Collection
Learn new skills, turn them into career evidence, and protect the main development edge
Design skill acquisition around representative practice, adaptive difficulty, retrieval, transfer and feedback; use current work and job-search behavior to create career options; and protect one temporarily dominant development target without turning focus into rigidity.
Explore in any order.
What’s included
1 anti-pattern · 9 heuristics · 9 principles · 30 protocols · 1 question
50 ideas
Target the bottleneck subskill
Whole skills often fail at one narrow point.
Read the idea ProtocolUse a representative task before designing the drill
A tidy drill can train the wrong game.
Read the idea HeuristicKeep difficulty in the diagnosable zone
The useful edge is hard enough to expose a gap and clear enough to diagnose it.
Read the idea ProtocolFade help when performance earns it
Support should leave because evidence changed, not because the lesson reached page ten.
Read the idea ProtocolRestore help when errors stop teaching you anything
Struggling is useful only while the struggle has information.
Read the idea PrincipleSeparate practice performance from retention
Today's fluent attempt and next week's usable knowledge are different measurements.
Read the idea ProtocolReturn after a delay before adding another layer
New material can hide a growing backlog of fragile knowledge.
Read the idea ProtocolRetrieve the decision rule, not only the answer
Knowing the label is useful; knowing what makes it apply is closer to performance.
Read the idea ProtocolChange the surface before claiming transfer
Transfer starts when the costume changes and the rule still works.
Read the idea ProtocolUse near-neighbour cases to train discrimination
Some errors are not missing knowledge; they are missing contrast.
Read the idea HeuristicInterleave only where choosing matters
Mixing is a design choice, not a ritual.
Read the idea PrincipleBlock briefly when the pattern is still forming
Contrast helps after there is something stable enough to contrast.
Read the idea HeuristicChange one difficulty dimension at a time
Harder is informative only when you know what became harder.
Read the idea ProtocolPractise the failure mode you cannot afford to discover live
Routine success does not train recovery from the edge case.
Read the idea ProtocolGive feedback at the smallest actionable unit
'Needs improvement' names a direction; useful feedback names the next controllable change.
Read the idea PrincipleMake feedback end in another attempt
A comment becomes training when it changes work.
Read the idea ProtocolCompare before-feedback and after-feedback work
Improvement is easier to see when the evidence sits side by side.
Read the idea HeuristicTreat easy fluency as a signal to change the task
Comfort can mean competence, or it can mean the practice stopped asking anything new.
Read the idea ProtocolKeep one cold benchmark
A warm-up can hide the skill you actually need on demand.
Read the idea PrincipleTrain accuracy before adding speed when errors are still structural
Speed rehearses whatever rule is already running.
Read the idea ProtocolShrink the attempt–diagnosis–correction loop
Repeating an uncorrected error is still repetition.
Read the idea ProtocolAdd target constraints after the core decision is stable
Realism is useful when it tests the skill, not when it buries the lesson under five new demands.
Read the idea ProtocolBuild a task bank for variation, not repetition count
Fifty questions can still be one question wearing different numbers.
Read the idea Anti-patternStop using hours as the proof of expertise
Time invested is an input. Skill is an output that needs its own evidence.
Read the ideaAll 50 ideas in this collection
- Target the bottleneck subskill
- Use a representative task before designing the drill
- Keep difficulty in the diagnosable zone
- Fade help when performance earns it
- Restore help when errors stop teaching you anything
- Separate practice performance from retention
- Return after a delay before adding another layer
- Retrieve the decision rule, not only the answer
- Change the surface before claiming transfer
- Use near-neighbour cases to train discrimination
- Interleave only where choosing matters
- Block briefly when the pattern is still forming
- Change one difficulty dimension at a time
- Practise the failure mode you cannot afford to discover live
- Give feedback at the smallest actionable unit
- Make feedback end in another attempt
- Compare before-feedback and after-feedback work
- Treat easy fluency as a signal to change the task
- Keep one cold benchmark
- Train accuracy before adding speed when errors are still structural
- Shrink the attempt–diagnosis–correction loop
- Add target constraints after the core decision is stable
- Build a task bank for variation, not repetition count
- Stop using hours as the proof of expertise
- Turn the current role into a skill laboratory
- Ask for one stretch responsibility, not a vague bigger role
- Increase challenge and resources together
- Remove one hindering demand before adding another learning load
- Run a low-cost career experiment before a full pivot
- Practise job search before it becomes an emergency
- Train search skill and search motivation together
- Rehearse self-presentation against a real role requirement
- Turn application outcomes into a search dataset
- Explore careers with a decision question
- Keep one option-opening move in the career plan
- Use task redesign to practise the next capability
- Seek decision complexity, not just more task volume
- Check whether learning is actually the career bottleneck
- Name one dominant development edge
- State what loses priority when the dominant goal wins
- Attach the focus block to an observable cue
- Pre-plan the most common derailment
- Rehearse the cue–response plan once
- Protect the dominant edge from attractive adjacent work
- Keep a return list instead of switching now
- Give secondary skills a maintenance floor
- Review the dominant focus on a trigger, not on every mood
- Switch when the evidence changes, not when novelty appears
- Measure dominance by protected quality attempts, not hours
- End a focus cycle with a transfer test and a new allocation decision