That's even worse. He took a simple side-by-side comparison of 2 or 3 images, and invested substantial effort into making it far more annoying and harder to compare by an uncontrollable permanent headache inducing slow flicker which is inferior to our built-in saccades. I hate the GIFs on his pages, they're by far the worst thing on his site and why I don't reshare his otherwise very interesting analyses, because they're a crime against the reader.
It works well for some of the comparisons, but for others I would really prefer static shots so I can take my time examining each. Especially the GIFs that compare more than two sources, such as the ones that include Here and Bing as well as Google and Apple maps, where waiting for the GIF to cycle through all four images is annoying.
If the author is willing to implement it, those things some websites use that let the reader click the image or drag a sliding bar over it to compare might work better.
Regardless, I do appreciate the author's insights.
No, it doesn't. When you saccade normally, you can do so in milliseconds (it's the fastest movement your body is capable of making and is literally subconscious) repeatedly and freely. The GIF yanks away control, forcing you to wait through the slow arbitrary preset cycle, while distracting you reading anything else. And that's in the good case of just 2 frames, when it's 3 or 4, it's much worse, because now you have to sit through the cycle possibly several times just to figure out what each frame is labeled and only then can you begin comparing! Imagine how much more pleasant and easy to read and accessible and less headache-inducing this article would be if it followed a systematic approach of presenting 1-4 images side by side, Apple then Google then misc.
Very interesting. Anecdotally, I've found the new Maps to be much better in its level of detail–at least from the brief time I was able to use it this summer–but from the article it looks like most of that might just be things to make the map look pretty? It seems like Apple is extracting shapes from satellite imagery to make their maps look better, but failing to include actual business and place information.
It's weird how out of date data in providers like Tom Tom is. I just pulled up the GIS data from Monterey County and it clearly has the same roads that the new Apple Maps shows for Parkfield. (I can't check Markleeville, as Alpine county charges a fee for their GIS data)
My personal favourite was when I lived in Hackney in London, and there was a business listing in Apple Maps (one of the few!) for a garage business that had closed down in 1975, a year before Apple was even formed. That one was pretty special.
Last year's article from the same author [1] about Goggle Maps' use of photogrammetry and other building scanning techiques was, in my opinion, one of the most interesting HN submissions ever (its comment section[2] is also worth a read).
He's written multiple articles about maps, even before that one: https://www.justinobeirne.com. He's very good at spotting details in maps–IIRC he worked on Apple's maps team before, so this isn't particularly surprising.
Sadly he took them down, disappeared into the memory hole the way so many Apple employee's work has a way of doing. I'm glad he's out and writing in public again.
For some reason I thought he is some random guy just passionate about maps, I was actually looking for donate me buttons on the site because I love reading his articles whenever they are posted to HN.
Complaining gets results. I'm a total believer in negativity. You could say I'm positive about it.
What I mean by "relentless" is that this article in the OP covers the changes in detail for the first half, but the second is the same content as his last piece:
But while this one has a lot of specific speculation, the Google article is about how cool they are, links to PR and basically says "look at all those buildings, they must have done it with computers". I suspect someone who worked at Google wouldn't have said that.
"...it’s a dramatically different map from before, with a staggering amount of vegetation detail"
"...what’s really remarkable about this new vegetation detail it how deep it all goes—all the way down to the strips of grass and vegetation between roads"
"...some of these upgraded buildings are spectacularly detailed"
"...Apple is filling its map with so many of them that Google now looks empty in comparison"
"...Apple hasn’t just closed the gap with Google—but has, in many ways, exceeded it"
Maybe not! The last half of the article is all about the faults; the closing line of the article (before footnotes) questions whether or not Apple is making the right maps at all.
And the footnotes are relentlessly negative about Apple:
"These building height regressions are surprising because they contradict TechCrunch’s claim that Apple’s buildings are now 'more accurate'."
