Evidence Chain Template

This worksheet answers one concrete question: What exactly do I know has changed?

It does not require every result to improve or turn life into a score. It places samples from the same task under different conditions side by side, so you can distinguish “I can do it now”, “I still can after a few days”, and “I can use it somewhere new”.

Copy it into a private note and use one copy for each main question. Do not record passwords, identity documents, customer data, medical privacy, or unauthorised third-party information. For health, safety, legal, financial, or persistent loss-of-control concerns, contact a trusted person and a qualified professional first.

1. Define the Question

Record start date:
Real situation and audience/user:
The one question this record answers:
Acceptance standard:
Possible risks and stop conditions:

The question should be observable, such as “Can I explain a project risk in English without a script and answer one follow-up?” Do not write “Am I better now?”

2. Save an Unaided Baseline

The baseline should happen before AI, peer feedback, or concentrated training whenever possible. Record the conditions so tools, extra time, and familiar material do not disappear from the comparison.

Date	Task and constraints	Tool used?	Result/errors	Raw evidence location
		No

What I could complete independently at baseline:
Where I stopped or needed a prompt:
What the baseline cannot prove:

3. Record Immediate Performance

Immediate performance says what happened under the current conditions. It does not automatically say that ability is stable.

Date	Conditions and prompts	What I completed	Quality/time/rework	Evidence location

If AI was used, state what it did and what you completed independently after closing it. A chat log is a process trace, not proof of ability by itself.

4. Test Delayed Retention

Wait for a period, remove recent prompts and the original answer, and complete a related task. Choose the interval according to the task and your capacity, but write it down.

Retest date:
Time since the last practice:
Prompts removed:
Related task:
Independent result:
Errors that remained:
Evidence location:

5. Test Transfer

Keep the core action and change at least one condition: topic, listener, time limit, material, tool permission, or purpose.

Changed condition	New task	Independent result	How did the other person/user understand it?	Evidence location

A failed transfer does not make earlier practice worthless. It shows that the change may be too large, prerequisite knowledge is missing, or the acceptance standard needs revision.

6. Separate Fact, Interpretation, and Hypothesis

Fact: I can point to a date, condition, sample, source, or direct observation.
Interpretation: How I understand those facts; it may need revision.
Hypothesis: The judgment I will test in the next action.
Conclusion I cannot draw yet:

For example, “the unaided version finished in 90 seconds” is a record. “It was clearer because my sleep improved” is an interpretation. “It may work if I can repeat it with an unfamiliar listener next week” is a hypothesis.

7. Let Evidence Choose the Next Step

This evidence most supports: continue / downgrade / change one variable / pause or stop
The one condition I will change:
Smallest next task:
Acceptance standard for the next task:
Support, permission, or professional judgment needed:
Next review date:

When evidence conflicts, do not hide the conflict in an average. Write the different conditions first, then decide whether another sample is worth collecting. When evidence is thin, shrinking the question is usually more honest than manufacturing a precise score.

8. Handover Summary

Main question and acceptance standard:
Most reliable evidence:
Least reliable or still missing evidence:
Repeated errors/risks:
Smallest next task and stop condition:
Evidence folder or link:

When handing this summary to your future self, a peer, or AI, ask them to work only from this record. They should not pretend to remember another session or fill in facts that are not present.

Related chapters: Evidence: How Change Becomes Visible | Toolkit Overview | Rhythm Ledger | Glossary of Terms and Methods