The Battle of Roshi Hall: Improving Apple Maps Japanese Data

Rereading my previous post something bothered me: the newer parts of Apple Maps like Japanese public transit, are clearly the result of hard work and are well done. It is older parts from 2012 that are buggy and full of incorrect information five years on.

I assumed that all the problems pointed out in Japanese blogger Train’s post, “Mistakes are Alive and Well in Apple Maps in 2017” were the fault of bad map data from Apple Maps Japanese primary map data (PMD) supplier IPC, but that assumption was too simplistic and incorrect. Something else is going on.

Searching for Something Else
I have written about the odd, long, verbose place names and amateurish place name categories in Apple Maps Japan many times (here and here). Blogger Train lists similar problems with arcane, old-fashioned, incorrectly named places and missing data. A quick comparison of his problems reveals that IPC has correct and up to date map data… but Apple isn’t using it.

In a similar vein, I have submitted countless place name corrections via the Apple Maps problem reporting system about a particular building near my office: Roshi Hall (Roshi Kaikan). Roshi Hall is a local community facility that houses classrooms, school clubs, and a small gym for children’s gymnastics, taiko drumming and so on.

Apple Maps ignores all corrections and insists on using an incorrect name. It displays the building as “Ikegami Sports Club” which is just one of many school clubs that use Roshi Hall. IPC lists the correct place name (as do Google and Yahoo Japan Maps). A wider inspection shows that IPC place names are correct and concise in a way they never are Apple Maps.

Who Supplies the Japanese Place Names?
Clearly Apple is not using the place names offered by IPC but are using a different data source. Who? The answer is complicated.

I suspect that Apple has a primary place name data (PND) supplier that it supplements with 2nd tier information from Tabelog (restaurants, cafes) TripAdvisor (Hotels), Bookings (Hotels) and Yelp Japan (hit and miss places of interest). The PND is not IPC but I have not discovered who is.

Looking at the evolution of Apple Maps Japan product, PND and PMD are the biggest problems Apple needs to fix if they want a good service that Japanese customers want to use.

Some Fixes
Let’s keep it simple and assume that the Apple Maps team has three choices:

  1. A Japanese product that is better than the competition
  2. A Japanese product that is as good as the competition.
  3. A Japanese product that is worse than the competition.

Choice #3: keeping the status quo is the easiest and cheapest choice, cheap in the short-term but more expensive later on. No major investment to improve quality risks losing Japanese customers in the long run.

Choice #2 is not necessarily more expensive but requires a significant investment in effort and system expertise: realize that IPC will never get Japan maps where it needs to go, drop IPC and the current PND supplier and cut a deal with Zenrin. Zenrin is the premier Japanese digital map data supplier used by Google Maps, Yahoo Maps and all the top digital map services.

The Zenrin choice offers Apple an opportunity to achieve Google Maps and Yahoo Japan Maps quality with their current system without reinventing the wheel. The Apple Maps Transit team already achieved this kind of success by using Jorudan, the same Japanese transit data supplier that Google uses.

Choice #1 is the most difficult and long-term choice, it also assumes successful completion of choice #2. A few ideas come to mind: a super smart and super connected Siri, untapped Japanese suppliers such as Recruit’s Jalan, detailed and nationwide Michi-no-eki (roadside rest area) with store-in-store information featuring popular local delicious omiyage, on-route live updates for transit, improved and detailed indoor mapping.

There are many smart and hard-working people at Apple with great ideas. It will be fun watching the choices that the Apple Maps team makes in 2017.