"Consider that just two years after it started adding algorithmically extracted buildings to its map, Google had already added the majority of the U.S.’s buildings. But after four years, Apple has only added buildings in 64% of California and 9% of Nevada."
"All of this new detail is not without cost. In many areas, Apple Maps’s roads are now harder to see than before."
"Part of the reason why Yelp’s place database is so much smaller than Google’s is because Yelp is largely focused on businesses with consumer-facing storefronts. And you can see the consequences of this on Apple’s map, especially with government-related places."
"Or maybe the issue is that Apple’s extraction algorithms just aren’t as good as Google’s yet?"
"Another advantage of the Local Guides program is that Google owns everything that’s contributed, including all of the photos."
"For instance, here’s the Six Flags Great America theme park that’s just seven miles away from Apple’s headquarters." (With a shot of Great America lacking detail in Apple's map.)
"It’s odd that Apple refuses to track trip start/end points but sees nothing wrong with mapping tennis and baskball courts in people’s backyards."
"It’s almost as if Google is saying this is now a map of destinations—all of the places it’ll be able to take you to, someday soon."
"I think Google’s ambitions here run far deeper than being just another Yelp or Foursquare. If you zoom out on everything Google is doing, you see the makings of a much larger, end-to-end travel platform."
It's kind of interesting -- it's as if halfway through he forgot that he was writing about improvements in Apple's mapping and decided he was writing about the promise of Google's mapping as a new kind of platform.
But either way, this is a classic rhetoric. It's the bit where you praise a specific thing to the skies, then pull back and reveal how in the big picture that thing doesn't matter so much. Note that I'm not saying he's wrong. I'm just saying that this is not, over all, an article that praises Apple.
Just to be clear, as far as I can tell from Google, the author actually works for Apple [1] on Apple Maps, at least at some point. So this blog may serve as an advertisement of his own work, despite the fact that it seems to be written in the voice of an observer.
"Justin O'Beirne designed and led the development of Apple’s cartography effort, and is the author of Maps for the Masses, an upcoming book about digital cartography." [2]
"led the development" pretty much implies he's not working there anymore.
Reading the article it's also pretty clear he doesn't have any insider-insights, particularly with the speculation about building shapes and how Apple supposedly outsourced that, as a manual task, to a couple of thousand Indians.
I doubt an Apple insider would be too keen on speculations like that because if it turns out to be true it would be kind of bonkers.
Except that the new map makes it seem as though California is some lush, green, tropical land. I think the green is way over-emphasized on the new map compared to the old, which is a bit more realistic looking to my eyes.
In the spring maybe. For much of the year it's brown because it's desert, nonnative grassland, or city. Big chunks are forest, especially in the north, but brown is still dominant.
Sure, but brown areas on the map marks different types of areas, so they obviously can't use brown. Imagine how much of the world would look in the winter, it would just be a completely white map.
The colors of the map obviously can't be based on the season.
The arbitrary transition from green to brown on the old maps, when in reality, there is no such transition, is the real problem. When navigating by foot those shapes SHOULD be accurate; I care less about the legal boundaries of the exact park and more about what i'm seeing as I walk around.
Why do you care that the map represents the green parts accurately? How does that enhance your walking experience?
IMO 99% of the time people use maps to get to destinations, so maps should be focused on destinations, not shapes (insofar as the shapes don't prevent you from getting to your destination).
The shape of e.g. a wooded area can be important for orientation when on foot.
But really, maps are nothing but shapes (streets, waterways, coastlines) that constrain navigation on the ground, so it's difficult to make sense of your proposed focus on destinations to the exclusion of shapes.
My scout leader always taught me "never navigate by woods". The reason? It doesn't take long to chop down a forest and invalidate your assumptions.
I remember taking part in 2-day team orienteering competition. We had walked very accurately on a bearing to where the checkpoint should be, but it wasn't there. My friend had taken the bearing on the corner of a wooded area. Looking more closely at the wooded area, we could see a large area of tree stumps next to it...
(We reorientated ourselves, found the checkpoint, and came second in the competition.)
