Module 13: Exam Strategy and Practice

Lesson 3: Recognizing Trick Questions

The CCA-F (and almost all professional certifications) use Distractors—answers that are "Correct in theory" but "Wrong in this context." If you pick these answers, you fail. To win, you must understand the "Tricks" used to separate an Amateur from an Architect.

In this lesson, we look at the three most common trick archetypes in the CCA-F.

---

1. The "Over-Engineering" Trap

An answer suggests a very complex, expensive solution that is not necessary for the specific goal.

• Goal: "Route a user to a department."
• Trick Answer: "Build a MAS with a Planner-Executor and a Supervisor and verify outputs with Opus."
• Real Answer: "Use Claude Haiku with a classification system prompt."

The Rule: Choose the simplest, cheapest solution that meets the requirements. If the requirements don't mention security, don't pay for security.

---

2. The "Model Confusion" Trick

Answers that swap the capabilities of Haiku, Sonnet, and Opus.

• Trick Answer: "Use Haiku for complex codebase-wide refactoring because it is the fastest."
• Reality: Haiku will hallucinate. You need Sonnet for refactoring.

The Rule: If a question requires "Reasoning Depth" or "Long Context Following," the answer is Sonnet.

---

3. The "Double Negative" Logic

The question uses language like: "Which of the following does NOT result in an UNRELIABLE output?"

• This is a trick to test your reading comprehension under stress.
• Translation: "Which of the following is Reliable?"

The Rule: Literally write the translation down on your scratchpad if you are confused. Never answer a "Not" question until you convert it to a "Positive" goal.

---

4. Visualizing the Trick Filter

graph TD
    Q[Question] --> T1{Is it too complex?}
    T1 -->|Yes| D[Distractor]
    T1 -->|No| T2{Is model correct?}
    T2 -->|No| D
    T2 -->|Yes| T3{Is logic positive?}
    T3 -->|No| D
    T3 -->|Yes| A[Verified Answer]

---

5. Summary

• Simplicity Wins: Avoid over-engineering unless requested.
• Model Mastery: Know when intelligence is required vs. speed.
• Logic Decoding: Simplify "Negative" questions before answering.

In the final lesson of this module, we look at the most productive part of practice: Reviewing Incorrect Answers Effectively.

---

Interactive Quiz

1. Why is a complex Multi-Agent System often a "Distractor" answer for simple routing tasks?
2. How do you decode a "Double Negative" question?
3. In what scenario would you choose Claude Opus over Claude Sonnet during the exam? (Hint: See Module 12).
4. Scenario: A question asks for the "Most Cost-Effective" way to summarize 10,000 documents. Option A is "Parallel Sonnet calls." Option B is "Sequential Haiku calls with Prompt Caching." Which is the trick and why?

---

Reference Video: