paper-to-game

Me and My Friends — design memo

0. Given

A gift, not one of Steffen’s: the beta line goes in the footer, no signature, no counter. One idea given: given the theme, perhaps the visuals should sort of rely on child-like drawings. Recorded as binding. It overrides the precedent rule on the register family (section 6): the last three games are Austere, Instrument and Drawn, and the Given puts this one in the Drawn family too, as Julian’s Window is. Everything inside the family differs: crayon on buff paper instead of ink and screentone on a white page; five crayon colours instead of ink and one rose; a seven-year-old’s hand instead of a comic letterer’s; wobbly ringed answers instead of panel borders.

1. Paper

Manuel Bagues and Natalia Zinovyeva, Gender Segregation in Childhood Friendships and the Gender-Equality Paradox, working paper dated 2 July 2025 (University of Warwick), 83 pages with appendix. JEL J16, I2. No DOI in the PDF; no code or data attached. Read from inbox/ as a typeset PDF: every number below is from typeset text or tables except where marked read off a figure. The paper can be read on the authors’ pages at Warwick; the memo cites by section, table and figure number.

2. What the paper hands over

Primary: an estimate. Two datasets and a long table of associations. Country level (HBSC, 37 countries, ages 11–15, 2002–2010): the share of close friends who are of the opposite sex (mean 27 per cent at 11; Finland 16, Portugal 39; Table A1, p. 17) correlates negatively with every measure of “vertical” gender equality (GII ρ = 0.43, GGGI ρ = −0.48, “men have more right to a job” ρ = 0.36), with GDP per capita (ρ = −0.39) and, most strongly, with the share of parents who say a child should learn independence rather than obedience or hard work (ρ = −0.67; Figure 3, Table 1). Individual level (Millennium Cohort Study, about 10,000 British children born 2000–01): children from richer families have fewer opposite-sex friends (lowest income fifth 65.6 per cent with a mixed friend group at 11, highest fifth 48.2; Figure 4; −3.9 points per standard deviation of income, Table 3), in 34 of the 37 HBSC countries too; but, inside the country, children whose mothers hold more gender-egalitarian views have more opposite-sex friends (+0.8 points per SD overall, +1.8 at age 11; Tables 3, A14). Clubs outside school reduce opposite-sex friendship (−2.1 points per SD), unsupervised time raises it (+3.3), a good relationship with the parents lowers it (−1.4), “think for yourself” parents’ children have 3 points fewer than “obey” parents’ (Tables 4, 5). Richer children are calmer, more cooperative, score higher, enter puberty later, and every one of those traits goes with same-sex friendship; together they account for about a quarter of the income gradient, the week (clubs, roaming, parents) for about an eighth (Table 6, pp. 33–34). Richer children are also more popular and more satisfied with their friends (Table A18): preference, not constraint. Downstream: children with opposite-sex friends at 7 pick subjects at 16 with a slightly higher opposite-sex share (0.067 SD at GCSE, 0.072 at A-level), still have opposite-sex friends at 14 (+6 points), hold more egalitarian views at 14 (0.09 SD) and want jobs with about 3 points more opposite-sex workers (Table 7). At the country level, where children mix, women’s propensity to graduate in STEM is higher (ρ = 0.43) and the Duncan segregation index lower (ρ = −0.47; Figure 5).

Secondary: an argument. Lareau’s two childhoods: concerted cultivation (clubs, lifts, supervised afternoons, self-expression) against natural growth (the street, whoever is there); and Charles and Bradley’s “gendered selves”, that prosperity lets children express gender-typical preferences in carefully arranged, gender-sorted settings. The paper does not formalise it, and says plainly that none of its estimates is causal.

3. The discovery, and its kind

A counterintuitive result, twice. The public belief: in rich countries where women and men are treated alike, boys and girls will be friends more. The paper: less, and already at eleven, and the countries where the playground is most sorted are the ones whose universities are most sorted by subject. Then the second turn, which is the paper’s own contribution: look inside a country and the equality part of the paradox reverses (egalitarian mothers’ children mix more), while the money part stays (richer families’ children mix less, everywhere). So the paradox is prosperity, not norms, and what prosperity buys: a week of dance class and football club and a lift home, and a calm, popular child who is happy with the friends they have. The game stages both turns and ends on the chain the paper then draws to subject choice at sixteen, read, not played.

