21–30 of 181 posts
this is the "Digital Twins" story right here - private corporate walled gardens with total control and feeding off of commons.
That's a curious use of an already established term, "Digital Twin"[0]. A digital twin is a virtual model designed to accurately reflect a physical object. 0: https://en.wikipedia.org/wiki/Digital_twin
> Ninety-eight percent of drivers who try Lyft’s maps stick with them — rather than switching to Google, Apple, Waze, or any other app. I find this surprising because my personal experience is the exact opposite. At least in NYC, drivers use the Lyft (or Uber) app for managing riders but use a dedicated mapping application for navigating to the destination once you're picked up.
First, the focus is on savings to Lyft, and not how this innovation benefits customers in any way.
Second, it is highly antagonistic to Google - who is probably still a partner to Lyft in many other aspects (advertising etc). Taking shots at your partners is generally considered bad for business.
> By late 2019, Lorimer and Justin Moore, then the head of engineering, were ready to form a team to “do the impossible” — create a home-grown mapping system. THE IMPOSSIBLE: A "home-grown" mapping system built on OpenStreetMaps? Lyft seems to really lack any imagination if _that_ was considered impossible.
It was impossible a few years ago, when the OSM data wasn't good enough and the cost of acquiring the data themselves was too high.
Then again, Apple Maps started getting sorted out around then too iirc.
Does anyone know if Lyft plan to contribute their map corrections back to OSM? They mention other companies doing so in this blog post, but don't make it clear whether or not they, too, intend to. I'm curious to see how good Lyft Maps is, too. Uber Maps is, frankly, a disaster in my area, frequently suggesting illegal and/or impossible turns. I try to submit map feedback every trip (very hard as a rider, as most form…
I've submitted a correction for an illegal turn instruction twice on Google maps. No reply. But my other corrections have been accepted.
Maps has been losing money for literally decades. I don't know if they've finally turned that around or not. In some parts of the world, they were still paying a 3rd party per lookup because they didn't own the data.
I also don't know the numbers for their API charges, but I wouldn't be surprised if the totals are mostly symbolic, for VP's to defend themselves as "Hey, look, we're getting all we can out of it!"
Does anyone know if Lyft plan to contribute their map corrections back to OSM? They mention other companies doing so in this blog post, but don't make it clear whether or not they, too, intend to. I'm curious to see how good Lyft Maps is, too. Uber Maps is, frankly, a disaster in my area, frequently suggesting illegal and/or impossible turns. I try to submit map feedback every trip (very hard as a rider, as most form…
I also find it consistently obnoxious that there is essentially no way to say "My driver was fine, but your platform did something stupid" when leaving feedback. At least, not without jumping through a million hoops. The arrogance of our industry is tiresome. I took a ride in SF a while ago and it predicted a 3 minute pick up time which turned out to be 25 minutes - no fault of the driver, and the only way I could co…
> Ninety-eight percent of drivers who try Lyft’s maps stick with them — rather than switching to Google, Apple, Waze, or any other app. I find this surprising because my personal experience is the exact opposite. At least in NYC, drivers use the Lyft (or Uber) app for managing riders but use a dedicated mapping application for navigating to the destination once you're picked up.
> By late 2019, Lorimer and Justin Moore, then the head of engineering, were ready to form a team to “do the impossible” — create a home-grown mapping system. THE IMPOSSIBLE: A "home-grown" mapping system built on OpenStreetMaps? Lyft seems to really lack any imagination if _that_ was considered impossible.
It was impossible a few years ago, when the OSM data wasn't good enough and the cost of acquiring the data themselves was too high.