Methodology · Reviewed 7 October 2026 · Model coefficients unchanged
Follow the calculation.
Not Married? is a preference simulator with jokes. Its useful output is the comparison between settings within the same model. It does not count actual compatible or available people, predict marriage, or measure anyone’s desirability. There are no profiles, introductions or matching records.
Public data and reporting dates
The browser requests the latest non-empty value for each of six World Bank WDI indicators separately. “Latest” does not mean the current year, and different indicators can come from different years. Public statistics themselves may be estimates. Open the result’s calculation panel to see the actual values and years used in that run.
| Dataset | Definition and use |
|---|---|
Total populationSP.POP.TOTL | All residents; not singles or available partners. |
Population ages 15–64 (%)SP.POP.1564.TO.ZS | Not a direct measurement of adults aged 18–55. The model scales this by 0.62. |
Female population (%)SP.POP.TOTL.FE.ZS | All ages. Published sex categories do not measure gender identity, orientation or mutual availability. |
Urban population (%)SP.URB.TOTL.IN.ZS | National urbanization, not population within a radius or a specific city. |
Current tobacco use, adults (%)SH.PRV.SMOK | A tobacco-use proxy; not a direct measure of cigarette-free and vape-free people in your age range. |
Bachelor’s or equivalent and above, age 25+ (%)SE.TER.CUAT.BA.ZS | Not education among all adults or among the selected age/partner group. |
The country reference tables were retrieved on ; the API reported a database update of 13 July 2026. Those dates are not reporting years. Each row shows its own reporting year. For the Saudi example below, population, age, female share and urbanization are from 2025; tobacco use and bachelor’s attainment are from 2024.
Inspect the dated reference snapshot (JSON) or the World Bank API documentation. WDI indicator pages identify upstream providers and display a CC BY 4.0 license. Our transformations and game assumptions are not supplied or endorsed by those providers.
When a request fails or a value is missing, the game uses its existing bundled population and regional defaults. Original source dates were not recorded for the bundled population list; it is explicitly labeled as an undated fallback, not retroactively attributed to a dataset. Missing values are not treated as zero. Cached country lists last up to 30 days and per-country indicators up to 90 days. Results can change when data loads, caches expire or sources revise their series. The reference snapshot does not silently replace the game’s live inputs.
The actual calculation
Starting pool = population × modeled eligible-age share × assumed single share × partner share.
Then apply the age, distance and selected preference multipliers in order.
“Limit” below means keep a value between the stated minimum and maximum. Every factor is part of a scenario, even when it begins with a public statistic.
| Component | Current implementation |
|---|---|
| Starting eligible-age share | Public age-15–64 share × 0.62, limited to 28–48%; otherwise a regional constant. This is an assumed proxy, not a measured adult count. |
| Single share | A regional assumption, from 43% to 50%; no marital-status or relationship-status dataset is used. |
| Partner share | Female share is limited to 46–54%, or defaults to 50.5%. Men use its complement. “Anyone” uses 98.5%, an assumed factor. Your own gender does not change the estimate. |
| Age | ((ageMax − ageMin + 1) / 38)^0.9, with the inner fraction limited to 0.025–1. Two intervals of the same width produce the same age factor even at different ages. |
| Distance | Country = 1. City = 0.07 + 0.09 × urbanShare. Nearby = 0.25 + 0.14 × urbanShare. Urban share is limited to 18–100% or defaults to 62%. No city boundary or kilometre distance is used. |
| Height | An illustrative normal distribution with regional male/female means and a 7 cm standard deviation. Minimum and maximum height use opposite tails. The raw height factor is floored at 0.008. |
| Education | School = 0.78. Bachelor’s uses the public attainment indicator, limited to 4–72%, or a regional assumption. Graduate = bachelor’s share × 0.24. |
| Lifestyle and intentions | Non-tobacco-use = 1 − tobacco prevalence (prevalence limited to 3–55%). No drinking, activity, wanting children and marriage use regional assumptions. Long-term = min(0.86, marriage share + 0.22); dating = 0.64. |
| Core dampening | Each raw core factor is limited to 0.003–1. Age and distance are applied directly; other core factors are raised to 0.84. This softens reductions but does not establish real statistical dependence. |
| Optional numerical preferences | Appearance shares are multiplied and raised to 0.62, then the extra-factor transformation below is applied. Hair length shares: 0.35 / 0.30 / 0.35; beard: 0.35 / 0.35 / 0.30. Appearance and career use undocumented regional priors, not comparable population measurements. |
