xhigh-task-18-r1GPT-5.6 SolNot archived yetK3Not archived yet
Runtime59:1323:53——
API calls398Not recorded——
Tool calls401145——
Failures / retries03——
Token throughputNot recorded329.2K——
AI review83/10088/100——
RUN RECORDS

Run records

Two runs of the same task version can be compared side by side.

18.A.01 · run-deepseek-v4-1-flash-exp-0910-task-18-r1DeepSeek-V4.1-Flash● Archived
Runtime
59:13
Token throughput
Not recorded
API calls
398
View archive record →
18.B.01 · run-muse-spark-1-3-xhigh-task-18-r1Muse Spark 1.3● Archived
Runtime
23:53
Token throughput
329.2K
API calls
Not recorded
View archive record →
RELATED TASKS

Related tasks

View all tasks →