But that's what I don't like about the apple map. Check out the image where he says:
> Regardless of how Apple is creating all of its buildings and other shapes, Apple is filling its map with so many of them that Google now looks empty in comparison:
Google does have vegetation and desert data, but you can see that it starts to fade around zoom level 5 in the US (at which you can only see a number of entire states and the names of major cities).
If you go somewhere else on the planet, e.g. along the Amazon river, you see that the vegetation data reappears, only to start fading again at zoom level 11 or so. It's too irregular and following satellite imagery too closely to have been drawn from human sources such as parcel data. It must have been built from imagery.
So it looks like it's a deliberate choice on Google's part to only use human-derived features such as a park's outline at most zoom levels, on most of the inhabited surface of the planet.
I used to work for Google, but not on maps. You can see the zoom level for yourself in the URL. Use the scrollwheel or its trackpad equivalent and the Z parameter in the address bar will change. That's where I got the numbers, not any special knowledge.
> it shows me where there's a public park I might visit
Google's map shows a relatively tiny bit of land that it labels as the regional park.
Apple Maps, meanwhile, highlights a much larger area that it gives the same label.
Based on the fact that Apple Maps is explicitly highlighting this area, I'm going to guess that it's correct and that entire highlighted area is, in fact, the regional park. So if you wanted to visit Garland Ranch Regional Park, wouldn't you prefer it to be accurately mapped, instead of represented as a tiny slice of green?
Or if you meant that tiny bit of green in Caramel Valley, Apple Maps has it too, it just has a lot more green too that might be interesting if your goal is to visit bits of greenery stuck in the middle of towns.
It's niche, but I could see this feature being useful to me. This is relevant when using maps to scout outdoor locations for whatever activity it is. I use maps to find public bodies of water to go fishing along the lake shore. Getting an idea of the vegetation around a lake before zooming in and looking at the satellite data (which is mentally tiring to me) is very nice. I find myself preferring Apple's Maps app to the Google Maps app on iOS even if Google has better maps in its app anyway, so I really welcome these improvements.
The first couple paragraphs and images led me to believe that the article was satire, and was explaining that everything in Apple Maps was gone except for an extremely detailed map of California. Then I started wondering if it wasn't satire.
It's a long read end to end, but I found that the author is consistently level-headed in his criticisms and praise. I do not think he was being sarcastic.
Seems like an odd conceit for Apple to make, particularly since they've been pushing the accuracy of their 3D maps, and since you can easily rotate the map in 3D to get a full view of everything. My guess is that their algorithm to calculate the height of buildings from satellite imagery messed up.
No, because if you view in 3D view using satellite mode, the buildings all look the correct height, indicating that the algorithm can work correctly. It seems to be specific to the 3D non-satellite (what they call "map") view. Which suggests it might be an intentional design decision to try to reduce masking. Very weird.
I thought about that too, but in these same examples, other buildings were made taller, making it more difficult to see what's behind them, so that can't be it.
>In other words, TomTom’s database somehow has roads from Parkfield’s boomtown days—roads that have been gone for more than 75 years. No wonder why Apple removed them.
Another possible explanation is the TomTom was using these unlikely to be visited streets as trap streets.
>In cartography, a trap street is a fictitious entry in the form of a misrepresented street on a map, often outside the area the map nominally covers, for the purpose of "trapping" potential copyright violators of the map who, if caught, would be unable to explain the inclusion of the "trap street" on their map as innocent. On maps that are not of streets, other "copyright trap" features may be inserted or altered for the same purpose.
Using streets from maps that predate TomTom’s maps defeats the purpose of a trap street. The plagiariser could just claim to have sourced the data from old, public domain, sources.
True, but that is assuming that those streets ever existed. As far as I can tell from the article that's just an assumption.
Although I think it would actually be possible to use once-real-but-no-longer steets as trap streets, as long as your sources are improbable enough. For example, if you take a map from 1908, remove one street, then add in another single street from 1920, etc., your exact combination of streets could serve as proof of copyright infringement.
