Dropdown Time Tax: Lifetime Selection Time Estimate
Residents whose place of residence appears near the top of an alphabetical dropdown menu (for example, Alaska, the second U.S. state) spend less time selecting it than residents near the bottom (for example, Wyoming). This form first measures how fast you select from a dropdown, then asks for your address one dropdown at a time, the way address forms do, and estimates what your position in each list costs you over a lifetime.
Step 1. Selection speed test
Step 2. Address
Statement of lifetime dropdown liability
Complete Steps 1 and 2 to generate your statement.
Technical notes
1. Purpose and scope
This form estimates the additional time a person spends selecting a place of residence from alphabetically ordered dropdown menus (HTML select elements) because of that place's position in each list. Only the position-dependent part of each selection is counted. Time spent opening the menu, reading the field label, and confirming the choice is the same for the first and the last entry and is excluded. Estimates are illustrative. They combine your own measured selection speed with the lists used by many address forms and with usage assumptions stated in Section 5, which are not drawn from any survey.
2. Measurement of selection speed (Step 1)
The test list contains 250 entries, equal to the length of the country list in Step 2, made up of 207 dish names and 43 words from languages other than English, all real and all written in the Latin alphabet, sorted with the same collation as the address lists. Unfamiliar entries were chosen so that the test measures scanning and scrolling rather than memory of where a familiar name sits; a familiar list may be faster (Section 7). Trial j (j = 1, …, 8) draws its target at random from the j-th eighth of the list, so targets span the top through the bottom. The timer starts when the target is displayed and stops when the matching entry is selected; a wrong selection does not stop it. For each trial the page records the target position i (the first real entry is 1) and the elapsed time t in seconds from the browser's high-resolution clock (performance.now). An ordinary least-squares line t = a + b·i is fitted to the 8 points. The slope b, in seconds per row, is your selection speed; the intercept a absorbs reading the target, opening the menu, and confirming. A negative slope is set to zero. The correlation coefficient r is reported as a check on fit quality. Users who select by typing the first letters of an entry (browser type-ahead) typically produce a slope near zero, which correctly implies that list position costs them almost nothing. If the test is not completed, a default slope of 35 ms per row is used, an estimate from an earlier mixed-method model in this series (40% mouse with 20 visible rows and a 12 rows/s wheel scroll, 8% keyboard type-ahead, 37% touch picker at 0.03 s per row, 15% browser autofill), which gave about 1.8 s for the 51st entry of a state list relative to the 1st, or about 36 ms per row.
3. Computation of the per-selection tax
For an entry at position i in a list, the per-selection tax is τ = b·(i − 1) seconds, the predicted extra time relative to the first entry of the same list. Each address field is a separate list and is taxed separately: country (Step 2, first field), first-level division (state, province, prefecture, region, or the term the dataset uses for that country), city, and, where shown, district. The same measured slope applies to every list. Lists longer than the 250-entry test list (for example, 1,064 cities in California and 1,759 in Bavaria) are an extrapolation beyond the measured range (Section 7). Each field also reports the time you actually took for that selection, measured from first pointer or key contact with the menu to the change of value, and the time your fitted line predicts (a + b·i).
4. Computation of lifetime totals
Lifetime liability for a field is L = τ · n · Y, where n is the number of times per year that field is filled from a dropdown and Y is the number of form-filling years (Section 5). The statement sums L over the fields you completed. The comparison figures under each field (the first and last entries of that list, and Alaska and Wyoming in the U.S. list) use the same formula and slope.
