There were very few Apple Maps rumors for WWDC this year, only one little paragraph from Mark Gurman <with comments>:
An updated Maps app will make it easier to set frequent locations, like home or work addresses, and then navigate there. Users will also be able to create groups of frequent places and add a photo to them. The current interface for navigating to suggested or past destinations can sometimes be confusing. <duh> This will increase competition with Google Maps and Waze apps <really? are you serious?>
Apple Maps didn’t get much love at WWDC18. It was only after WWDC was over that Eddy Cue offered a sneak peak to Matthew Panzarino to let everybody know Apple Maps 2.0 was in the works and well on its way. Panzarino’s piece is the logical starting point for any WWDC19 Apple Maps 2.0 discussion. The best approach is to discuss things using the 3 groups outlined in my previous post: Collection, Processing, Presentation. As usual I’ll discuss things from the Japanese perspective.
Eddy Cue outlined Apple Maps 2.0 as a dual approach of using anonymous iOS device data and Apple Maps vans to collect high quality map data while getting faster updates from devices vs. the next scheduled drive:
“The truth is that Maps needs to be [updated more], and even are today,” says Cue. “We’ll be doing this even more with our new maps, [with] the ability to change the map in real time and often… In the new map infrastructure, we can change that relatively quickly. If a new road opens up, immediately we can see that and make that change very, very quickly around it…”TechCrunch Apple is rebuilding Maps from the ground up June 29, 2018
In short: Traffic, real-time road conditions, road systems, new construction and changes in pedestrian walkways are about to get a lot better in Apple Maps.
High quality in-house map data collection is a vital step, but there are limitations. The Google Maps Japan meltdown proved that even Google can’t do it all. When Google dropped premier Japanese map data supplier Zenrin, Google Maps Japan quality instantly crashed. Japan has very high density urban areas and very remote rural areas that cannot be effectively mapped from a van no matter how much fancy recording equipment it has. Zenrin has a 1,000 person ‘ground truth’ team just for mapping and updating those kind of places, inside and out, on site and on foot.
Apple Maps Japan data collection has kicked into high gear, but Apple Maps quality in Japan has suffered from poor locally supplied map data all along. Apple’s Japanese map supplier Increment P (IPC) bundles sub-supplier map data but doesn’t have a ground team to match Zenrin quality for hard to map areas.
Panzarino explained at length how the high-resolution image data collection effort fits with Apple’s in-house data qualification toolkit to identify problem areas with machine learning, so that the human team can quickly vet problems and update corrected map data for the trouble area:
The coupling of high-resolution image data from car and satellite, plus a 3D point cloud, results in Apple now being able to produce full orthogonal reconstructions of city streets with textures in place. This is massively higher-resolution and easier to see, visually…This is hugely important when it comes to the next step in Apple’s battle for supremely accurate and useful Maps: human editors.
Apple has had a team of tool builders working specifically on a toolkit that can be used by human editors to vet and parse data, street by street.
Many hundreds of editors will be using these tools, in addition to the thousands of employees Apple already has working on maps, but the tools had to be built first, now that Apple is no longer relying on third parties to vet and correct issues.
And the team also had to build computer vision and machine learning tools that allow it (Apple) to determine whether there are issues to be found at all.
There we have it: Apple is using in-house machine learning and no longer relies on 3rd party vetting or correction. How is this working out? Answer: not so great. At least in Japan. Let’s take a quick look around the Ikegami Honmonji Temple area.
Example #1: Ikegami Hall is completely missing in the map view even though it is in the satellite view.
Example #2: Duplicate Five-story Pagoda pin locations. The Manji character marked pagoda is correct while the grey one from Foursquare is the wrong location and duplicate information that needs to be removed or merged. <Kudos to Apple here for respecting local culture and using the traditional Buddhist temple Manji character, while Google Maps does not>
The conclusion here is that Apple Maps 2.0 isn’t living up to Eddy Cue’s stated goals, at least in Japan:
- In example #1 machine learning is supposed to identify problem areas when the satellite and map views don’t match up, but fails. The human team is not alerted to the problem and cannot fix it.
- In example #2 the system cannot distinguish between incorrect 3rd party supplied duplicate data and the real thing. In my experience Foursquare Japan and Yelp Japan have no human location vetting, most of their product is worthless. Apple faces a choice: is it better to show nothing, or is it better to show unvetted 3rd party data that has a high risk of being incorrect leading users to the wrong place? My suggestion: don’t use any 3rd party data that has not been vetted by Apple Maps van collected in-house map data.
Cartography and the Maps UI is where it all comes together.
Apple has a team of cartographers on staff that work on more cultural, regional and artistic levels to ensure that its Maps are readable, recognizable and useful.
For instance, in the U.S., it is very common to have maps that have a relatively low level of detail even at a medium zoom. In Japan, however, the maps are absolutely packed with details at the same zoom, because that increased information density is what is expected by users.
Panzarino got it wrong here. Users in Japan don’t want a map view packed with details. The difference is not cultural, it’s simply that high density metropolitan areas like Tokyo have much more information packed into a given area than American cities. Presenting high density information in clean easy to read cartography is challenging.
Yahoo Japan Maps and Google Maps have both evolved their cartography away from detail packed, point of interest cluttered views to cleaner cartography. Yahoo Japan Maps cartography is the best because they deploy good design with smartly edited zoom level assignment: this information is important at default zoom level, this other information belongs at zoom-in level 2, etc. This clean approach shows only the important details for the given zoom level for quick navigation. The differences in readability comparing Tokyo area views of Yahoo Japan Maps, Apple Maps and Google Maps are immediately noticeable. Here is Gotanda Station:
Apple Maps 2.0 fails here too. The cartography is less readable, recognizable and useful than the competition. The easiest fix would be for the Apple Maps cartography team to stop stuffing so many Point of Interest (POI) icons at the same zoom level and intelligently rank information to display at different zoom levels.