I hope they have some manner of “shearing off” the nodes that represent those streets and break the graph into two disjoint ones. That way they’re off on their own little island without a ferry connection and problems only arise if you ask for a route to, or are starting from within, somewhere that actually doesn’t have any ground truth.
In the accompanying aerial photograph, the outlines of the old streets are visible - together with a couple of others that do not appear on either map, and which run at an angle to the old grid.
Before Apple Maps added transit mode and hand-edited all the stops, it was displaying half the Caltrain stations as "Southern Pacific". The data sources are more than happy to give you data from 1940.
I think this article illustrates a critical issue with Apple's software. The stance on privacy means little machine learning can take place. Little machine learning means that Apple has to resort to more manual techniques. This, in turn, leads to the far inferior (and often dangerously incorrect) place databases and issues with their Maps application.
And then users will simply turn to an alternative which absolutely does not care about their privacy, undoing all of Apple's effort in protecting their users' privacy.
Apple needs to strike a balance between protecting users' privacy and performing analytics. Perhaps send the data off-shore to a location not under US jurisdiction, I don't know. But it is clear that Apple cannot keep up with its competitors with its current practices.
I think they do a pretty good job with the little data they do collect. For example, I almost exclusively use Apple Maps–it's not like it's unusable. It might help that all my navigation needs fall inside of California, but at least here I haven't had many issues.
It's incredibly detailed in pointing out the differences between old and new Apple Maps, as well as comparing map images from Apple, Google, Tom Tom, and others.
The article demonstrates through the use of animated gifs that Apple Maps has become significantly better over time and in many cases surpasses the quality of competitors.
> The article demonstrates through the use of animated gifs that Apple Maps has become significantly better over time and in many cases surpasses the quality of competitors.
Except for actually finding & navigating to places, where it's somehow getting worse rather than better.
Vegetation & building outlines are the primary aspects Apple Maps is improving upon per the article, whereas places are getting worse.
Two years ago Apple released information on its website about "differential privacy", which is a branch of data analysis that deals with collecting information for analysis that cannot be tied directly back to an individual person.
How can violating someone's privacy improve their place database? Extract "I'm waiting at <restaurant name> now" from people's iMessage chats???
Google has been connecting place names by OCRing storefronts taken with their street view cars. Apple claim to be doing the same thing (but apparently doing a far worse job of it)
If you look up somewhere in Google Maps and then go there, with your phone in your pocket, Google knows where you went. It also knows how long you stayed there.
Now they know exactly where it is, when people like to go there, and if the opening hours seem accurate.
Not limited to them of course, this is also check-in companies' business model, like Foursquare.
The balance they are striking is that they are collecting trip information, but anonymizing it by only sending segments of a given trip and not the whole thing (as well as by not associating trip data with the user).
The workaround will probably be that they end up buying the data from a provider who doesn't care much about privacy.
How they don't yet have a web presence boggles my mind though. I send in small corrections to Google Maps all the time when I notice errors, Apple Maps won't let me do that.
I cannot fail to notice how many # of ads, notifications, various asks, pop-ups, disclaimers, endorsements, contact buttons or share buttons there are on this page compared to what we see in any other piece of content with as much detailed information.
The author definitely enjoys compiling these amazing essays and share this knowledge.
Guessing that the methods and output of human generated maps are used as labelled training data that will be used to scale out when they apply ML to other areas.
Apple is really stepping up and showing Google it has some power. When they broke the map relationship with google in 2012 I was super skeptical, as were many. But in a very apple-like way - it's starting to come together.
I still use Google for nav but I have a feeling Apple might win me over on this one.
Funny that the author mentions the Markleeville Courthouse as being across the street from the General Store, when it is, in fact, next door, as shown in the picture. I'm sure plenty of us on HN have cycled through Markleeville a couple of times in mid-July...
