RUBRIC.md

Judging Rubric — System Pulse

You are reviewing two anonymized implementations (A/ and B/) of the same spec. You do not know which is which. Be honest, be specific, be brief.

Score each implementation 0–10 on each axis

1. Functionality (0–10)

Does it actually run? Are the numbers correct? Does the polling honor the ≥1Hz target? Does it handle missing GPU? Does the historical view work?

2. Design quality (0–10)

Visual polish, layout, information hierarchy, density. Does it look like something a human would willingly leave on their screen?

3. Code quality (0–10)

Organization, idiomatic use of the chosen stack, maintainability, dependency hygiene. Would a senior engineer be okay inheriting this?

4. Decision quality (0–10)

Read DECISIONS.md. Were the choices defensible? Is the reasoning honest about trade-offs, or is it post-hoc justification of arbitrary picks?

5. Polish (0–10)

Edge cases, error states, install ergonomics, README clarity, "feels finished."

Required output (per implementation)

## Implementation [A or B]

Functionality:    X/10 — one-line reason
Design quality:   X/10 — one-line reason
Code quality:     X/10 — one-line reason
Decision quality: X/10 — one-line reason
Polish:           X/10 — one-line reason

Total: XX/50

Verdict (3 sentences max): ...

Would I ship this? Yes / No / With one fix: <what fix>

Then a final comparison

## Comparison

Which is stronger overall, and why (3 sentences max):
Single biggest differentiator:
What does NOT differ meaningfully between the two:

Do not speculate about which thinking level produced which. Score the artifact in front of you.