4 ms·
This is true for freeways and possibly multi lane arterials. It's absolutely not true for small neighborhood streets, which seem to be the shortcuts at issue he
by enoch_r 7y ago
This is true for freeways and possibly multi lane arterials. It's absolutely not true for small neighborhood streets, which seem to be the shortcuts at issue here.
- JMTQp8lwXL 7y agoIt seems improbable, though, that Waze is assuming you're doing 65 mph on residential streets (which, presumably, are around 25mph legal limit) to save time. It's faster to take residential streets if the freeways are completely backed up to the point they're moving slower.
- nullc 7y agoFrom the article: “If it’s a single-lane residential street and the speed limit is 15 mph, but people are speeding through at 35, they will model the time off that data.” The article also appears to state that it doesn't model the load dependent throughput-- 'The instant the time penalties work out,” DwarfLord claims, “Waze will just as happily send a thousand Wazers down'. Which isn't surprising, since routing a lot of cars independently is computationally tractable but adding capacities to the problem makes it NP-complete in theory and much harder in practice.