Decision log and open questions

Record every choice with the alternative it beat, and every question nobody could answer, with who owns it, by when, and what happens meanwhile.

From the first day to the last. Keep it open while you write every other document. Feeds: every document that relies on a choice, and the change requests.

• Log a decision the day it is made, with the alternative you rejected and why.
• Log a question the day you find you cannot answer it.
• Give each question an owner, a due date and a fallback that holds until the answer comes.
• Link what each decision and each question affects, by ID.
• Say how you would get each answer, and what each possible answer would change.
• Close a question by logging its answer as a new decision, never by editing an old one.

1. What did you reject, and why? Answered in Decisions sheet, columns "Rejected alternative" and "Why".
2. What happens if a question is never answered? Answered in Open questions sheet, column "Fallback": each one lets the process run meanwhile, for example DEC-12 for OQ-02. Where a fallback carries a risk of its own, the risk register logs it: RSK-02 for DEC-12, and RSK-11 for keeping every record while OQ-04 is open.
3. Who owns each question, and by when? Answered in Columns "Owner (STK)" and "Due": a role for each, and a date in October 2026.
4. Which decisions change if an answer comes back? Answered in Column "What changes with each answer", and "Affects" on both sheets.
5. How would you get each answer? Answered in Column "How to ask (technique)". No interview took place, so none is reported.

• Decisions with no alternative. You can spot one when the "Rejected alternative" cell is empty or says "none", so nobody can tell what was weighed.
• Questions with no owner or date: the same question has been open since the first week, and nobody is chasing it.
• Questions you could answer by reading: the answer sits in a source you already cite.
• Decisions and questions with no link to what they affect: an answer arrives, and nobody knows which documents to change.
• Decisions rewritten after the event: a decision's wording no longer matches the documents that were issued under it.

• Dates never go down as the IDs go up. Each decision was logged on the day it was made, from DEC-01 on Monday 3 August 2026 to DEC-16 on Tuesday 29 September 2026. Each question was logged on the day it was found. DEC-15, logged on Friday 18 September 2026, was issued in version 1.1 on Tuesday 22 September 2026, together with the other registers that CR-01 changed. Rejected: grouping decisions by topic under renumbered IDs, which would hide when each was taken.
• No interview took place, so the two elicitation columns say how each question would be asked and what each answer would change, not what anyone said. DEC-01, DEC-04, DEC-05, DEC-06 and DEC-11 record how this fictional case was set up, so that every MINE tag has a decision to point to. In your own log, these would be facts or assumptions in the brief, confirmed by the sponsor. Rejected: invented interview notes, which the files promise not to contain, and premises of the case left untagged.
• Decisions that choose between two readings of a clause or of guidance name the legal adviser, STK-10, in the Consulted column. The sponsor, STK-06, approves; the adviser is asked first, and the RACI in document 16 has a row for it. Rejected: the sponsor alone deciding how a clause is read, with no one who can speak for the law on record.
• Every open question has a fallback, so none of them blocks the baseline. OQ-02 and OQ-08 have decisions as fallbacks (DEC-12, DEC-10), so the fallback is logged and can be argued with. Rejected: holding the baseline until every question is answered, which would wait on people outside the analysis.