Re:Waze Hijacked L.A. in the Name of Convenience
by enoch_r on Monday September 23, 2019 @10:25PM
(#21055891)
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.
[ Parent
| on HN ]
Re:Waze Hijacked L.A. in the Name of Convenience
by JMTQp8lwXL on Monday September 23, 2019 @10:34PM
(#21055943)
It 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.
Re:Waze Hijacked L.A. in the Name of Convenience
by nullc on Tuesday September 24, 2019 @12:42AM
(#21056631)
From 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.