Tamelabo 日本語

What If, Calculated#01

What If a City Had No Red Lights? We Simulated 2 Towns | What If, Calculated #01

Short answer

  1. It depends on the town: with no crashes, the countryside got 24% faster while the big city just tied with normal signals.
  2. Add a rule that one in twenty near-misses becomes a crash, and the big city sees 27 crashes with only a third of its cars arriving.
  3. Even the countryside had 2 crashes and ended up slower — signals turned out to be a crowded city's safety device.

Video

Pressing play loads the YouTube player on this page. Watch on YouTube

Try it

The lab bench for this episode. It runs right here on the page.

Open full screen

Explained

What if every traffic light stayed green — would a town actually get faster? (all-green intersections · what-if traffic simulation) We built two fictional towns, a big city and a countryside town, and calculated what happens when no light ever turns red.

This episode's page (run the simulation yourself, explanations, sources): haccha-meccha.com/en/moshimo/01

The two towns are fictional, built for this calculation: a "big city" (1 km², 25 intersections, 900 cars leaving in a 5-minute morning rush, 24 cars per km of road) and the "countryside" (a 1.5 km square, 9 intersections, 100 cars, 2 cars per km of road).

The answer depended completely on the town. With cars that never collide, the countryside got 24% faster, while the big city came out almost tied with normal signals. Then we added a rule that one in twenty near-misses becomes a real crash: the big city had 27 crashes, only a third of the cars arrived, and even the cars that did arrive took 22% longer than with signals. Under the same rule, the countryside had 2 crashes and also ended up slower than with signals (+76%). Traffic signals turned out to be the safety device of a crowded city.

How we calculated: Fictional towns. First we compared everything with cars that never collide. At an intersection, the car that came first goes first; a car that stopped takes a 2-second look before moving off (based on research showing a start-up loss of about 2 seconds). Then we ran it again with crashes: one in twenty near-misses (moments that need hard braking) becomes a crash (from Bird's ratio in workplace safety). A crashed intersection is closed for 10 minutes and cars detour (the 10 minutes is our own setting). Even on green, a single lane can only send about one car every 2 seconds through an intersection (a basic traffic-engineering value). Cars drive on the left, as in Japan. Each number is the middle value of 3 runs with different random seeds.

More from What If, Calculated: youtube.com/playlist

Subscribe: youtube.com/@tamelabo_jp

Shorts

Sources

Where the facts and numbers in the video and on this page come from.

  1. University of Tokyo IIS (start-up lost time)transport.iis.u-tokyo.ac.jp/poster/031-060/46.pdf
  2. MHLW Japan & KEYENCE (near-miss-to-accident ratios)keyence.co.jp/ss/general/manufacture-tips/heinrich-la…
  3. SHRP2 naturalistic driving data (USA)transportationops.org/Bigdata/NDS
  4. National Police Agency, Japan (traffic signal stock)npa.go.jp/bureau/traffic/seibi2/annzen-shisetu/hy…
  5. Shinsei Kogyo (Traffic Signal Day, 1931)shinseibt.co.jp
  6. Tokyo Metropolitan Government (daytime population)toukei.metro.tokyo.lg.jp/tyukanj/2020/tj20gaiyou.pdf
  7. Statistics Bureau of Japan (2020 census, place of work/school)stat.go.jp/data/kokusei/2020/kekka/pdf/outline_04.…
  8. CiNii (start-up lost time and large vehicles)cir.nii.ac.jp/crid/1390305488974376576
  9. MHLW Japan (Heinrich's law)anzeninfo.mhlw.go.jp/yougo/yougo24_1.html
  10. City of Kawasaki (intersection traffic counts)city.kawasaki.jp/300/cmsfiles/contents/0000177/177750/11…
  11. MLIT Japan (Road Traffic Census)mlit.go.jp/road/census/h22-1/data/kasyorep.pdf
  12. Google (Project Green Light)sites.research.google/gr/greenlight
  13. Miovision (Pittsburgh)miovision.com/rapidflow
  14. Agency for Natural Resources and Energy, Japan (2018 Hokkaido blackout)enecho.meti.go.jp/about/special/johoteikyo/blackout.html
  15. Cabinet Office, Japan (White Paper on Disaster Management 2019)bousai.go.jp/kaigirep/hakusho/h31/honbun/0b_1s_01_04…
  16. National Police Agency, Japan (White Paper on Police 2019)npa.go.jp/hakusyo/r01/pdf/03_tokushu.pdf

Voice, music, art

  • Music: “楽しく軽快に” by yuki, “考え中” and “突破口の手がかり的なBGM” by 鷹尾まさき, “Shall we meet?” by MATSU (OpenTracks). Some sound effects: Freesound (CC0), Kenney (CC0). Fonts: Zen Maru Gothic, Inter, IBM Plex Mono (SIL Open Font License).

More

More from What If, Calculated

New in other series

Subscribe on YouTube