4. The core narrative

In rich countries, where mums and dads say girls and boys can do anything, you would think boys and girls would be friends more. They are friends less. The reason is not what the parents believe. It is money, and what money buys: rich kids go to dance class and football club and get driven everywhere, and they are happy with the friends they make there, who are mostly all girls or all boys. Kids with no clubs play out on the street with whoever is there. And the kids who had friends of both kinds at seven pick slightly different subjects at sixteen, and the countries where the playground mixes are the ones where the university mixes.

5. Who lives in the paper

6. Precedent

  Position Format Feedback device Chassis
Leavitt’s Cloud experimenter at a light box discovery with a prediction close the fold (dots lining up into one tooth), a stroke dropped on a chart Austere (a glass negative in a dark room): velvet, glass, violet ink; Spectral + Josefin; pills
Jevons’s Noon controller: a nation setting a pace (its own memo says agent) challenge, scored against the best path the conservation sheet: a curve drawn, coal squares emptied Instrument (Jevons’s own plate): cream, rule blue, soot, furnace orange; Bodoni Moda; engraver’s labels
Julian’s Window subject / witness choosing where to look discovery, no score the page’s weather and a strip of sixteen frames Drawn (shōjo manga): white page, ink, one rose; Delius + Italiana; panel buttons

What this game does differently. Position: the parents of a street, answering the paper’s own survey (an agent, plural), after a prologue in which the player is a country. Not an experimenter, not a witness, and not a nation drawing a curve: the controls are the four survey questions the paper measures. Format: a challenge the player cannot meet the way they expect. A target is given in the aim’s units (kids in ten with a friend from the other side), the obvious levers go the wrong way, and the game names the shortfall as the result. Neither a scored allocation over time nor a measurement game. Device: one of our own, the crossing: a child’s drawing of the playground with boys on one side and girls on the other, friendship lines drawn one by one, red only when a line crosses the middle, and a crayon ruler from 0 to 10 under it where each try drops a mark. No fold, no sheet, no weather. Chassis: crayon on buff sugar paper (Drawn, by the Given); a light page after a dark one; five crayons; a child’s print hand; wobbly rings round the chosen answer. Shares nothing with Julian’s ink-and-one-rose but the family.

7. The aim and the verdict

Aim, part one: Build a country where boys and girls are friends. Aim, part two: Get the big-house kids mixing like the small-house kids. Yardstick, both parts: Of ten kids, how many have a friend from the other side, a crayon ruler with ten marks, read in the kids the aim counts.

Verdict leads with:

8. Three concepts

A. Build a Country, then Look Inside (controller for ninety seconds, then agent: the parents of a street; a challenge that cannot be met the expected way). Controls: three three-way dials for the country (how rich the houses, whether jobs go to dads or to both, what parents say a child should learn), then three for the big-house families on one street (what the mums think about mums working; what a child should learn; whether the week is clubs, a mix, or the street). Deliberation: which dial will mix the playground? Then: which lever closes the gap between the big houses and the small ones? Sees after: the playground drawn line by line, red lines across the middle, the ruler mark dropping; after the tries, Finland and Portugal placed on the same ruler. Moment: the “good” country comes out at the Finland end; then, inside, the mums’ dial goes the other way, and the gap will not close. Fun: anticipation (ring the bell), agency (the dials move the mark, the wrong way), discovery (twice), mastery (the second country is built to test one dial). Second playthrough: a different hypothesis per dial, the corners few real countries live in, the street with the levers in a different order.

B. One Childhood (agent: a single parent; a picture book of five ages; no score, fates in words). The player answers the MCS’s own questions as they come (nine months, three, seven, eleven) for one child in a dealt house; at seven, eleven, fourteen the child draws “me and my friends”; at sixteen a subject list. Deliberation: what kind of parent to be. Moment: the cultivated, cheerful child draws all girls or all boys. Fun: anticipation, discovery. Second playthrough: the other childhood. Fails the walkthrough on size: the paper’s per-child effects are a few points on a coin that lands mixed six times in ten, so one child’s drawing is noise and the player learns nothing from it; ten children would have to be drawn to show a pattern, and then the player is a street, which is A.

