Exercise 2: Adding a Fifth Question — Possible Solution ==================================================================== THE NEW QUESTION ------------------------------ 'Who is allowed to publish an OrderPlaced event?': ['OrderService will publish'], Added to this chapter's own QUESTIONS dictionary alongside the existing four, using the same keyword-matching approach. RESULTS ACROSS THE EXPANDED FIVE-QUESTION SET ------------------------------ BAD_ADR: 1 of 5 GOOD_ADR: 5 of 5 BAD_ADR's total moved from 1-of-4 to 1-of-5 (its one answerable question, "what was decided," still counts, but the new question adds nothing it could answer). GOOD_ADR's total moved from 4-of-4 to 5-of-5 - because GOOD_ADR's own Decision section already states "OrderService will publish an OrderPlaced event," the exact phrase the new question's keyword searches for. WHY GOOD_ADR ANSWERED THE NEW QUESTION WITHOUT BEING EDITED ------------------------------ This is a genuinely useful confirmation, not a coincidence: GOOD_ADR's own Decision section was already written specifically to state WHO does WHAT ("OrderService will publish..."), not just that publishing happens in the abstract. A well-written Decision section tends to answer follow-up questions its author never explicitly anticipated, simply because it states the mechanism precisely rather than vaguely - the same property that let it answer this chapter's own original four questions in the first place. WHY BAD_ADR STAYED STUCK AT 1 ANSWERABLE QUESTION ------------------------------ BAD_ADR's own Decision line ("We will use events for order processing") never names WHICH service does the publishing - it describes the category of solution (events) without describing the actual mechanism. No new question added to the test could be answered by a sentence this vague, regardless of what that question asks. WHY THIS WORKS AS AN ANSWER ------------------------------ The new question follows this chapter's own established keyword-search format, both ADRs are re-tested against the full expanded set rather than just the new question in isolation, and the reason GOOD_ADR's score improved without any edits is traced to a specific property of its own writing (precision about the mechanism) rather than treated as a lucky coincidence.