Question: Explain why machine learning teams may choose to build another model change before reviewing mistakes. Compare the appeal of immediate implementation with the small time cost of error analysis and the possible cost of skipping it.

Sample answer: Teams often rush to implement a change because building something feels like visible progress. By contrast, reviewing errors can seem like a pause, even though it is usually a short task. For example, inspecting 90 examples at about one minute each takes only 90 minutes, which is less than two hours. That small investment can reveal whether a proposed fix is actually worth pursuing. If the team skips that review, it may spend a week or more on a change that turns out to have little value. So the real trade-off is a little manual checking now versus the risk of wasting much more engineering time later.

Key points:

• Implementing a fix can feel more productive than pausing to inspect errors.
• Reviewing a modest sample can be done quickly; 90 examples at one minute each is 90 minutes.
• Skipping the review can lead to a week or more of wasted effort.

Rubric: The answer should explain the attraction of immediate implementation, state that error review is quick and low-cost, and describe the risk of wasting substantial development time if the review is skipped.