C. The Kid’s Week (subject: an eleven-year-old; wordless; the reckless one). A crayon week, seven afternoons; the parents’ money has filled some with clubs and lifts; the player taps where to go when there is a choice and, in each place, taps who to play with from who is there (the dance class: nine girls and a boy; the street: whoever). Lines form over the week. Deliberation: who do I pick, given who is here. Moment: a week of clubs leaves no one to pick from. Fun: agency, discovery, a screen with no words on it. Why it loses: the player becomes the demand side, and half the paper’s explanation is that rich children prefer their same-sex groups and are happy in them; a player who is the preference cannot find that out. And the room’s composition is invented: the paper never sees the club. Its walkthrough is in section 9.

9. The first five minutes

A (the pick). 0:00: buff paper, a crayon sun, two stick figures with a red line between them, the title in a child’s hand: Me and My Friends. Three lines: thirty-seven countries, half a million children, one question, “how many boy friends and girl friends do you have?” The aim on its own line. BEGIN. 0:20: a crayon country: a house, a shop with people in it, a mum and dad with a speech bubble. Three rows of three cards under them. The stranger taps the big house (the house grows a car), both the same (the shop fills with mums and dads alike), think for yourself (the bubble changes). Obvious choices; this is the country they would want to live in. RING THE BELL. 0:50: a playground draws itself: five boys on the left, five girls on the right, lines appearing one at a time between friends, black, black, black, one red across the middle, black. The ruler mark drops: six of ten. The game: Six of your ten kids have a friend from the other side. That is the Finland end of the ruler. The stranger frowns: six? 1:10: ANOTHER COUNTRY. The dials are back in the middle. They keep the big house and the equal jobs and try do as you’re told: the mark goes up a bit. 1:50: third country, everything down: small house, dads first, do as you’re told: nine of ten, red lines everywhere. They grin and are slightly annoyed. 2:10: the reveal: their three marks on the ruler, Finland and Portugal drawn beside them with their dial pictures, and the line: Every rich, equal, think-for-yourself country comes out at the Finland end. The paper calls it the gender-equality paradox, and it is on the playground at eleven, before anyone has chosen a subject. Then: But a country is just a lot of streets. Was it the equality? Look inside one. 2:40: a street in Britain, five small houses and five big. Two playgrounds: the small-house kids, seven of ten; the big-house kids, five of ten. The new aim. 3:00: three cards for the big houses: what the mums think (mums stay home / either / same as dads), what a child should learn, the week. The stranger, having just learnt that equality sorts the playground, picks mums stay home, do as you’re told, out on the street. DRAW IT. 3:30: the big-house playground redraws: the mark lands at about six. The game: Out on the street: up by about one kid in ten. Do as you’re told: up a touch. Mums stay home: down a touch. Inside one country it runs the other way: the kids of mums who say “same as dads” have a friend from the other side a little more often, not less. The stranger has to think: so the country dial and the family dial point opposite ways. 4:00: second try, same as dads with the rest kept: about six, a hair higher. Third try: everything the game said helps: still short of seven. 5:00: the game names it: You cannot get there from here. Pulling every lever as far as it goes closes about half the gap. The paper, looking at what families actually do, puts the week at about an eighth of it and the children themselves at a quarter: the big-house kids are calmer, get on with everyone, and like the friends they have. The rest nobody can explain. Two big-house kids say “I like my friends.” 5:30: the verdict page.

C (the reckless one). 0:00: a crayon week. Monday: a dance class, drawn; nine girls, one boy; the player is the boy; tap someone. The player taps the girl nearest. A line. Tuesday: football, ten boys. Tap. Wednesday: the street, four boys, three girls, a dog. Tap. By Friday there are five lines; two red. The week ends, the drawing appears: me and my friends. Then a second week, in a smaller house: more street, fewer clubs, more red. The player understands the room decides. And does not find out that the rich child is happy, because the player has been the child and was not asked. And the dance class was ours. It loses, and the game keeps its one best thing: kids drawn by a kid.

