Daily Ethnicity Quiz: The Arithmetic of a Daily Score

Ten guesses is a small sample

A daily set contains a fixed number of faces, and that number is the sample size behind your score. Ten guesses is a small sample. Treat the day's total as an estimate of how well you read the map, and the estimate arrives with a spread that most players never think about. Two people of identical skill can finish the same ten faces with scores several hundred points apart, and nothing about either player explains the gap. The gap is the sample talking. This is not a reason to distrust the score. It is a reason to read one day as one day, and to stop treating a personal best as evidence of a permanent change in ability. A round of ten faces is one measurement, taken once, on a map that mixes sparse regions with crowded ones.

Averages are the standard remedy for a noisy measurement, and the daily format supplies them for free. Play for a month and you have roughly three hundred faces behind your number; play for a year and you have several thousand. The day-to-day scores will still jump, because the draw changes every morning, but the running average settles. It settles quickly at first and then barely moves, which is how any average built from limited data behaves.

The arithmetic of a lucky day

Set out an illustration with round numbers. Suppose scores on ten faces behave like ten roughly independent draws, each totalling about half the available points with a spread of a few hundred. The standard deviation of the daily total then scales with the square root of ten, or about three times the spread of one face. On a scale where a day is worth 5,000 points, that is a swing of several hundred points produced by nothing more than which faces came up and in what order.

The consequence is blunt. A few hundred points between two players on the same day says almost nothing, because the gap sits inside the spread. It becomes meaningful only when it survives many days, which is another way of saying it needs a bigger sample. The faces are not perfectly independent, either: one shared misreading of a crowded region can cost several faces at once, which makes a single day's total wider still. This is arithmetic, not a judgement about anybody's eye.

Why a hot streak proves less than it feels

Francis Galton gave the pattern its name in an 1886 paper in the Journal of the Anthropological Institute, "Regression towards mediocrity in hereditary stature." He had spent years measuring sweet pea seeds and then the heights of 930 adult children and their 205 parents. Daughters' heights were multiplied by 1.08 before comparison, and he found that extreme parents produced children who landed, on average, closer to the middle than the parents themselves. He put the regression ratio at roughly two parts in three, and the word he chose for the effect has outlived the peas.

The lesson travels far beyond height. Whenever you pick the best result out of a noisy series and then measure again, the next result will tend to be worse, and it will tend to be worse for reasons that have nothing to do with skill changing. Your best day of the year is a day when the draw and your attention lined up; the following day is simply a normal day. A streak in a daily quiz is the same story told more slowly. The run is real, but part of it is the draw being kind, and the average is waiting to pull the number back.

A year of daily sets

Three hundred and sixty-five daily sets of ten faces come to 3,650 guesses in a year. The wobble of an average falls with the square root of the sample size, so going from ten guesses to 3,650 cuts the spread by a factor of about nineteen. A yearly average is therefore a far more stable number than any single day, and it is the figure worth comparing between players. It is also the figure nobody screenshots.

A stable number is not a complete one. A 3,650-guess average measures how closely your pins track a fixed catalogue of documented distributions. It does not measure your eye for faces in the street, it transfers poorly to people the catalogue never recorded, and it says nothing at all about the identity of anybody. The number gets precise with repetition; its scope stays exactly where it started.

Who decides when today starts

Every daily game needs a moment when the board clears, and that moment is an editorial choice rather than a fact about the world. Midnight is a convention. The inhabited world spans twenty-six hours of clock offset, from twelve hours behind UTC to fourteen ahead, so one fixed reset instant lands in the middle of the afternoon for some players and in the middle of the night for others. Whoever the boundary is set for gets to finish a day's puzzle while a counterpart elsewhere is still asleep.

A single shared reset is the only arrangement that lets one global leaderboard exist, so it is the right choice. It just means that "today" on the scoreboard is not the same interval as today on your wall calendar, and that a score posted at three in the morning is competing against a day that began for another player at a different hour. The date label belongs to the game's design, not to your performance.

The catalogue is the part that does not reset

The quiz changes every day. The material it draws from does not. Each entry in the catalogue carries a documented description, a named author, a year and a set of coordinates. Inuid, the northernmost entry, sits at about 78 degrees north and was named and defined by Neumann in 1952 and again in 1966, with the Mongolian fold frequent among its recorded traits. Robust Polynesid was described by Eickstedt in 1934, after Sullivan had called it an Indonesian-like element in 1923, and its range runs from Hawaii and Tonga to the Cook Islands and Tahiti. Shillukid, named by Lundman in 1967, is anchored to the savannahs of the Upper Nile.

Some entries describe patterns that have since become scarce. Barrinean, identified by Tindale and Birdsell on their 1941 expedition to Queensland, is recorded as practically extinct, pushed back first by neighbouring groups and later by European colonists; around 1900 a few were still surviving in the rainforest of the far north. That is what a historical catalogue is for. It records what observers described in named places at a named time, which is why an entry can be checked, and why some of what it holds no longer has living representatives to point to.

The map is older than its players

The catalogue describes distributions recorded against a map roughly 1,500 years old. The players are not. In 2020 the United Nations estimated 281 million people living outside the country of their birth, up from 173 million in 2000 and 84 million in 1970, with about two thirds of them concentrated in just twenty countries. A quiz that asks you to place a face somewhere on a historical distribution is therefore not asking where you are, and it is not asking where the person beside you is.

That distance is the point of the exercise, and it is also its limit. The scores compare players against a fixed map of documented appearance patterns. They do not license any inference about an individual, because the catalogue was assembled for historical description rather than identification. Opening the daily set and matching a composite against a record is one task. Reading a real stranger is a different one, and no score here prepares you for it.

Frequently Asked Questions

Does a higher daily score mean I am better at reading faces?

No. The score measures how closely your pins track a fixed catalogue of documented distributions, on a map rebuilt from records roughly 1,500 years old. It does not measure your skill with real faces, and it transfers poorly to unfamiliar people. Ten faces in one day is also a small sample, so a difference of a few hundred points between two players on the same day sits inside the normal spread. Look at a running average over months before reading anything into one number.

Why does my score swing so much from day to day?

Because the daily set is small and the faces change. Ten guesses is a short sample, and the day's total carries a standard deviation of roughly three times the spread of a single face. The faces are also not independent: one crowded region can cost several faces at once. Averages settle as the sample grows, which is why a monthly or yearly average holds steady while individual days jump around. The swing is the format doing exactly what a small sample does.

When does the daily ethnicity quiz reset?

The board clears at one fixed moment chosen by the operator, and every player worldwide shares that boundary. Because the inhabited world spans offsets from twelve hours behind UTC to fourteen ahead, the reset lands in the afternoon for some players and the middle of the night for others. A single instant is what makes one global leaderboard possible. It also means the "today" on the scoreboard is not identical to today on your own wall calendar.

Related Phenotypes

Faces from the encyclopedia that appear in this article. Open any entry for its full description, distribution, and references.

• Inuid

  Arctic

  Eskimid proper type, ranging over the vast, sparsely populated tundra along the icy coasts of Alaska, Nunavut, and Labrador (Canada) to Gree...

• Barrinean

  Australia

  Australid subtype of the forest belt associated with the Australian pygmies. Could be the remnant of an ancient Negrito strain that once occ...

• Shillukid

  East Africa

  Nilotid variety similar to Dinkaid, but with slightly modified features. Typically found in the savannahs of the Upper Nile of South Sudan. ...

• Robust Polynesid

  Polynesia

  Relatively recent Polynesid variety that developed after the Southeast Asian migration waves had reached Oceania. Most likely a brachycephal...