| Personality and income | Personality assumptions: funny 0.38, calm 0.43, outgoing 0.38, introvert 0.40, ambitious 0.34, golden 0.23, blackcat 0.21. For multiple choices, raw = min(smallest individual share, product^0.65). Income assumptions: stable 0.64, top25 0.25, top10 0.10, trust 0.012. These are not measured joint distributions or monetary cutoffs. |
| Extra-factor dampening | Each extra raw factor is limited to 0.003–1 and raised to 0.72. For example, the “top 10%” setting applies 0.10^0.72, approximately 19%, after this transformation. |
| Mutual interest | A game factor: limit(0.18 + singleShare × 0.16, 0.22, 0.31). It is not a measured probability that someone will like you. |
| Rounding | Round after the core filters, then apply optional numerical preferences, then round again. If mutual interest is enabled, multiply that rounded number and round again. Public counts of 100+ display two significant digits. The exact trace is an arithmetic check, not an accuracy claim. |
Regional defaults are shared across groups of countries; they are not country surveys. The result panel lists actual source inputs, effective multipliers and intermediate values. Its loss bars depend on calculation order and are not causal estimates. The new single-change comparisons recalculate from the same answers with only the named preference changed; they do not sum the bars.
Worked example: Saudi Arabia
Women, ages 24–38, nationwide, long-term relationship; height, education, lifestyle, children, appearance, personality, career and income unrestricted; mutual interest off. The example uses the dated public inputs above. It is generated and tested against the calculator function, not a separately invented formula.
| Step | Value | Origin or operation |
|---|---|---|
| Population | 36,973,555 | World Bank SP.POP.TOTL, 2025 |
| Eligible-age proxy | 0.454597 | 73.322138816712% (2025) × 0.62; within the 0.28–0.48 limits |
| Single share | 0.49 | Gulf model assumption; not a statistic |
| Partner share | 0.46 | Female population 39.574660635055% (2025), forced up to 46% by the existing model |
| Starting pool | 3,788,540.515227 | Population × eligible-age proxy × single share × partner share |
| Age 24–38 | 0.433188 | 15 inclusive years / 38, raised to 0.9 |
| Nationwide | 1 | No distance reduction |
| Long-term intent | 0.829078 | (0.58 + 0.22)^0.84; 0.58 is the Gulf marriage assumption |
| Before rounding | 1,360,643.719234 | No appearance, personality, career or income restriction |
| Internal result | 1,360,644 | Rounded integer; mutual interest off |
| Public display | About 1,400,000 | Illustrative estimate, not identified people |
The table prints extra decimals only to make the arithmetic inspectable; calculations retain full precision internally. “1 in” and percentage remaining use the modeled starting pool, not total Saudi population and not a personal success probability.
Choices used only for presentation
City names, local flavor, style, cars, housing, homemaker preference, head/face covering and everyday priorities do not have prevalence factors. They appear in jokes and share cards. Selecting homemaker clears the separate numerical career filter, so that UI change can increase the estimate by removing a career restriction. No city population is loaded. Your own gender changes wording and colors, not the calculation. “Surprise me” selects a personality assumption at random; it is revealed and retained in shared links.
What the estimate cannot establish
- It cannot establish how many people are single, reachable, interested in you, or compatible with you.
- The fixed female-share clamp materially changes some countries. In this Saudi example, 39.6% is changed to 46%. This is a known model limitation, not a correction to the source statistic.
- Age 18–25 and age 30–37 have the same width and the same age multiplier. The model does not describe the different demographics of those groups.
- The city and nearby options are national fractions. They cannot distinguish Riyadh from Jeddah, or New York from a small town.
- Population includes children and people already in relationships; subsequent game assumptions do not turn it into an observed list of available adults.
- Traits overlap. Powers of 0.84, 0.72, 0.65 and 0.62 are unvalidated adjustments, not measured joint probabilities.
- Source age bands, definitions, population universes and reporting years differ. Binary source sex data do not fully represent gender identities or orientation.
- Zero means this model rounded to zero. Large results can follow a large population or generous assumptions. Neither proves anything about your worth or prospects.
Use one-change comparisons to see which assumptions drive the output. Keep values and boundaries that matter to you; the calculator cannot tell you which preferences you should give up.