Use a repeatable method, then follow the rubric provided for the particular task. The exercises below are independent original examples, not samples from the unfinished Handshake Project Hedgehog pack.
Read the requirements
Separate the requested result, required facts, restrictions and output format. Ask what a successful answer must accomplish.
Identify the evidence
Establish what the reference actually says. Distinguish visible or supported facts from assumptions.
Check each response against the same criteria
Look at accuracy, instruction following, completeness and relevance. Add other criteria only when the task requires them. If a rubric is provided, use it.
Explain the consequential difference
A failure that changes the requested result matters more than a harmless stylistic preference. Severity and preference labels still depend on the task's definitions.
Write a concise justification
State the judgment, identify the relevant requirement, give concrete evidence and explain its effect. Do not assume confidence or length makes an answer correct.
Original worked example 1 — Reference accuracy
Reference: Cedar Library opens at 9 a.m. Monday–Friday, opens at 10 a.m. Saturday and is closed Sunday.
Task: Give exactly two bullets: one for weekdays and one for the weekend.
Response A
Weekdays: opens at 9 a.m. Weekend: opens at 10 a.m. Saturday; closed Sunday.
Response B
Opens at 9 a.m. every day. Free parking is available.
Show the explanation
A meets the two-bullet format and preserves the relevant schedule. B incorrectly extends the weekday time to the weekend, contradicts the Sunday closure and adds an unsupported parking claim. The important difference is the accuracy of the information requested.
Original worked example 2 — Importance of an error
Reference: An order has four notebooks at $5 each, before tax.
Task: State the total before tax in one sentence.
Response A
The total before tax is $20.
Response B
The total before tax is $25.
Show the explanation
A gives the correct requested result. B's arithmetic changes the main answer, so the difference is substantive. A stylistic preference would not repair the wrong total. In a real task, use the supplied rubric to assign the exact severity or preference label.
These examples teach a method. They do not predict a platform's assessment questions or passing score.