I've always found Google vs Apple maps discussions to be interesting, as I've always vastly preferred Apple Maps data for walking around places; Google Maps too often had poor building shapes that looked nothing like the real thing, or hid business names at the scale I was trying to use. Perhaps that's because I live in the Bay Area. When I'm overseas, in particular, I don't hesitate to go to Google Maps first.
> And the office’s large headcount (now near 5,000) suggests some sort of manual / labor-intensive process.
My partner briefly worked for a human-powered 3d mapping firm; they would get satellite and plane/drone photography of a large swath of land, split it up into block-sized chunks, and then each worker would take a block and use an in-house program to model the buildings at a pretty impressive level of detail. Workers got paid per-block and blocks were priced based on their complexity. They've been doing this for over 10 years by this point, so it's not an entirely unknown or uncommon thing to handle this kind of work manually.
Is this even sustainable in a "World" Scale? Because at this rate I don't see this brings Apple to cover all the major cities of world in 10 years time, let alone majority of lands which aren't in these locations.
That's your use case. I've used maps a lot to try to find hidden or secluded parks -- often with great success.
I'm talking postcard-perfect views around the Sydney harbor, small beaches around various bays, isolated parks near beautiful rivers and a whole lot of other places that are really beautiful and probably only known to the locals. Lots of these places were a brisk walk away from high-traffic areas too.
This is not to dismiss your question on the directions, though, which is certainly an important part of using maps on a phone.
I am being overly crabby. I do appreciate the author's enthusiasm for maps. But still, I stick to Google Maps because Apple Maps has a reputation for giving terrible directions. That's where the Google Maps moat is. The directions are generally good and will route you around traffic, which is crucial.
That's an interesting point, as all maps are (by definition) only representations, invariably leaving something out - the question is what to leave in the map.
I have a knack for finding good driving roads with Google Maps.
It doesn’t really work in satellite view because the trees make for a ton of visual noise at the relatively low resolutions shown.
A good indicator for me is green in the map, and no street view (street view tends to cover busier routes, I drive roads street view wouldn’t bother with)
I don't know if it has been updated since, but literally two months ago it could not get you from Custer State Park to Mount Rushmore. It frequently asked me to go in the complete wrong direction. I can't trust it when it can't connect two popular areas that are relatively nearby.
That said, it was fine for anything that was on major highways.
In the area Apple's new maps are in, in my experience (important to underline that): yes, they do. I've been using these new maps since they appeared in the iOS 12 beta, so close to half a year, and they're pretty good. And, yes, to a comment you made below this one, Apple Maps is getting good at handling traffic. I find them equal to Google, although not as good as Waze. (Yes, I know Google owns Waze, but I'm standing by this. Waze does better than Google Maps here.)
As the linked blog post shows, they're definitely not perfect; I haven't encountered anything particularly egregious, but the fun thing about maps is that virtually everyone will run into what are ostensibly "edge cases" eventually. Apple Maps has steered me wrong before, but so has Google and Waze. (Waze has gotten much better at destinations in the last few years, after their purchase by Google, although they still have a disquieting penchant for "shortcut" directions that include "make this uncontrolled left turn across three lanes of traffic." Not only does Apple rarely do that, Google rarely does that.)
I find vegetation massively helpful, because I have a pretty good idea of the shape of the forest areas around my city (both from driving around and also from looking at satellite/aerial photos). I can actually find a lot of places faster looking at the shape of the forest, parks, etc. than by either looking at the streets or even searching!
It's just making the map data match the real world better, and making it useful for more than just driving.
It depends greatly on use case. When I’m hiking somewhere “in the country”, it’s of tremendous help to see every tree (and even what type) and shed on a map as it helps me with orientation.
376 comments
[ 4.0 ms ] story [ 326 ms ] threadAll in all, what I've learnt is how many petabytes must be involved in mapping the US and the world and the 3D aspects.
If the author is willing to implement it, those things some websites use that let the reader click the image or drag a sliding bar over it to compare might work better.
