Evidence workbook
Start with one decision you can explain and one canvas that fits it. The twelve techniques below are options, not a required sequence. Copy the prompts into your own notes and leave an honest gap wherever support is missing.
Start with a common record
Use these fields with every canvas:
| Field | Your notes |
|---|---|
| Project and context | What was being built, when, and under which constraints? |
| Your role | What did you own? What did collaborators own? |
| Claim | What specific statement should a reader assess? |
| Source | Artifact, date/revision, relevant observation, and access conditions. |
| Evidence status | Recorded observation, recollection, inference, or proposal. |
| Limits | Missing evidence, uncertainty, alternative explanations. |
| Disclosure | What can be shared with this audience? |
| Next check | What would most improve confidence in the account? |
If you start from memory, label the draft accordingly. Return to available source material before treating its details as verified. Do not invent measurements to complete a field.
01. Provenance chain
Trace one feature through the failures and revisions that shaped it.
- Current feature and project: …
- Earlier failure or constraint: …
- Decision or correction at each step: …
- Your role at each step: …
- Evidence for each link, including gaps: …
Do not turn a remembered sequence into a documented history. Mark links that still depend on recollection.
02. Structural rhyme
Compare a mechanism you know with a mechanism in another domain.
- First domain and observed mechanism: …
- Second domain and its supporting source: …
- Shared constraint or failure mechanism: …
- Conditions required for the comparison: …
- Where the analogy breaks: …
Present the comparison as an analogy unless you can substantiate a stronger equivalence. Do not substitute a familiar metaphor for evidence.
03. Forensic number
Put a consequential measurement in its actual context.
- Quantity, value, units, and uncertainty: …
- Instrument or source record and date: …
- Test conditions and sample: …
- Baseline or comparison conditions: …
- Decision the measurement informed: …
Use recorded precision. Label an estimate or remembered value. A decimal place does not make a claim more reliable.
04. Compressed war story
Draft one sentence that identifies a problem, decision, and supported result.
- The essential sentence: …
- The constraint behind the decision: …
- Your contribution and collaborators: …
- Evidence available for the result: …
- What the short version leaves out: …
Keep a longer account behind the sentence. Check that compression has not removed a qualification or turned a team outcome into individual credit.
05. Constraint personality
Use a memorable description to explain a constraint you encountered.
- Constraint and plain technical description: …
- Behavior that surprised you: …
- Event or observation behind the description: …
- Metaphor you would use: …
- Where the metaphor stops being useful: …
A metaphor is a teaching aid. Pair it with the actual mechanism and conditions so the reader does not mistake personality for a physical explanation.
06. Decision heuristic
Explain a rule you use, where it came from, and when it fails.
- Rule in one sentence: …
- Conditions where it applies: …
- Origin event or supporting observation: …
- Known exception or failure case: …
- How you would check it on a new project: …
State the exception as carefully as the rule. One successful project does not establish a universal practice.
07. Counterfactual
Describe a risk you identified and the action taken to reduce it.
- Observed condition: …
- Predicted failure mode: …
- Evidence or model supporting the prediction: …
- Action and your role: …
- What was observed afterward and what remains hypothetical: …
Distinguish observed outcomes from avoided-outcome estimates. Do not present an unobserved catastrophe or savings figure as a measured result.
08. Threshold vocabulary
Explain a term used to recognize a changing system state.
- Term and plain-language meaning: …
- Observable condition it describes: …
- Threshold or criteria, if defined: …
- Event that led to its use: …
- Limits and possible misunderstanding: …
Define local jargon for an outside reader. Attribute a borrowed term; do not imply an established industry definition where there is none.
09. Tool as character
Explain why a particular tool helped you make a consequential decision.
- Tool and task: …
- Constraints that shaped the choice: …
- Alternative considered: …
- Assumptions you could inspect: …
- Outcome and the tool’s limitations: …
Describe a choice in context. Tool preference is not evidence that the tool is generally superior.
10. Negative-space expertise
Document a practice you avoid and the experience behind that boundary.
- Practice you avoid: …
- Specific conditions that make it risky: …
- Origin event or supporting source: …
- Exception where the practice may be reasonable: …
- Alternative you use and how you check it: …
Replace a blanket prohibition with a scoped lesson. Repeat for up to three practices if the evidence warrants it.
11. Temporal signature
Explain why you address a problem at a particular project stage.
- Problem: …
- Stage where you intervene: …
- Previous timing and observed consequences: …
- Why the earlier or later timing helps: …
- Evidence and costs of the timing decision: …
Compare with a documented project sequence. Avoid unsupported claims about what most engineers do.
12. Mentor echo
Describe advice that changed how you make a decision.
- Advice, marked as quotation or paraphrase: …
- Attribution, where appropriate to share: …
- Context in which you received it: …
- Decision where it proved useful: …
- Limits or later revision of the advice: …
Mark remembered wording as a paraphrase unless verified. Preserve attribution and sharing permissions.
Review before sharing
Read the draft against the Source Standard and Depth Standard. Check the material claims, individual credit, conditions, and disclosure boundary. A missing source is a reason to qualify or remove a claim, not to embellish it.
Keep the working record and its revision history. Share the smallest account that supports the intended discussion, with clear limits and a path to further review where appropriate.