Self-scoring is how you improve between practice sessions. The rubric is a shared language for candidates and interviewers: it says what separates a 2 from a 3 and a 3 from a 4.
Score | What the answer contains | What it sounds like |
|---|---|---|
0 | No answer, or a wrong one | Silence, or a confident wrong statement |
1 | Vocabulary only | Names setup, hold, derate, corner; no mechanism |
2 | Partial mechanism | Explains the check on the circuit but no cost or evidence |
3 | Mechanism and trade-off | Explains the check, offers more than one remedy, says what each costs |
4 | Mechanism, trade-off, evidence and limits | Adds numbers, the report that closes it, and what the answer does not cover |
def score(a):
if a.get("wrong"):
return 0
if a.get("evidence") and a.get("trade_off") and a.get("mechanism"):
return 4
if a.get("trade_off") and a.get("mechanism"):
return 3
if a.get("mechanism"):
return 2
return 1 if a.get("vocabulary") else 0
answers = {
"wrong": {"wrong": True, "mechanism": True},
"vocabulary only": {"vocabulary": True},
"mechanism only": {"vocabulary": True, "mechanism": True},
"mechanism and trade-off": {"vocabulary": True, "mechanism": True, "trade_off": True},
"full": {"vocabulary": True, "mechanism": True, "trade_off": True, "evidence": True},
}
s = {k: score(v) for k, v in answers.items()}
assert s == {"wrong": 0, "vocabulary only": 1, "mechanism only": 2, "mechanism and trade-off": 3, "full": 4}
print("PASS: scores", s)Each level adds one framework step on top of the previous one: vocabulary, then mechanism, then trade-off, then evidence. Missing steps, not missing knowledge, usually cost the points: a candidate who knows the material but never offers a trade-off or names the evidence scores 2 on a question they could have answered at 4.
Question | A 4 includes |
|---|---|
Why do we need hold fixes after setup fixes? | Speed-ups erode hold slack; hold fixes use the setup slack that remains; the setup-plus-hold sum at the endpoint bounds what delay insertion can do; evidence is a before-and-after table across all scenarios. |
How does OCV change slack? | Late and early derates on launch and capture, the common-path credit, the cost on a worked path, the method options and what each misses. |
3: mechanism and trade-off are present; the evidence step is missing.