- The “Desirable Difficulty” test: is your struggle productive?
- Four upgrades that instantly make practice less safe—and more effective
- Time to build a 35min “unsafe” practice session
- Common mistakes when people try to “make practice harder”
- How to know you’re improving (simple measurement system)
- Quick swaps: from “safe routine” to “growth routine”
- FAQ
- Bottom line
If you don’t make mistakes in practice, you’re mostly rehearsing—time to learn something. Trade your “safe block reps” for blocks of deliberate practice: one small target at a time, clear feedback after each rep, and do reps that feel a little unstable! Build difficulty by spacing them out, mixing things up (interleaving), and quick self-tests (retrieval)—so performance in your practice session will look worse, but your learning will get better. Check yourself one day later (did you retain this?) and two weeks later (can you do this in a new context? Not just one you are usually successful in today)—not just how good it felt to do it today.
A comfortable routine is great! I know what is coming, I will probably get that rep, I will leave here feeling competent. That familiarity and reliability is the problem! When your practice is trained to feel comfortable it may also be trained to repeat an existing level of performance; not create a new level of performance.
This article is about how to turn your practice into a learning machine. The first step is to diagnose your safe routine. Next, you will redesign your practice sessions using proven techniques from research (deliberate practice, spacing, interleaving, retrieval). Most importantly figuring out how to show yourself that you are learning.
So what does the comfortable, predictable, “safe routine” really look like? First: You do the same warm up, in the same order, the same tempo/intensity, in the same comfort zone. You dance around the parts you missed (or touch them, and go back to what feels good). You measure your sessions by feel, (not by how much material Harry Potter you got through today. You can do it perfect in practice—until you need to do it different speeds, different contexts, different pressure, different order.
Safe practice often boosts what psychologists call performance in acquisition—how good you appear while doing the drills—but can fail to boost retention and transfer (what you can still do later, and in new situations). That’s part of why random/interleaved practice can feel worse in the moment but yield better long-term outcomes. ( Source)
The fundamental divide: practicing for performance vs practicing for learning
Performance is what you can do right now in familiar conditions. Learning is a change that sticks—something you can reproduce tomorrow, next week, and under slightly different conditions.
Deliberate practice (in the Ericsson etc. research tradition) is not “just doing the thing a lot.” It’s working on a specific weakness, at the edge of current skill, with adjustments. Comfortably repetition may be all that’s needed for warmup, for maintenance, but it’s rarely enough for growth. ( Source)
Why “safe” practice seems productive (even when it isn’t)
- Fluency illusion: when it’s smooth, we think we’ve learned it.
- Short feedback loop: because the drill’s so similar, it’s just “auto-correcting” from round-back to round-back without really reconstructing it in memory.
- Low decision-making: you’re never choosing the right move/solution—you already know what’s next in the routine.
- Low stakes: If you never test yourself (from memory, under time pressure, in mixed conditions), you never see the real gap.
The research on “test-enhanced learning” (retrieval practice) says that actively recalling information can boost long-term retention more than even more studying—despite feeling harder, and producing more errors in practice. That mismatch between “feels good” and “works well” is precisely why safe routines may be quietly stuck. (journals.sagepub.com)
The “Desirable Difficulty” test: is your struggle productive?
Not all difficulty is helpful. You want the kind that produces deeper processing, better discrimination, and stronger recall—without devolving into random chaos. Good practice difficulties tend to yield worse “practice” performance while improving retention and transfer. Interleaving and spacing are classic examples. (journals.sagepub.com)
| If you notice… | It usually means… | Adjustment to make it productive |
|---|---|---|
| You miss sometimes, but you can explain why | Good signal: your error is specific | Keep the challenge; add a tiny constraint (tempo, time limit, variation) |
| You miss randomly and can’t diagnose the cause | Too hard / too many variables | Reduce degrees of freedom (slow down, isolate a component, add clearer feedback) |
| You never miss, even on “hard days” | Too easy / auto-pilot | Raise difficulty (interleave, change context, increase speed, remove prompts) |
| You improve within the session but lose it tomorrow | Likely blocked/massed practice illusion | Add spacing + short retrieval tests + mixed reps |
Four upgrades that instantly make practice less safe—and more effective
1) Replace “time goals” with “target + feedback”
A safe routine is time-based (“practice for 60 minutes”). A growth routine is target-based: one micro-skill, one measurable standard, one feedback method.
- Musician: “Clean shift from position 3→5 at 84 bpm with no string noise (recorded).”
- Athlete: “10 out of 12 free throws after a 20-second heart-rate spike.”
- Engineer: “Solve 3 array problems without looking up the pattern; write a post-mortem for mistakes.”
- Language learner: “Retell a story for 2 minutes using past tense accurately (no notes).”
This lines up with deliberate practice: specificity, feedback. Intentional adjustment—rather than mindless repetition. (andymatuschak.org).
