Why Mental Math Speed Is Overrated — and What Accuracy-First Training Looks Like

Every math teacher has met this parent, and honestly, I've been this parent. You watch your eight-year-old work through a page of subtraction, and it's slow. Painfully slow. Fingers come out. Lips move. A single problem takes fifteen seconds while the kid next door apparently rattles off times tables like a vending machine. And a quiet panic sets in: my child is behind.

I run a math academy in Korea, a country not exactly famous for relaxed attitudes toward arithmetic. I've watched thousands of elementary students come through the door, and I want to tell you something that took me years to fully believe: speed is one of the least useful things to train directly, and one of the most reliable things to get for free if you train something else first.

That something else is accuracy.

Where the Speed Obsession Comes From

Speed is seductive because it's visible. You can time it. You can compare it. You can put it on a chart and watch the line go up. Accuracy at a low speed looks like nothing much is happening, while a fast kid looks like a genius at a dinner party.

There's also a historical reason. Before calculators were in every pocket, arithmetic fluency was a job skill. Shopkeepers, bookkeepers, and engineers computed by hand all day, and being fast genuinely paid. That world is gone, but the curriculum culture it created — the timed drill, the "mad minute," the flashcard race — stuck around long after its original purpose expired.

And finally, speed is often a real correlate of understanding. Strong math students usually are fast. Parents observe this and reverse the arrow: if fast students are strong, then making my child fast will make them strong. But correlation runs the other direction here. Speed is downstream. It's the exhaust, not the engine.

What Speed Actually Measures

When a child answers 7 × 8 instantly, one of two very different things is happening. Either they've retrieved the fact from memory as a single stored unit, or they've computed it quickly through some route — doubling 7 × 4, or subtracting 7 from 63, or something they invented themselves.

Both are fine. What's not fine is the third possibility: they've memorized a sound without meaning. "Seven-eights-fifty-six" is a syllable pattern, and plenty of children can chant it flawlessly while having no idea that it describes eight groups of seven, or the area of a 7-by-8 rectangle, or why 7 × 80 must therefore be 560. That child looks fast on the flashcard and falls apart in fourth grade when the problems start requiring the meaning rather than the sound.

Here's the practical test. Ask your child a fact they know, then ask a follow-up: "So what's 7 × 9?" A child with real structure will often say "63 — it's seven more." A child with only sound will restart the chant from the beginning, or guess. The first child is going to be fine even if they're slow today. The second one is fast and fragile.

The Hidden Cost of Racing

Pushing speed before accuracy isn't just neutral. It actively creates three problems.

It rewards guessing. Under time pressure, a rational child learns that producing an answer beats producing the right answer, because blank space definitely loses. That habit is extraordinarily hard to unlearn, and it shows up years later as a student who writes down a plausible-looking number and moves on without ever checking whether it makes sense.

It burns working memory. Working memory — the mental desk space where you hold a problem while you work on it — is limited, and anxiety eats it. This is well documented in the research on math anxiety: when a child feels threatened by a timed test, some of the capacity they'd normally use for computing gets consumed by worrying about the clock. So the child performs worse, concludes they're bad at math, and gets more anxious next time. It's a genuinely vicious loop, and timed testing is one of the most efficient ways to start it.

It hides the actual problem. If a child is slow at 14 − 8, speed drilling treats the slowness. But the slowness is a symptom. The real issue might be that they don't have a solid grip on making ten, or that they're counting backwards one at a time because nobody ever showed them a better route. Drilling faster just makes them count backwards faster, which has a hard ceiling and teaches them nothing.

What Accuracy-First Training Actually Looks Like

Accuracy-first doesn't mean slow forever, and it definitely doesn't mean easy. It means the standard for a practice session is "every answer correct and explainable," not "as many as possible in three minutes." Here's the concrete swap.

Instead of this	Try this	Why it works better
"Do 40 problems in 3 minutes"	"Do 10 problems, all correct, no time limit"	Sets correctness as the goal, so the child slows down where they're unsure instead of guessing
Marking the page and moving on	Marking the page and reworking every miss	The error is the lesson; skipping it means practising the same mistake tomorrow
Flashcards on every fact equally	Flashcards only on the 5-10 facts they actually miss	Most kids know most facts; drilling the known ones wastes the session's energy
"Faster! Come on!"	"How did you get that?"	Reveals whether the answer came from structure or from luck, and builds explaining as a habit
A long session on Sunday	10-15 focused minutes most days	Short, frequent, low-stakes practice suits how memory consolidates and keeps frustration low

The single most important line in that table is the second one. Reworking errors is where almost all the learning lives, and it's the step families skip most often because it feels like punishment. Reframe it out loud: "Great, we found one to fix. That's the whole point of practising." Children take their emotional cues about mistakes directly from your face, so make your face bored and cheerful rather than disappointed.

One more piece of practical advice: keep a small, ugly notebook of the specific facts and problem types your child gets wrong. Not a beautiful chart — a running list. After three weeks you'll see patterns you'd never notice otherwise. It's almost never "multiplication." It's "anything crossing a ten," or "he's fine until there's a zero in the middle."

How You'll Know It's Working

Because you're not timing anything, you need different signals. Watch for these:

• Fewer restarts. The child stops going back to the beginning of a sequence to find an answer.
• Self-correction. They say a wrong answer, pause, and fix it themselves. This is a huge milestone — it means an internal checker has switched on.
• Estimation appears. Unprompted, they say things like "it should be around 40." That's number sense showing up.
• Transfer to new formats. They handle 8 × ? = 56 nearly as easily as 8 × 7, which memorized-sound kids typically cannot.
• And then, quietly, speed. Facts they've used correctly and repeatedly become automatic on their own timetable. You didn't train it. You made it possible.

When Speed Does Matter

I don't want to oversell this. Fluency genuinely matters, and a child who is still counting on fingers for basic facts in fifth grade is at a real disadvantage — not because of the seconds lost, but because the effort of computing crowds out the harder thinking the problem actually requires. If you're solving a multi-step word problem and half your brain is occupied with 6 + 7, you have less brain left for the problem itself.

So the goal is real automaticity. The disagreement is only about the route. Accuracy-first gets you there through correct repetition, and it gets you there with a child who still likes math. Speed-first sometimes gets you there faster, and sometimes gets you a fast, anxious guesser. Given how long the math road is — a decade or more — I'll take the slower start every time.

If it helps to have something structured to practise with, we make free printable worksheets in the Korean drill style, built around short daily sets rather than timed races: free printable math drill sheets (English). Print ten problems, do them properly, fix the misses, and stop. That's a good day.