---
name: deliberate-practice
description: Use when asked to design, audit, or fix practice so it actually builds skill — a goal just past current ability, full concentration, immediate specific feedback, and repetition with correction — or when someone has practiced for a long time without improving. Distinct from [[getting-feedback]] (how to receive feedback well), [[always-improving]] (the general habit of continuous improvement), and [[smart-criteria]] (how to word a goal), each of which it draws on.
---

# Deliberate Practice

Deliberate practice is practice with four specific features that ordinary
effort lacks. Hours of effort alone do not build skill; effort that has
all four features does. This skill audits the practice a user does now,
then turns a skill they want to improve into a concrete practice design
and session plan.

## The four features

1. **A clear goal just past current ability.** Not something already
   doable, and not something impossibly far ahead. Aim for the edge
   where you fail about half the time if you focus completely.
2. **Full concentration without interruption.** Not background effort,
   multitasking, or partial attention. The brain reorganizes itself
   under conditions of focused struggle; divided attention produces no
   restructuring.
3. **Immediate, specific feedback.** It tells you exactly which piece
   failed, soon enough to correct the next attempt. A vague sense of
   progress or general encouragement does not count.
4. **Repetition with correction.** Not just repetition: each attempt is
   adjusted based on what the previous one revealed.

If any one is missing, the activity is ordinary practice, however hard
it feels.

## Input required

- **The skill to improve**, stated narrowly enough to practice (a
  specific piece, not a whole discipline).
- **Current level**, with evidence: what the user can do reliably,
  what they fail at, and how often.
- **Current practice**: what they do now, for how long, and with what
  distractions.
- **Time available** per session and per week, and how long the user
  can concentrate before quality drops.
- **Feedback sources** available: a coach, a recording, a test suite,
  a scoreboard, a peer, or an answer key.

If details are missing, state assumptions or ask the questions below
once, then proceed.

## Questions to ask first

Ask in **one batch of at most five questions**, skipping any already
answered:

1. Which exact skill or sub-skill do you want to improve?
2. What can you do reliably today, and where do you fail?
3. What does a typical practice session look like now (length,
   setting, distractions)?
4. How will you find out, right after each attempt, what went wrong?
5. Is there a coach, mentor, or expert whose methods or standards you
   can borrow?

## Method

1. **Audit current practice.** Score each of the four features as
   present, partial, or missing, with one line of evidence each. Most
   stalled practice fails on goal difficulty, feedback, or correction,
   so look there first.
2. **Narrow the target.** Split the skill into sub-skills and pick the
   weakest one that matters. Practice one at a time.
3. **Set the edge goal.** Write one goal that the user would fail about
   half the time with full focus. Calibrate by the observed failure
   rate (see the table below).
4. **Protect the concentration.** Define a session length the user can
   sustain at full attention, remove interruptions in advance (phone,
   notifications, open tabs), and stop when focus is gone rather than
   practicing on in a degraded state.
5. **Build the feedback loop.** Decide how each attempt will be judged,
   and how soon. Prefer feedback that names the failed piece. If
   feedback is slow, shorten the attempt so feedback arrives sooner.
6. **Run attempts with a correction step.** After each attempt, write
   down what failed and the one specific change for the next attempt.
   Then try again with that change.
7. **Review and raise the goal.** At the end of each session, and again
   weekly, note what moved. When the goal is met reliably, set a new
   edge goal just past it.

### Calibrating the goal

| Observed result over a few attempts | Meaning | Action |
|---|---|---|
| Succeeds almost every time | Goal is inside current ability | Raise it |
| Fails about half the time with full focus | At the edge | Keep it |
| Fails almost every time, cause unclear | Too far ahead | Split into a smaller step |

### Feedback when there is no coach

Feedback must still name the failed piece. Substitutes, in rough order
of usefulness:

- **A recording** replayed against a standard or an expert example.
- **An automatic check**: a test suite, a score, a timer, an answer key.
- **A rubric** with specific criteria, applied to each attempt.
- **A peer** asked to look for one named thing, not "what do you think".
- **Self-review** against written criteria, done immediately, before
  the next attempt.

### When progress stalls

Check the four features again, in this order: is the goal still at the
edge, is the feedback still specific, is each attempt still being
adjusted, and is concentration still full. A plateau usually means the
goal stopped being at the edge, or corrections stopped being recorded.

## Output format

A practice plan with these parts:

- **Audit:** the four features, each scored present, partial, or
  missing, with evidence.
- **Skill and sub-skill** being targeted.
- **Edge goal**, worded so success or failure is observable.
- **Session design:** length, setting, and interruption controls.
- **Feedback loop:** what is measured or observed, by whom or what, and
  how soon after the attempt.
- **Correction log:** a short template.
- **Review rule:** when to raise, hold, or shrink the goal.

Example correction log:

| Attempt | Result | What failed | Change for next attempt |
|---|---|---|---|
| 1 | Missed 4 of 8 | Lost place after key change | Scan ahead one bar before playing |

## Ground rules

- **Check all four features.** Name which are present and which are
  missing in the user's current practice.
- **Do not equate hours with progress.** Time spent on what is already
  easy is not deliberate practice.
- **Specific over encouraging.** Feedback that cannot change the next
  attempt is not feedback.
- **Do not invent research claims.** Describe the four features as
  given; do not add statistics, study results, or named hour counts.
- **Match the user's real constraints.** A plan that needs a coach the
  user does not have must offer a feedback substitute.

## Self-check before answering

- Did I audit the user's current practice against all four features?
- Is the goal at the edge, failing about half the time, not trivial or
  hopeless?
- Is the session short enough to keep full concentration?
- Does the feedback name what failed, and arrive in time to use on the
  next attempt?
- Does each attempt end with a recorded correction?

## Common pitfalls

- **Comfortable repetition** — replaying what is already mastered feels
  productive but changes nothing.
- **Goals too far ahead** — constant failure with no way to locate the
  cause.
- **Multitasking** — practicing with music, messages, or a show on
  produces no restructuring.
- **Vague feedback** — "that was better" gives nothing to correct.
- **Repetition without adjustment** — repeating the same attempt the
  same way, hoping for improvement.
- **Practicing the whole instead of the weak part** — run-throughs
  hide the sub-skill that is failing.
- **Overlong sessions** — continuing after focus has gone.
- **Never raising the goal** — staying at a level once it is reliable.

## Example prompt

> I want to get better at sight-reading music. Audit how I practice now,
> then design my deliberate practice: a goal just past what I can do,
> how to keep my focus, how I get immediate specific feedback, and how
> I correct each attempt.

## Learn more

- [[getting-feedback]] for receiving the feedback that drives correction.
- [[giving-feedback]] if a coach or peer will supply it.
- [[always-improving]] for the wider habit of continuous improvement.
- [[smart-criteria]] for wording the edge goal so it is measurable.