← Back to Daily Builds

Daily build note · October 21, 2026

Poll Line

Simulate one Election Day polling place: voters arrive through the day and queue at check-in, the booths and the scanner. See the longest line, who waits past 30 minutes, the bottleneck station, the fix, and the best time to go vote.

Civic / Data Toys Interactive demo live Static HTML/CSS/JS

What shipped

Poll Line is a polling-place line simulator you can watch. Enter registered voters, turnout, the share who already voted early or by mail, polling hours, and an arrival pattern (commuter, steady, midday, or a campus-style late surge). Then staff three stations: check-in poll books, voting booths with minutes to mark the ballot, and the ballot scanner. A seeded discrete-event simulation sends every Election Day voter through all three first-come, first-served queues with randomized service times. A stacked chart shows how many people wait at each station, minute by minute, with a polls close marker. Anyone already in line at close still votes, so the line keeps draining after close. A second chart shows the average wait if you arrive in each 15-minute slot, and outlines the best half hour to go. Tap, drag, or use the arrow keys to scrub the day. A readout gives the line and your expected wait at that time, and a station strip shows queue dots and busy stations. Stats cover average, median, 90th-percentile and longest wait, how many wait over 30 minutes, the longest line, the number in line at close, and when the last voter leaves. The verdict chips are SMOOTH, MANAGEABLE, SOME WAITS, LONG LINES, MELTDOWN, and EMPTY, plus a BOTTLENECK chip and a LINE AT CLOSE chip. A greedy Fix it step keeps adding a station at the bottleneck until nobody waits past 30 minutes. You can also copy an hour-by-hour markdown report stamped in America/Chicago time. State saves under poll-line-v1.

How to try

  • The demo opens on Suburban school gym: 3,200 registered, 60% turnout, 42% already voted, so about 1,114 voters come on Election Day to 11 booths. The average wait is about 9 minutes, but the after-work rush builds a line of about 61 people around 6:10 PM and pushes a few voters just past 30 minutes: SOME WAITS, with the booths as the bottleneck. Best time to go is 9:30–10:00 AM. Fix it says one more booth brings the longest wait down to about 19 minutes.
  • Load College campus. Slow same-day-registration check-ins meet a late-afternoon surge, so about 764 people are in line when polls close and the last voter leaves around 10:21 PM: MELTDOWN.
  • Long midterm ballot takes 8 minutes to mark, so the booths fill and about 1 in 4 voters waits more than 30 minutes (LONG LINES). Three more booths fix it. Rural township hall comes out MANAGEABLE and Early-vote county comes out SMOOTH.
  • Press Reroll day to replay the same setup with different random arrivals. Then try trading booths for check-in stations and watch the bottleneck chip move.

Why this (vs neighbors)

Election Day is Tuesday, Nov 3, 13 days out, and "when should I go vote, and why was the line so long last time?" has a real, visual answer. Walk-in Wait Board handles a single business queue and Bus Bunch shows how headways fall apart. Poll Line chains three stations, so you can see a bottleneck move from check-in to the booths as you add staff. It also has a public-interest benchmark to aim for: the 2014 Presidential Commission on Election Administration recommended that no voter wait more than 30 minutes.

Scope trim

No real precinct data or polling-place lookup, no provisional-ballot or accessibility-station lanes, no voters who leave the line, no multi-precinct sites, and no share URL. One room, one day, three stations.

Limitations

  • Arrival patterns are hand-shaped curves (sums of bell curves plus a small crowd waiting at open), not fitted to real turnout logs. Service times are randomized around your averages (about 0.45× plus an exponential tail), so a single day is a noisy sample. Reroll a few times.
  • Queues are strictly first-come, first-served with no balking, line-jumping, or broken equipment. Real lines can be worse (a jammed scanner) or better (poll workers moving to wherever the line is).
  • Verdict rules: MELTDOWN means a longest wait of 90 minutes or more, or a 90th percentile of 60 minutes or more. LONG LINES means at least 10% of voters wait more than 30 minutes, or the 90th percentile is above 30. SOME WAITS means anyone waits more than 30 minutes. MANAGEABLE means a longest wait of 10–30 minutes, and SMOOTH means under 10.

Suggested next steps

  1. Fit the arrival curve to a county's published check-in timestamps and offer one-click presets by precinct.
  2. Add voters who give up after a set wait, so the tool can count lost votes as well as minutes.
  3. Add a staffing budget mode: given N extra poll workers, put them where they cut the most wait-minutes.