5. Assumptions
The values below may be changed. Results update immediately. Values marked "estimate" have no survey source and are the author's assumptions.
| Symbol | Quantity | Value | Basis |
|---|---|---|---|
| n0 | Country dropdown selections per year | Estimate. International sites, travel bookings, app sign-ups. | |
| n1 | State / province selections per year | Estimate. About one per month: checkout, shipping, tax and benefit forms, which often ask for the state without the country. | |
| n2 | City selections per year | Estimate. Most forms use a free-text city field; dropdowns are less common. | |
| n3 | District selections per year | Estimate. Applies only where a district field is shown. | |
| Y | Form-filling years | U.S. life expectancy at birth, 78.4 years in 2023 (CDC NCHS Data Brief 521), less about 16 years before a person fills in forms independently. Estimate. | |
| b0 | Default slope if Step 1 is skipped | ms | Estimate; see Section 2. |
6. Lists, hierarchy and ordering
The address lists reproduce the Countries States Cities Database (Section 8), an open dataset that is widely used to populate country, state and city dropdowns on websites and in form libraries. It lists 250 countries and territories. The second field lists that country's first-level divisions as the dataset records them, labelled with the division type the dataset gives for most of them (for example State, Province, Prefecture, State (Land)); where types are mixed the label reads "State / province / region". For the United States this list has 60 entries, because, like many U.S. address forms, it includes the District of Columbia, five territories, the U.S. Minor Outlying Islands and three Armed Forces postal regions; Alaska remains second and Wyoming last. The third field lists the populated places the dataset assigns to the selected division, excluding records the dataset documents as administrative units rather than settlements (types county, regency, prefecture, parish, banner, province, oblast, area, abandoned and historical) and merging exact duplicate names. In Japan the municipalities of Tokyo include its special wards (for example Shibuya-ku), as Japanese address forms do. A fourth field, district, is shown only where address forms in that country conventionally ask for it and the dataset supports it: in China, where the order is province, prefecture-level city, then district or county (for example Xinjiang, Ürümqi, Tianshan District). For the four Chinese municipalities (Beijing, Chongqing, Shanghai, Tianjin) the third field lists districts directly. Countries without recorded divisions, or divisions without recorded places, end the address at the last available field. All lists are sorted with the browser's English collation (Intl.Collator, base sensitivity), so accented letters sort with their base letter; other sort orders would shift positions slightly. Live sites vary: some place a default country first, some use search boxes, autocomplete, free-text fields, or postal-code ordering (in which Alaska, AK, is first and Wyoming, WY, is still last). Those designs change the result and are not modelled.
7. Limitations
(a) The linear model assumes each additional row costs the same time. Mouse users scroll only past rows hidden below the open menu (about 20 in Chromium-based browsers on Windows and Linux), so the true cost is lower near the top of a list and higher further down; the fitted slope averages over this. (b) Eight trials give an imprecise slope; one slow trial can move it noticeably. Repeat the test to check. (c) The test uses unfamiliar words. People who know roughly where their own country sits may jump there faster than they scan an unfamiliar list, so the measured slope may overstate their cost; the actual time recorded for each address field allows a comparison. (d) For lists much longer than 250 entries people usually switch strategy (type-ahead, browser find, or a free-text field), so city figures for very long lists overstate the cost. They are the time you would spend if you kept scrolling at your measured rate. (e) Selection counts per year are assumptions, and lifetime figures are proportional to them. (f) The dataset is community-maintained; some divisions are incomplete, some names are transliterated inconsistently, and city coverage varies between countries. (g) Times are continuous clock time: "2 hours" means 2 hours of selecting summed over a lifetime.
8. Data sources
Address lists: Countries States Cities Database (github.com/dr5hn/countries-states-cities-database), files countries.csv and states.csv from the repository and cities and counties from release assets, retrieved October 25, 2026, licensed under the Open Database License (ODbL) 1.0; the database draws in part on GeoNames (CC BY 4.0) and OpenStreetMap contributors. Chinese prefecture-level cities and districts are the dataset's level-2 and level-3 county records. Speed-test list: compiled for this piece from published dish names and dictionary words; no source dataset. U.S. life expectancy: National Center for Health Statistics, "Mortality in the United States, 2023", Data Brief No. 521 (December 2024). Chromium visible-row limit: kMaxVisibleRows = 20, Chromium popup menu source. The page makes no network requests; all data are embedded.