Daily Decision Audit Log

Template Overview

Humans suffer from severe outcome bias: when a decision succeeds, we attribute it to brilliance; when it fails, we blame bad luck or forget the conditions that made the choice reasonable at the time. The Daily Decision Audit Log acts as a tactical decision journal. By documenting critical engineering, managerial, or financial decisions in the moment—along with assumptions, alternatives discarded, and predicted outcomes—it creates a feedback loop for sharpening judgment.

Markdown Template Structure

Copy the block below or click Download .md File to add this template directly to your personal vault:

---
title: "Decision Log: [Brief Decision Headline]"
date: [YYYY-MM-DD]
decision_maker: "[[Your Name / Role]]"
decision_stakes: "[High / Medium / Low]"
decision_type: "[Technical Architecture / Product Scope / Hiring / Commercial]"
review_date: [YYYY-MM-DD]
tags: [decision-journal, mental-models, rationality, risk-management, judgment]
---

# Daily Decision Audit Log — [Decision Headline]

## 1. Decision Framing & Context
- **The Core Problem / Dilemma:** Should we migrate from cloud Whisper APIs to 100% on-device Core ML transcription for all desktop users?
- **Urgency & Timing:** Decision required before Q4 software architecture freeze.
- **Mental State at Time of Choice:** Calm, well-rested, high conviction based on customer privacy interviews.

## 2. Alternatives Evaluated & Trade-Off Matrix
| Alternative Option | Key Advantages | Fatal Flaw / Unacceptable Trade-Off | Why Discarded |
|---|---|---|---|
| Option A: Cloud-Only API | Zero client hardware requirements | Massive monthly server API costs; violates customer privacy guarantees | Discarded |
| Option B: Hybrid Cloud/Edge | Fast fallback for old laptops | Complex multi-path codebase; confounds user privacy mental model | Discarded |
| Option C: Pure On-Device (Chosen) | Complete privacy; zero server bills; uncompromised offline capability | Requires Apple Silicon M-series hardware; higher app download size | **Selected** |

## 3. Assumptions & Expected Probabilities
- **Assumption 1:** Over 85% of target customers in legal, medical, and engineering fields use Apple Silicon Macs (Probability: 90%).
- **Assumption 2:** Local transcription accuracy on 4-bit Whisper models is indistinguishable from cloud endpoints for 95% of speech audio (Probability: 85%).
- **Assumption 3:** Customer satisfaction will increase due to zero network latency and privacy guarantees (Probability: 95%).

## 4. What Would Change My Mind? (Falsification Criteria)
- If customer support tickets regarding unsupported Intel hardware exceed 15% of total inbound volume within 30 days of launch.
- If thermal throttling causes system audio stutter during multi-hour recordings on M2 MacBook Air laptops.

## 5. Review Checkpoint & Outcome Assessment
- **Scheduled Review Date:** [YYYY-MM-DD] (60 days post-launch).
- **Actual Outcome (Filled at Review):** `[To be completed on review date]`
- **Did the Decision Succeed? (Y/N):** `[Pending]`
- **Lessons Learned Regarding Judgment:** `[Pending]`

Filled Real-World Example

Here is a realistic example demonstrating how this template captures actual deliberations:

---
title: "Decision Log: Rejecting Cloud User Accounts"
date: 2026-10-09
decision_maker: "[[Julian Thorne]]"
decision_stakes: "High"
decision_type: "Product Architecture"
review_date: 2026-12-09
tags: [decision-journal, mental-models, rationality, risk-management, judgment]
---

# Daily Decision Audit Log — Rejecting Cloud User Accounts

## 1. Framing
- Dilemma: Should we require users to create an online account to use Oyma?
- Decision: Absolutely not. Zero accounts, zero emails, zero tracking cookies.

## 2. Rationale
- Our core competitive wedge is sovereign privacy. Adding account logins introduces server dependencies, password reset flows, and compliance liabilities.

## 3. Review Criteria
- Review in 60 days: measure conversion rate from download to daily active note creation.

When to Use and When to Avoid

When to Use

• Use whenever making significant technical architectural bets, hiring choices, or product scope cuts.
• Use to inoculate yourself against hindsight bias and revisionist memory.
• Use to train junior leaders in disciplined risk evaluation and mental model application.

When to Avoid

• Avoid using for trivial low-stakes choices (e.g., choosing which font to use in a slide).
• Avoid using for decisions that can be easily reversed in 5 minutes with zero cost.

On-Device AI Integration

Summarize your recording in Oyma with this local instruction:

Analyze the daily decision audit log:
1. Examine the stated assumptions for cognitive biases (e.g., confirmation bias, optimism bias).
2. Stress-test the falsification criteria to ensure they are measurable and objective.
3. Formulate the single strongest counterargument against the chosen option.
Output in structured Markdown.

Your data remains securely on your Mac with zero cloud exposure.