Why do different countries count age differently?

The reckoning that changed by law in 2023

The international convention is so familiar that it looks like the only possibility: a person is zero at birth and gains a year on each anniversary. It is one system among several, and a major alternative was in everyday use very recently.

East Asian age reckoning counts a newborn as one year old, on the reasoning that time in the womb counts, and then adds a year to everyone simultaneously at the lunar or solar new year rather than on individual birthdays. Under it, a baby born in late December is one at birth and turns two a few days later.

The consequence is that a person could be up to two years older by traditional reckoning than by international count, with the gap depending on how close their birthday sits to new year. In practice this meant two ages in circulation for the same person — the traditional one used socially, and the international one used for law and medicine.

South Korea is the clearest case because it maintained three systems at once: traditional Korean age, international age used in most legal contexts, and a third calendar-year age used for school entry and conscription that counted the difference in birth years. Legislation passed in December 2022 and effective from June 2023 standardised on international age for administrative and legal purposes, which changed the stated age of most of the population overnight.

The practical relevance is for anyone handling ages across cultures: an age reported in conversation may not be the age on a document, and the difference is not an error. Where age matters, the date of birth is the unambiguous datum and the age is a derived value.

The arithmetic is less obvious than it looks

Calculating completed years between two dates is a classic source of off-by-one errors, and the reason is that months have different lengths.

The correct method takes the difference in years, then subtracts one if the current month and day have not yet reached the birth month and day. Subtracting dates and dividing by 365.25 gives an approximation that is wrong near birthdays, which is exactly when people check.

Leap years are where implementations fail. Someone born on 29 February has no birthday in most years, and jurisdictions differ on when they legally age. Several treat 1 March as the operative date, on the basis that the person has completed the year once February ends. Others use 28 February. The distinction is usually academic and occasionally is not — where an age threshold governs eligibility on a specific day, the choice determines the answer.

Time zones add a subtler problem. Age computed from a date of birth and the current date depends on where "now" is evaluated, so a system running in one zone can consider someone a day older than a system running in another. For anything where the threshold matters — age verification, benefit eligibility — the birth date should be treated as a plain calendar date rather than a timestamp, and compared against a date in a defined zone.

Duration in years is also not reversible through addition the way durations in days are. Adding one month to 31 January has no obvious answer, and different libraries return 28 February, 2 March or an error. Anywhere a calculation chains month or year arithmetic, the ordering of operations changes the result, which is why date libraries document their clamping behaviour and why rolling your own is a poor idea.

Dog years are not seven to one

The rule that one dog year equals seven human years is the most widely repeated fact about pets, and it is wrong in a way that matters for their care.

Its origin appears to be arithmetic convenience rather than evidence — average human lifespan divided by average dog lifespan, some decades ago, giving roughly seven. It has no biological basis and it fails obviously at both ends. A one-year-old dog is not equivalent to a seven-year-old child; it is sexually mature and close to full size, which puts it nearer a human teenager. And a fourteen-year-old dog is not equivalent to a ninety-eight-year-old human, since plenty of dogs reach fourteen and very few humans reach ninety-eight.

Ageing in dogs is front-loaded and then slows. A widely cited approximation treats the first year as roughly fifteen human years, the second as about nine more, and each subsequent year as four to five. That maps far better onto observed development, and it explains why the first two years of a dog's life involve so much more change than any pair of years afterwards.

Size complicates it further, and in the opposite direction to most mammals. Larger dogs age faster and die younger — a great dane is elderly at seven while a small terrier may be middle-aged at ten. Any conversion that ignores breed size is discarding one of the two strongest predictors.

A 2020 study proposed a more principled mapping based on epigenetic changes — patterns of DNA methylation that track biological ageing in both species — and produced a logarithmic relationship rather than a linear one. Its practical output resembles the front-loaded approximation: rapid early equivalence that flattens with age. The detail matters less than the shape, which is that dog ageing is not proportional to human ageing at any fixed ratio.

The reason this is worth getting right is veterinary. Screening recommendations, dietary changes and the point at which a dog is considered senior all depend on realistic ageing, and a seven-to-one rule places those milestones in the wrong years — late for large breeds and early for small ones.

Ages that are counted in other units

Several contexts deliberately avoid years, because years are too coarse at the scale that matters.

Infant age is counted in weeks and then months, since development over the first two years is fast enough that a year-based figure conveys almost nothing. Paediatric guidance additionally uses corrected age for babies born preterm — chronological age minus the weeks of prematurity — when assessing milestones, so a baby born two months early is compared against a point two months behind the calendar.

Gestational age runs on its own clock entirely, measured in completed weeks and days from the last menstrual period rather than from conception, which is why a pregnancy is described as forty weeks while actual gestation is about thirty-eight.

Employment and legal contexts often count in service years or working days rather than calendar time, and the two diverge quickly. A qualifying period expressed in working days excludes weekends and public holidays, so thirty working days is six weeks rather than a month — a difference large enough to change a deadline materially.

Where any of these are computed programmatically, the same principle applies as with age itself: store the underlying date, compute the derived figure on demand, and be explicit about which convention is in use. A stored age is wrong within a year of being written, and a derived one never is.

Questions