10. The pick, and why

A. The paradox is the paper’s title and its first act is a ninety-second trap a stranger walks into with their own hands: the country they would want to live in sorts its playground. The second act is the paper’s own move, the within-country reversal, and it lands as a second surprise the stranger has to think about, because they have just learnt the opposite. The target in part two is honest: it cannot be reached with the paper’s levers, and the game says so, with the paper’s shares. The sentence afterwards has no “guess”: I built the country I’d want, and its playground was the most sorted; I looked inside and it wasn’t the equality, it was the money and the week it buys. Originality: no game in the table has had the player build a country from three survey answers and then prise it open; none has drawn its data as a child’s drawing.

Author test. Bagues and Zinovyeva know the paradox and will enjoy watching someone build Finland on purpose. What they will enjoy more is that the game’s second act is their own argument: the dials that cannot be told apart from outside (rich, equal, permissive travel together, ρ 0.59–0.72) come apart inside the country, and the game says plainly where the paper’s own numbers run out (a quarter, an eighth, the rest unexplained). The corners of the country dials where few real countries live are flagged on screen, which is the caveat they wrote themselves (p. 6).

Runner-up: B, One Childhood. Switch if the author wants a single child’s story over a street’s shares and accepts that the drawings are then drawn from the paper’s probabilities with a declared magnification.

What the reckless concept would have given. The room deciding: a week of clubs with no one of the other kind to pick. Kept: the week is one of the three family dials, with clubs at one end and the street at the other, and the result line says what the week did. Given up: being the child, and a screen with no words.

11. The shape

Two parts and a verdict, about five to seven minutes. Part one: up to three countries built from three dials, each reset to the middle before every try; after each, the playground and the ruler; after the third, or after the second if the player stops, the reveal with Finland and Portugal on the ruler. Part two: the street (no choice: the two playgrounds and the gap), then up to three tries at the three family dials for the big houses, each reset to the middle; after the third, the shortfall named, the paper’s shares, the children’s bubbles. Verdict: three lines, the chain to sixteen, ANOTHER STREET, How to play, About, the beta line. Presets are the paper’s: the dial positions are the paper’s minimum, mean and maximum (Table A1), the quintiles are Figure 4’s, the levers’ ends are one and a half standard deviations. No score; the ruler is the yardstick and the verdict is in words. Seed visible in the URL; the seed sets only where the kids stand and which of them hold the lines.

12. Controls, mechanism, feedback, moment

Controls. Part one: three rows of three crayon cards (house: small, middling, big; jobs: mostly dads, dads first, both the same; parents say: do as you’re told, work hard, think for yourself); tapping a card redraws that piece of the country picture at once (the preview), and RING THE BELL commits. Part two: three rows of three (mums: stay home, either way, same as dads; a child should learn: do as you’re told, work hard, think for yourself; the week: all clubs, a bit of both, out on the street); tapping redraws the family picture; DRAW IT commits. Every row starts in the middle each try.

Mechanism. Reconstructed estimates, played as a toy (section 16). Part one: the country’s share of opposite-sex friends is the age-11 mean plus a weighted sum of the three dials, each weighted by its own reported correlation and scaled so that all-up lands on Finland and all-down on Portugal; the share of children with no opposite-sex friend follows from the share by the line through Finland and Portugal (Figure A9); the ruler reads ten times one minus that. Part two: the big-house share is the top quintile’s (Figure 4) plus the paper’s age-11 marginal effects for each lever times the distance of the lever from what a big house typically does. Robustness in section 17.

Feedback. The crossing: the playground redrawn, lines appearing one by one, red across the middle; the ruler mark dropping; earlier tries’ marks staying as small pencil ticks; the result stated by the game in its own voice (what you set, what came of it, what the mark did, what that means).

Moment. Part one, first ring of the bell: six of ten, “the Finland end”. Part two, first draw: the mums’ dial moved it the other way, and the mark fell short of seven.

13. The look

Register: a child’s crayon drawing on buff sugar paper (Drawn family, by the Given). From the paper’s subject: the data are seven-year-olds’ answers to “are your friends mostly boys, mostly girls, or a mixture?”, which is the drawing every child has made, me and my friends, stick figures in a row. The whole game is that drawing, done by a seven-year-old with a box of five crayons on the rough buff paper of a primary classroom.

