#45 · How does Google Maps know the road is red?
99 phones in a handcart made empty Berlin streets turn red. Phones as traffic sensors, road segments, historical patterns, and stream aggregation — the pattern behind live traffic.
Ek line yaad rakho: Maps ne sadak nahi dekhi. Usne phones gine.
The story
In February 2020 the artist Simon Weckert pulled a handcart with 99 secondhand smartphones, all running Google Maps, down empty streets in Berlin. Maps showed those streets as a traffic jam (red) and rerouted drivers (The Register, 3 Feb 2020).
How live traffic works
- Phones are the sensors. Google says "aggregate location data can be used to understand traffic conditions on roads all over the world" — phones with location on report where they are and how fast they move, anonymously (Google, "Google Maps 101", 3 Sep 2020).
- Roads are split into segments. Each short stretch gets the average speed of the phones on it right now.
- Speed → colour. Normal speed = green, slow = orange, very slow = red.
- ETA uses history too. Maps combines live conditions with historical traffic patterns for that road and time; Google says its ETAs have been accurate for over 97% of trips.
99 phones moving at walking pace on one segment look exactly like 99 cars stuck in a jam — the average speed drops, so the road turns red.
The interview answer: stream aggregation
- Partition by key (road segment id / geohash) so all events for one segment land on the same worker.
- Windowed aggregation (tumbling or sliding window of a few seconds to a minute): average / median speed, count of devices.
- Guard against weird inputs: minimum distinct devices, outlier filtering, compare with the historical baseline — the Weckert cart is exactly the case a count-only rule misses.
- Serve from a fast store (segment → current speed) that the map tiles and the routing/ETA service read.
Words to say: stream processing, partition by segment, time window, aggregation, historical baseline.
Sources: The Register (3 Feb 2020); Google blog "Google Maps 101: How AI helps predict traffic and determine routes" (3 Sep 2020). Segment speeds in the reel are illustrations.
Go deeper
Finished reading?
Mark it done to track your progress in Reels.
#46 · How does Uber find the nearest car so fast?
Thousands of cars, and your cab in two seconds. Uber's H3 hexagon grid, k-ring search, and the geospatial index pattern behind "find nearby".
#44 · How does your phone keyboard really work?
You type "kal" and the keyboard already offers "milte". Touch decoding, an on-device language model, federated learning and the typeahead pattern (trie + top-k per prefix) interviewers ask about.