Unfortunately, that effort requires a group of humans with expert local area knowledge. An Apple Maps engineer explained the dilemma to me once, “Yahoo Japan Maps has the luxury of focusing all of their product development on just the Japan market.” It’s a luxury that neither Apple nor Google have.
WWDC19 Wish List
Here is my wish list for Apple Maps Japan 2.0 using the same categories, including transit which is a separate app and service layer within Maps.
- Traffic and Real Time Road Conditions: these important features are missing in Japan and absolutely must be added. Car navigation with Apple Maps in Japan is worthless without them.
- Sign a deal with Zenrin: because Apple Maps can never re-invent the digital map wheel for their Japanese product. Don’t make the same mistake Google Maps did.
- Offline turn by turn navigation: Apple Maps turn by turn navigation completely dies in underground roads or in rural areas without a network connection. It’s like flying blind. Dedicated Japanese turn by turn navigation systems handle this without a problem. Apple Maps 2.0 needs to match the same level of performance to be a reliable car navigation service.
- Fix stuff: Improve machine learning to identify problem areas for humans to fix, or hire humans who can identify and fix problems in Japan maps.
- Vet Stuff or Don’t Use It: If Apple Maps cannot internally vet 3rd party social networked geo trash from notoriously unreliable Yelp, Foursquare and TripAdvisor, don’t use it.
Presentation (Cartography and Maps UI)
This is where most of the action is covering how the map looks and how users interact with it.
- Apple Maps Cartography 2.0
Google Maps and Yahoo Japan Maps constantly tweak and evolve their map design, changing contrast, colors, text sizes, and more while pushing map information updates. Meanwhile Apple Maps cartography is fossilized in 2012 debut era design garb. I can only assume 2 things. Either Apple thinks so highly of the current Apple Maps cartography design language that it will never change it. Or Apple is creating a whole new cartography design. Let’s hope for the latter.
- Fix the Point of Interest overload: with smarter zoom level editing
- Eliminate Separate Map View/Transit View Modes
Toggling back and forth between 2 basic view modes in Apple Maps is passé. It desperately needs a revamp. Yahoo Japan Maps leads the way here by collapsing separate road map and transit maps into a single comprehensive map view that covers 99% of what users need, while offering a real rail map for the 1% who need a real rail map. It’s a time saver and smart way to eliminate toggling map views. More on this in the transit section.
- Recents 2.0
The current version of Recents is an old shoebox filled with crap: tapped places, liked places, Siri searches, suggestions, liked train stations to receive train delay notices, home, work, and stuff I have no idea why it’s even there. There are so many improvement suggestions I don’t know where to start. I’ll keep it simple and say, Apple please figure out what Recents is supposed to do, so that we don’t have to.
- Nearby 2.0:
Nearby suffers the same problems as basic processing, Apple Maps 2.0 needs to do a better job of filtering out the junk. Anybody can list 10 nearby cafes, but only smart editors can give me 10 that are worth visiting. Also follow Yahoo Japan Maps nearby approach of keeping everything on one screen, with minimal pinch and zoom. Avoid Google Maps approach of turning Nearby into stealth advertising.
- Live Weather Layer: this is Yahoo Japan Maps insanely great secret weapon. I always use it to find when its raining and where, with a time slider to predict if I need an umbrella at my destination. It’s a life saver and must have Apple Maps 2.0 feature. Once you use it, you can never use another map that doesn’t have it.
- Nearby Transit Time Widget
Google and Apple both use the same transit data supplier, but Google Maps uses it much better than Apple Maps. Most people already know where they are going and how to get there. What they really want to know is: when is the next train? Google Maps does this via a handy widget that offers location based nearby station train times and bus times without having to open the map or tap on a station. This is incredibly simple and convenient. Apple Maps 2.0 needs to offer it.
- Siri Transit Support
Siri does not support transit requests. Siri can navigate you to the nearest station but after that you are on your own. The ability to ask Siri for transit times is an important Apple Maps 2.0 feature.
- Transit Route Search 2.0
This is another area where Apple Maps has stood still while Yahoo Japan Maps and Google Maps continually push out improvements: route suggestion sorting by fare, transit time and number of transfers, train car position information for faster transfers and exits. Apple Maps 2.0 Transit needs to catch up with the competition.
- Location Based Transit Alarms on iPhone
Apple Maps transit has wonderful integration with Apple Watch but it could be improved with destination and transfer point alarms/alerts that also work on iPhone.
- Improved Apple Pay Transit Card Integration
Apple Maps has some basic integration with Apple Pay Suica but it could be improved by incorporating user Suica Commute Plan information for better route searches with more accurate fare information. Apple Maps integration with HOP and Ventra cards in Apple Pay Wallet would be a great feature for those transit regions.
- Adaptive Transit Times
The problem with transit route suggestions on Apple Maps, Google Maps and Yahoo Japan Maps is that once the user selects a route suggestion, transit times are locked in and cannot change on the fly. All too often a users selects a route and time but catches an early or later train, and has to input a new search to reset the transit time. But this is often impossible to do on the fly as transit route searches add a ‘time to station’ buffer. Transit times that adapt and automatically update to transit conditions would be a great feature to have in Apple Maps 2.0 transit.
We’ll find out at the WWDC19 keynote on June 3 at 10:00 a.m. PDT.