"Left foot forward. Left foot forward. Left foot forward." A seven-year-old is issuing instructions to her father, who is playing a robot and following them literally. He walks into the kitchen doorframe, because she never said stop. She laughs, then goes quiet for a second and says she needs to count the steps first.
That is debugging. No screen was involved, the whole thing took four minutes, and the child has just met something that will still be true when she is twenty-five and writing real software: the machine does exactly what you said, not what you meant.
What coding means before the age of ten
Parents often assume early coding means a child typing something that looks like real programming. Before ten, almost none of the value is in the typing. It sits in four habits of thought, all of which show up in ordinary life long before any device does.
- Sequencing. Order matters and cannot be improvised. Socks before shoes. Butter the bread before you close the sandwich. A child who can tell you the six steps of getting ready for school, in the right order, is doing the thing.
- Pattern recognition. Noticing that something repeats, and that the repeat can be described once instead of listed twenty times. Red, blue, red, blue becomes swap colours each time.
- Cause and effect. If I change this, that happens. Predicting the result before pressing anything, then checking.
- Breaking things down. Turning clean your room into books, then clothes, then floor. Big tasks are small tasks in a row.
Add debugging, which is closer to a temperament than a skill: when something does not work, look for the specific step that is wrong instead of concluding you are bad at it. That single habit transfers to maths, music and most other things, which is a large part of why school systems bother with this at primary age at all.
Blocks, text, and the reading tax
Block-based tools, the ones with coloured instruction tiles you drag together, exist for a specific reason. Text programming charges a tax that has nothing to do with logic: precise spelling, exact punctuation, fluent typing, and keywords that are almost always in English.
A nine-year-old who forgets a semicolon has not made a thinking mistake. She has failed a spelling test that was hiding inside a maths lesson. Blocks remove that tax and leave the reasoning intact.
| Aspect | Block-based | Text-based |
|---|---|---|
| Typing needed | Almost none | Fluent, both hands |
| Reading load | Short labels, usually translated | Full lines, English keywords |
| Typical errors | Wrong order, missing repeat | Punctuation, spelling, indentation |
| Feedback | You watch the character move | An error message you have to read |
| Usually workable from | Around 5 or 6 | Around 10 to 12 |
This matters more outside English-speaking countries than most guidance admits. A child in Jakarta or Lima working in a text language is decoding foreign words while also learning what a loop is. Block tools translate their labels; text languages do not translate their keywords. Staying with blocks longer is not falling behind, and there is no rush. A child who understands loops thoroughly in blocks tends to pick up the text version in a couple of weeks later on.
Unplugged activities build the same thing
The robot game at the top of this article is a complete lesson. So are several other things that need no device at all, which is worth knowing in a household with one shared phone, a metered data plan, or a firm view on screen time.
- Human robot. The child gives step-by-step instructions and the adult follows them with deliberate literalness. Walking into the furniture is the point.
- The sandwich algorithm. The child dictates how to make a sandwich. You do exactly as told, including putting the knife into the jam before anyone opened the jar.
- Repeating patterns with objects. Lay out beads or bottle caps in a repeating pattern, then challenge the child to describe it in one sentence rather than listing every item.
- Route directions. Give directions from the front door to the shop as a numbered list. Then walk it and see where the list breaks.
- Board games with conditional rules. Anything of the form if you land here, then this, is teaching conditionals in plain language.
- Dance and clapping routines. Eight counts, repeated four times, with a variation on the last one. That is a loop with an exception.
Ten unplugged minutes twice a week does more than an hour of tapping through a tutorial the child does not understand.
What is realistic at 5, 7 and 10
At five, expect sequences of three to six steps and a lot of trial and error. A five-year-old can drag four arrow blocks to move a character to a target, and will often solve it by trying combinations rather than by planning. That is fine and entirely age-appropriate. What they usually cannot do is predict the outcome before pressing go, or explain why a sequence failed. Sessions run about eight minutes. Loops are mostly out of reach, though repeating a short pattern by hand is not.
At seven, planning starts to appear. A seven-year-old can hold a goal, choose an order deliberately, and, the real shift, say what they expect to happen before running it. Loops become genuinely usable: they can see that four identical blocks could be one block repeated four times, which is the first properly abstract idea in the whole subject. Ask which step went wrong, rather than pointing at it, and they will often find their own bug. Simple conditionals make sense in context: if the character touches the wall, turn.
At ten, a child can plan a small project before starting, keep several rules running at once, and reuse a solution they built earlier. Ten is where a first game with a score, a timer and a win condition becomes realistic, and where variables stop being mysterious. It is also a reasonable point to try a text language, if the child types with some fluency and actually wants to. If they do not want to, blocks are still doing the work. Tools aimed at this age, including the coding activities in something like Kid Genius World, tend to sit right at this transition.
Across all three ages, the useful measure is not how many puzzles were completed. It is whether the child can tell you what they were trying to make.
The same idea, different school calendars
Parents comparing notes across countries often worry that their child is behind. The variation is mostly in timing and naming rather than in substance.
England's national curriculum has included computing from the first years of primary school since 2014. Japan made programming education part of elementary schooling from 2020, taught largely through existing subjects such as maths and science rather than as a standalone class. Estonia has run national programmes encouraging technology and programming in primary schools for over a decade. Australia's digital technologies curriculum begins in the earliest year levels. India's 2020 education policy brought coding into the middle school years. Plenty of other systems introduce it later, or leave it to individual schools and after-school clubs.
What is consistent is the approach at primary age. Nearly every system that starts early treats this as a way of thinking rather than as training in a particular language, and no country is teaching six-year-olds to write production software. A child who begins structured coding at eleven with a solid grasp of sequencing is in a better position than one who has been clicking through block puzzles since five without ever forming a plan.
Signs it is going well
Certificates and completed-level counts are poor indicators. A child can finish forty puzzles by trial and error and understand very little. Four other things are worth more:
- They predict before they run it. Saying that it is going to go too far is a bigger milestone than a correct answer.
- When it fails, they change one thing at a time instead of shuffling everything.
- They reuse something from a previous project without being prompted.
- They start building something nobody asked for.
That fourth one usually arrives somewhere between eight and ten, and it tends to look unimpressive: a cat that walks across the screen and announces that the child's brother is smelly. Take it seriously anyway. Someone deciding for themselves what to make, and then making the machine do it, is the entire skill in miniature.