Break a complex skill into a useful component
When a whole task is too broad to improve efficiently, isolate a meaningful component: one problem type, one paragraph structure, one movement sequence, one software function, one pronunciation contrast, or one decision pattern.
Practice with a criterion
Know what 'better' means for this attempt: accuracy, clarity, required steps, speed only where appropriate, form, completeness, or another relevant criterion. If the criterion is unclear, feedback becomes vague.
Write the error, not only the score
Record what you did, what the expected result was, why your earlier approach seemed reasonable, and the corrected rule or example. A score may summarize a result, but an error log can explain what to practice next.
Difficulty must remain proportionate
The source sometimes speaks strongly about struggle and optimal difficulty. DEPTH-01 does not assume that more struggle is always better. Useful challenge depends on task, safety, prerequisite knowledge, accessibility, fatigue, stress and feedback availability.
Feedback source matters
Use appropriate feedback for the domain. A teacher, answer key, compiler/test suite, supervisor, coach, source text, qualified professional, or measured outcome may each be useful in different contexts. AI feedback should not be treated as automatically correct.
Reflection
Questions to sit with
- What exact component am I practicing?
- What criterion tells me whether the attempt improved?
- What did the error teach me about the next attempt?