14. What is cut

The three vertical-equality measures (collapsed to one dial, “who gets the jobs”); the three ages 11, 13, 15 (the game is eleven); the quality-of-friendship questions; siblings, father at home, single-sex schools (controls, not the story); the authoritative/authoritarian distinction is kept as a card but its effect is a touch; the ordered-probit against the probit (the game shows shares); the adult Facebook comparison (Figure A11); the puberty index; the romance at seventeen (Table A20); the policy paragraph of the conclusion (one sentence on the verdict screen).

15. Divergences

16. Mechanism kind and claim boundaries

Kind: reconstructed estimates, played as a toy. Every number the mechanism runs on is one the paper reports (correlations in Figure 3 and p. 14; marginal effects in Tables 4, 5 and A14; shares in Figure 4 and the text), and the calibration check reproduces the paper’s own landmarks from them. But the paper reports associations and says, in three places, that none is causal; the game lets the player set a dial and watch the share move. So:

Correlation is not causation. The dials are levers only in the game. On every result the game speaks of “the paper’s 37 countries” and “ten thousand British children”, and the verdict’s last line says the paper shows patterns, not levers. An average effect is not an individual response. The game never draws one child’s fate from a marginal effect: it draws a playground of ten whose red lines are the share, and says “of ten kids”.

Where the discovery comes from. Part one’s discovery (all dials up lands at the Finland end) follows from the three reported negative correlations and from nothing the designer chose; it survives the multivariate alternative construction on the diagonal (section 17). Off the diagonal, the two constructions disagree whenever the equality dial moves without the parents’ dial, and there the game adds the paper’s own hedge (p. 6: “these correlations may simply reflect underlying socioeconomic development”): from outside it cannot tell which does the work. Part two’s discoveries (the mums’ lever goes the other way; the gap cannot be closed) follow from the signs of Tables A14, 4 and 5 and from the size of the Figure 4 gap against them; the “cannot be closed” result is checked with the typical-big-house construction removed. The 1.5 SD lever ends are the designer’s and only change how much of the gap closes (about half, against the paper’s observed eighth plus quarter), which the shortfall text states as the difference between what families could do and what they do.

17. Calibration plan and result

Checked headlessly against the paper (calibration.js, 42 checks, calibration-result.txt): all dials up reproduces Finland’s 16 per cent share and 41 per cent with no opposite-sex friend; all down reproduces Portugal’s 39 and 6; the middle reproduces the age-11 mean (27 per cent) and about a fifth with none (p. 17: 21 per cent); each dial lowers mixing when raised, with swings ordered as the reported correlations (0.67 > 0.43 > 0.39); the alternative multivariate construction from the same correlations orders the diagonal identically, and every setting where the two constructions disagree in sign carries either the corner line or the hedge line on screen. Part two: the small houses sit at 65.6 per cent (about seven of ten); the big houses with levers in the middle within 0.03 of the top fifth’s 48.2; each lever’s sign matches its table (mums +, think-for-yourself −, work-hard between and near obey, street +), the week moves most and the mums’ views least; no setting of the three levers reaches the small-house share, the best setting is the one the paper’s signs imply, and it closes about half the gap; both survive with the typical-big-house adjustment set to zero; the Table 5 income effect per SD times the 2.8 SD between top and bottom fifths gives the Figure 4 gap. The drawing: over 300 seeds and all 27 settings, exactly round(kids) children hold a red line, twelve lines every time, and the red-line count within one of share times twelve. Words: no decimals anywhere in result text.

18. Playing it in numbers (build.md §3)

No score, but the obvious strategies were run (loop-log.txt, all 27 settings of each part). Part one, kids of ten with a friend from the other side: all up 5.9 (Finland end), middle 7.7, all down 9.4 (Portugal end); a lone dial from the middle moves the mark by 0.6 (money), 0.7 (jobs) or 1.0 (parents). Part two, big-house kids of ten against the small houses’ 6.6: worst 4.1 (mums stay home, think for yourself, all clubs), middle 5.0, best 5.7 (mums same as dads, do as you’re told, out on the street). The decisions matter (a spread of 3.5 in part one and 1.7 in part two, both legible on a ruler of ten) and no setting reaches the target, which is the sentence the player should leave with.

19. Build log