Regardless, I do appreciate the author's insights.
Last year's article from the same author [1] about Goggle Maps' use of photogrammetry and other building scanning techiques was, in my opinion, one of the most interesting HN submissions ever (its comment section[2] is also worth a read).
[1] https://www.justinobeirne.com/google-maps-moat
[2] https://news.ycombinator.com/item?id=15965653
Sadly he took them down, disappeared into the memory hole the way so many Apple employee's work has a way of doing. I'm glad he's out and writing in public again.
https://archive.is/Fnw4W https://archive.is/AHmEF https://archive.is/WpOCJ
For some reason I thought he is some random guy just passionate about maps, I was actually looking for donate me buttons on the site because I love reading his articles whenever they are posted to HN.
Nobody else would know what to look for.
https://daringfireball.net/linked/2018/09/07/eddy-cue-the-in...
Complaining gets results. I'm a total believer in negativity. You could say I'm positive about it.
What I mean by "relentless" is that this article in the OP covers the changes in detail for the first half, but the second is the same content as his last piece:
https://www.justinobeirne.com/google-maps-moat
and quite like the one before that:
https://www.justinobeirne.com/a-year-of-google-maps-and-appl...
But while this one has a lot of specific speculation, the Google article is about how cool they are, links to PR and basically says "look at all those buildings, they must have done it with computers". I suspect someone who worked at Google wouldn't have said that.
"...it’s a dramatically different map from before, with a staggering amount of vegetation detail"
"...what’s really remarkable about this new vegetation detail it how deep it all goes—all the way down to the strips of grass and vegetation between roads"
"...some of these upgraded buildings are spectacularly detailed"
"...Apple is filling its map with so many of them that Google now looks empty in comparison"
"...Apple hasn’t just closed the gap with Google—but has, in many ways, exceeded it"
And the footnotes are relentlessly negative about Apple:
"These building height regressions are surprising because they contradict TechCrunch’s claim that Apple’s buildings are now 'more accurate'."
"Consider that just two years after it started adding algorithmically extracted buildings to its map, Google had already added the majority of the U.S.’s buildings. But after four years, Apple has only added buildings in 64% of California and 9% of Nevada."
"All of this new detail is not without cost. In many areas, Apple Maps’s roads are now harder to see than before."
"Part of the reason why Yelp’s place database is so much smaller than Google’s is because Yelp is largely focused on businesses with consumer-facing storefronts. And you can see the consequences of this on Apple’s map, especially with government-related places."
"Or maybe the issue is that Apple’s extraction algorithms just aren’t as good as Google’s yet?"
"Another advantage of the Local Guides program is that Google owns everything that’s contributed, including all of the photos."
"For instance, here’s the Six Flags Great America theme park that’s just seven miles away from Apple’s headquarters." (With a shot of Great America lacking detail in Apple's map.)
"It’s odd that Apple refuses to track trip start/end points but sees nothing wrong with mapping tennis and baskball courts in people’s backyards."
"It’s almost as if Google is saying this is now a map of destinations—all of the places it’ll be able to take you to, someday soon."
"I think Google’s ambitions here run far deeper than being just another Yelp or Foursquare. If you zoom out on everything Google is doing, you see the makings of a much larger, end-to-end travel platform."
It's kind of interesting -- it's as if halfway through he forgot that he was writing about improvements in Apple's mapping and decided he was writing about the promise of Google's mapping as a new kind of platform.
But either way, this is a classic rhetoric. It's the bit where you praise a specific thing to the skies, then pull back and reveal how in the big picture that thing doesn't matter so much. Note that I'm not saying he's wrong. I'm just saying that this is not, over all, an article that praises Apple.
"Justin O'Beirne designed and led the development of Apple’s cartography effort, and is the author of Maps for the Masses, an upcoming book about digital cartography." [2]
[1] https://www.crunchbase.com/person/justin-o-beirne
[2] https://qz.com/author/justinobeirne/
Reading the article it's also pretty clear he doesn't have any insider-insights, particularly with the speculation about building shapes and how Apple supposedly outsourced that, as a manual task, to a couple of thousand Indians.