2) Interleave instead of block (even if it hurts your ego)
Blocked practice (AAAA, then BBBB) feels safe because you keep doing the same thing. Interleaving (ABACBC…) forces your brain to select and reconstruct the right action each time. In motor learning research, this is closely related to the contextual interference effect: random practice often looks worse during practice but can improve retention and transfer. (tandfonline.com):
- Pick 2–4 related variations (not 12). Example: three chord changes, three serve targets, three bug patterns.
- Do short sets (e.g., 3–6 reps) of Variation A, then switch—before you feel “warm.”
- Add a decision cue: roll a die, use a randomizer app, or pre-write a mixed list.
- Track outcomes per variation so you don’t lie to yourself with “overall it felt fine.”
3) Space your repetitions (stop cramming the same problem)
Massed practice (cramming the same thing back-to-back) is the definition of safe: the last rep teaches the next rep. Spacing makes you retrieve and rebuild the skill once forgetting has started—that is a key ingredient in durable learning. Distributed practice effects are robust across nearly any domain of learning. (home.cs.colorado.edu)
4) Finish with a test (not a “victory lap”)
Safe routines usually end with something you’re best at, because that encompasses the warm fuzzy feeling. Growth routines finish up with a short test that creates an honest score—preferably without prompts, notes, or “one more warm-up rep.” Retrieval practice is a robust way to reinforce a memory and make accessing a skill easier over time. (journals.sagepub.com)
- Music: Report progress on your one cold take at the target tempo and grade it 0-2 (for time, tone, accuracy).
- Sports: Report progress on 10 non-rehearsed game-like reps (eg fatigue, time constraint, random defender, smaller target).
- Academics: Test progress by closing the book and writing what you recall for 3min, check it and repeat.
- Coding: Test progress by re-solving a smaller version from scratch, then compare to earlier solutions.
Now that we summarize the ideas, here’s the only 35 min practice session template we’ll ever need (copy/paste).
Time to build a 35min “unsafe” practice session
- 2min — Define Target: (one micro-skill)+(one metric) – (accuracy, tempo, number of helps, other quality)
- 6min — Diagnose: 2-3 testing reps to find your point of failure; write out the failure “hypothesis” in one sentence (“I miss because…”).
- 12min — Deliberate reps: 3 “mini-sets” of 4-reps with rapid feedback (be sure to use video or a metronome or a unit-test or advice from a coach based on that specific failure). Only adjust 1 variable every batch. (andymatuschak.org).
- 10 min — Interleave: rotate 2–4 variations (short sets). Expect practice performance to dip. (tandfonline.com)
- 3 min — Retrieval test: one cold attempt (no prompts). Record score.
- 2 min — Log + schedule: write what worked, what didn’t, and when you’ll retest (tomorrow / 48 hours). Use spacing. (home.cs.colorado.edu)
Common mistakes when people try to “make practice harder”
- They add difficulty but remove feedback. Hard with feedback = learning, hard without feedback = randomness.
- They change 5 variables at once (so you never learn what caused improvement / failure).
- They interleave too early, with too many variations (start with 2-4 variations not 10)
- They chase lack of comfort (i.e. training fatigue) instead of skill (you want to feel adaptation not exhausted muscles).
- They never run delayed check (not running a test tomorrow means you’re just measuring today… kind of.)
How to know you’re improving (simple measurement system)
If you over-rely on how practice feels today, you’ll bias toward safe routines. Instead, measure retention (does it stick?) and transfer (does it travel?); the best techniques for durable learning may differ from the techniques that feel the best in practice, research suggests. (journals.sagepub.com)
| Check | When to do it | What to do | What “good” looks like |
|---|---|---|---|
| Retention check | Next day (or 48 hours) | 1–3 cold reps, same standard as practice | Slight improvement or stable performance with less effort |
| Transfer check | Weekly | Change one condition (speed, order, context, problem type, pressure) | Performance holds up under change; fewer “falls apart” moments |
Quick swaps: from “safe routine” to “growth routine”
| Safe habit | Why it stalls | Growth swap (do this instead) |
|---|---|---|
| Always start with your favorite pieces/drills | You peak early and avoid weaknesses | Start with 5 minutes on your current bottleneck (fresh focus) |
| Repeat the same drill until it feels easy | You’re training short-term adjustment | Stop earlier; space it; revisit later with a cold test |
| Practice in perfect conditions only | Skill becomes context-dependent | Add one constraint: different tempo, surface, environment, or prompt removal |
| Measure success by “minutes practiced” | Time ≠ progress | Measure by scores on retention/transfer checks |
FAQ
Q: This mean I should practice mistakes on purpose?
Q: If interleaving and random practice suck, how do I keep my motivation up enough to practice through the pause I’m creating?
Q: How much of my practice should be unsafe?
Q: What if I don’t have a coach to give me feedback?
Q: Is spaced practice always better than cramming?
Bottom line
If your practice routine feels “safe” consistently (and given that you’re not enrolled in a boot camp or training for the amateur Olympics), you’re probably optimizing for comfort and fluency, not skill. Harsh, well-designed discomfort contains skill-getting features like specific targets, tight feedback loops, spaced and interleaved reps, and tiny tests that show you what you really changed. Do that instead (and your practice may feel worse) and your skills will travel farther.