I doubt an Apple insider would be too keen on speculations like that because if it turns out to be true it would be kind of bonkers.
The colors of the map obviously can't be based on the season.
IMO 99% of the time people use maps to get to destinations, so maps should be focused on destinations, not shapes (insofar as the shapes don't prevent you from getting to your destination).
But really, maps are nothing but shapes (streets, waterways, coastlines) that constrain navigation on the ground, so it's difficult to make sense of your proposed focus on destinations to the exclusion of shapes.
I remember taking part in 2-day team orienteering competition. We had walked very accurately on a bearing to where the checkpoint should be, but it wasn't there. My friend had taken the bearing on the corner of a wooded area. Looking more closely at the wooded area, we could see a large area of tree stumps next to it...
(We reorientated ourselves, found the checkpoint, and came second in the competition.)
> Regardless of how Apple is creating all of its buildings and other shapes, Apple is filling its map with so many of them that Google now looks empty in comparison:
https://static1.squarespace.com/static/54ff63f0e4b0bafce6932...
Yeah, the google view has a lot less green, but it's really a different kind of green -- it shows me where there's a public park I might visit.
Apple's green is showing vegetation, but finding that regional park gets lost in the noise.
If I wanted to find a place to visit, Google is the superior layout. If I want to get a sense of vegetation, I would just turn on satellite view!
If you go somewhere else on the planet, e.g. along the Amazon river, you see that the vegetation data reappears, only to start fading again at zoom level 11 or so. It's too irregular and following satellite imagery too closely to have been drawn from human sources such as parcel data. It must have been built from imagery.
So it looks like it's a deliberate choice on Google's part to only use human-derived features such as a park's outline at most zoom levels, on most of the inhabited surface of the planet.
Google's map shows a relatively tiny bit of land that it labels as the regional park.
Apple Maps, meanwhile, highlights a much larger area that it gives the same label.
Based on the fact that Apple Maps is explicitly highlighting this area, I'm going to guess that it's correct and that entire highlighted area is, in fact, the regional park. So if you wanted to visit Garland Ranch Regional Park, wouldn't you prefer it to be accurately mapped, instead of represented as a tiny slice of green?
Or if you meant that tiny bit of green in Caramel Valley, Apple Maps has it too, it just has a lot more green too that might be interesting if your goal is to visit bits of greenery stuck in the middle of towns.
In the apple layout, it gets lost in the noise of private-land vegetation.
It's the thought that they just got in their cars and drove up to buildings and harbours and cloverleafs and checked.
That bit just says a lot about ... accuracy.
Another possible explanation is the TomTom was using these unlikely to be visited streets as trap streets.
https://en.wikipedia.org/wiki/Trap_street
>In cartography, a trap street is a fictitious entry in the form of a misrepresented street on a map, often outside the area the map nominally covers, for the purpose of "trapping" potential copyright violators of the map who, if caught, would be unable to explain the inclusion of the "trap street" on their map as innocent. On maps that are not of streets, other "copyright trap" features may be inserted or altered for the same purpose.
Although I think it would actually be possible to use once-real-but-no-longer steets as trap streets, as long as your sources are improbable enough. For example, if you take a map from 1908, remove one street, then add in another single street from 1920, etc., your exact combination of streets could serve as proof of copyright infringement.
It brings up an interesting issue though - ideally if you have trap streets you would never want to use them in route finding...
And then users will simply turn to an alternative which absolutely does not care about their privacy, undoing all of Apple's effort in protecting their users' privacy.
Apple needs to strike a balance between protecting users' privacy and performing analytics. Perhaps send the data off-shore to a location not under US jurisdiction, I don't know. But it is clear that Apple cannot keep up with its competitors with its current practices.
It's incredibly detailed in pointing out the differences between old and new Apple Maps, as well as comparing map images from Apple, Google, Tom Tom, and others.
The article demonstrates through the use of animated gifs that Apple Maps has become significantly better over time and in many cases surpasses the quality of competitors.
Except for actually finding & navigating to places, where it's somehow getting worse rather than better.
Vegetation & building outlines are the primary aspects Apple Maps is improving upon per the article, whereas places are getting worse.
Here's one example article on the subject:
https://machinelearning.apple.com/2017/12/06/learning-with-p...
Google has been connecting place names by OCRing storefronts taken with their street view cars. Apple claim to be doing the same thing (but apparently doing a far worse job of it)
Now they know exactly where it is, when people like to go there, and if the opening hours seem accurate.
Not limited to them of course, this is also check-in companies' business model, like Foursquare.
How they don't yet have a web presence boggles my mind though. I send in small corrections to Google Maps all the time when I notice errors, Apple Maps won't let me do that.
The author definitely enjoys compiling these amazing essays and share this knowledge.
Thank you, Justin!
I still use Google for nav but I have a feeling Apple might win me over on this one.
I've always found Google vs Apple maps discussions to be interesting, as I've always vastly preferred Apple Maps data for walking around places; Google Maps too often had poor building shapes that looked nothing like the real thing, or hid business names at the scale I was trying to use. Perhaps that's because I live in the Bay Area. When I'm overseas, in particular, I don't hesitate to go to Google Maps first.
My partner briefly worked for a human-powered 3d mapping firm; they would get satellite and plane/drone photography of a large swath of land, split it up into block-sized chunks, and then each worker would take a block and use an in-house program to model the buildings at a pretty impressive level of detail. Workers got paid per-block and blocks were priced based on their complexity. They've been doing this for over 10 years by this point, so it's not an entirely unknown or uncommon thing to handle this kind of work manually.
In a world sense, the economics of making the map are very cheap.
In an ideal future, one can imagine submitting 3d models of buildings to a civic dataset as part and parcel of getting zoning approval.
Figuring out how many trees there are.
Do the directions work?
I'm talking postcard-perfect views around the Sydney harbor, small beaches around various bays, isolated parks near beautiful rivers and a whole lot of other places that are really beautiful and probably only known to the locals. Lots of these places were a brisk walk away from high-traffic areas too.
This is not to dismiss your question on the directions, though, which is certainly an important part of using maps on a phone.
But also likely the most common one. It's not fair to lump it in with the dozens of other map use cases.
Would you even trust an "interpreted" map with levels of green for what you're doing?
It doesn’t really work in satellite view because the trees make for a ton of visual noise at the relatively low resolutions shown.
A good indicator for me is green in the map, and no street view (street view tends to cover busier routes, I drive roads street view wouldn’t bother with)
I generally use Google Earth to do this, and I'm extremely skeptical that Apple Map's foliage would work half as well.
Sadly for Apple Maps... No.
That said, it was fine for anything that was on major highways.
As the linked blog post shows, they're definitely not perfect; I haven't encountered anything particularly egregious, but the fun thing about maps is that virtually everyone will run into what are ostensibly "edge cases" eventually. Apple Maps has steered me wrong before, but so has Google and Waze. (Waze has gotten much better at destinations in the last few years, after their purchase by Google, although they still have a disquieting penchant for "shortcut" directions that include "make this uncontrolled left turn across three lanes of traffic." Not only does Apple rarely do that, Google rarely does that.)
It's just making the map data match the real world better, and making it useful for more than just driving.
When I’m blasting down a highway, not so much.
http://gis.cambridgema.gov/dpw/trees/trees_walk.html
Or pick a tree for adoption?
https://www.cambridgema.gov/IWantTo/AdoptATree
Cambridge MA arborist has mapped all its trees.
"It's not just what it looks like and feels like. Design is how it works."
Most subtle changes I noticed on the examples on street names (e.g. "W 9th st" to "ninth street", sans "w") are actually a downgrade.