NPR

Why Mobile Maps Sometimes Lose Their Way

Mapping streets is easy. The trick is pinning down businesses and giving accurate turn-by-turn directions, as many people discovered when Apple launched its apology-worthy Maps app for iOS 6. Rakesh Agrawal, principal analyst for reDesign mobile, talks about how mobile maps are made--and what can be done to improve them.

Transcript

IRA FLATOW, HOST:

This is SCIENCE FRIDAY. I'm Ira Flatow.

UNIDENTIFIED WOMAN: Starting route to Empire State Building: Head northwest on West 43rd Street.

FLATOW: That's the voice of Apple's maps app for iOS 6. She sounds confident enough, but how do you know she'll actually lead you to the correct destination? Because as users all over the world have figured out, Apple's maps and their driving directions have some serious problems. Apple has even apologized for it.

For example, you search for Yellowstone, and instead of finding Yellowstone National Park, the map suggests Yellowstone Imports on 32nd Street here in New York. A second search for Yellowstone gave me Yellowstone Realty in New Jersey. So how about searching for Yellowstone with the map centered over Yellowstone Park? That should be a no-brainer, right, huh? No. A pin pops up on the side of Interstate 94 outside Billings, Montana, a four-hour drive from the park.

And while Apple may be getting the bad press and feeling the heat right now, other mapmakers like Google, MapQuest and Bing have glitches of their own, which makes you wonder: How do they actually make these mobile maps? Why is it so hard to get them right? And how soon before they fix these errors?

That's what we'll be talking about, and my guest here to talk about it is Rakesh Agrawal. He is principal analyst for reDesign mobile in San Francisco. He's been writing about mobile apps for VentureBeat.com. He joins us from KQED.

Welcome to SCIENCE FRIDAY.

RAKESH AGRAWAL: Glad to be here.

FLATOW: It's OK if I call you Rocky? I understand your best friends do that.

AGRAWAL: Absolutely.

FLATOW: Tell us, you know, what is so difficult? Every map has got problems with it. What is so difficult about making these maps? How do you make them? How do you start out?

AGRAWAL: Well, the United States, the way it starts out is there are maps that come from the government called TIGER maps. And that's a basis for a lot of the mapping that's done in the United States. And then they have people who drive around in cars and vans that record map data.

So these maps - or these vans will have cameras and GPS sensors, and they drive around. They take pictures of all the signs, and they input all that data into the mapping databases. Increasingly, we're starting to get maps that are being built by individuals who are driving around.

When you are out there with your vehicle using some of these apps, that's generating data that can be used to figure out where new roads are being built.

FLATOW: So you, by using the app, help make the app better?

AGRAWAL: Absolutely.

FLATOW: Hmm. One of the criticisms about Apple's map and other maps have been that they've been pretty bad at pinpointing businesses and points of interest. In downtown Manhattan, a kindergarten is placed on top of a Rite-Aid drugstore, for example. Coffee shops may be on the wrong block, stuff like that. How do they actually figure out where that stuff is? Is it by driving around in those cars?

AGRAWAL: Well, the business data typically starts from the Yellow Pages. So most businesses have a telephone number, and the phone company knows what that is, and they publish that in a book. And historically, what's happened is that those books get shipped overseas, and somebody either types them in or, increasingly now, they're scanned in, and that gets put onto a CD-ROM and gets uploaded into one of these databases.

That was a historical way of doing this. Now you have a lot of businesses that provide the data directly to companies like Google or Apple or Yelp, for example. But all this data changes constantly. New businesses open up, old businesses go under, businesses move. So it's a really tough, moving target to keep track of where all these businesses are.

And every database is going to have errors in it just because this changes so frequently. One of the things that's starting to happen now is users are starting to add their own venues. So you can go into Yelp and say: There's a new restaurant that's opened up near me. Here's where it is. Or this restaurant that I went to is closed now, so please take it off the maps.

So increasingly, users are contributing that data, but somebody who's been at it as long as Google has is going to have better data than somebody who is just getting into the space, like Apple, where they don't have all the processes to collect that data from users.

FLATOW: Yeah. So as I was driving by Citi Field, where the Mets play, and it still said Shea Stadium on it this week, it's been three years, I should phone something in to somebody and let them know that.

AGRAWAL: Absolutely. But the other thing is you still want that to come up when somebody searches for Shea Stadium because a lot of people probably still call it Shea. I still call the ballpark out here, even though it's officially AT&T Park, I still call it Pac Bell Park. So if somebody's doing that search for Pac Bell Park, you want them to see AT&T Park.

FLATOW: Now, you actually took Apple maps and Android for a test drive in Washington state to see how well they performed. How did they compare?

AGRAWAL: You know, clearly, Google was better. There was a lot less mapping problems in Google. They've also done, overall, a better job of user interface and getting the little details right. So, for example, I was on a road that was primarily known as State Road 20, but had an alternative name of Sims Way. And Apply kept referring to it at Sims Way, which is not what was on the street sign.

So if you were trying to follow the Apple maps, you wouldn't necessarily know where you were going. So little details like that, Google was much better at.

FLATOW: Mm-hmm. And I thought - there's a funny anecdote you were telling us. You turned on the Australian voice on your GPS unit in Washington and had sort of a strange result.

AGRAWAL: Exactly. So one of the things that happens is a lot of these things get abbreviated. So Washington typically gets abbreviated WA. And, you know, the right thing to do there is to expand that to Washington Route 20. Because I had turned on the Australian accent, they expanded it to Western Australia Route 20.

(LAUGHTER)

FLATOW: So they just - they didn't think you just didn't want to hear a g'day. They thought you were in Australia when you turned it on.

AGRAWAL: Exactly.

FLATOW: Why did Apple make this break with Google and not have a Google app for this?

AGRAWAL: Well, there were two big reasons. Apple has wanted to have turn-by-turn driving directions on their platform, which is something that's been available on Android for a couple of years now, and it's been a big competitive disadvantage. People want to be able to get turn-by-turn directions, and apparently they weren't able to get that from Google. So they really felt that that was a competitive need.

And then the other big thing is, as we've talked about, the more data you have, the better you can make your maps. And up until now, Apple has been providing that data to Google. And by providing their own maps, they now have the ability to capture all those millions and millions of iPhone users, their data and use that to improve their own product, as opposed to improving somebody else's product.

FLATOW: Now, there are third-party apps like INRIX and NAVIGON and GPS Drive, to just name a few. Are they as good, or do they work the same way? Do they have the same problems?

AGRAWAL: The core data problems are going to affect pretty much any company that is licensing data from the commercial data providers, and most of the companies in the space license data from two providers. Google is the only one that's really gone out there and made a huge effort to collect their own data, which is why they're doing a much better job.

They've also made a huge effort to go out and collect user-generated data. So, on the whole, I would expect Google to be tremendously better than the other providers. There's going to be some variance depending on certain markets might have a better provider. But on the whole, especially in the United States, I would expect Google to have the best results.

FLATOW: And last question to you: You suggest in one of your articles that mobile maps might be a good way of getting rid of ugly billboards. How does that work?

AGRAWAL: Absolutely. You know, when I'm going down the street on the interstate, you've got ads all over the place that just clutter up the entire landscape. And oftentimes, you go to places not because it's where you want to go. It's because they have the biggest or most obnoxious sign.

With mobile maps, you can search for exactly what you want. If you're a fan of Friendly's, you can see that there's a Friendly's five miles down the road, whereas the Cracker Barrel has a big sign, and it's only one mile down the road. But maybe you can wait the extra four miles.

FLATOW: Yeah, maybe we can. Well, we can all hope for that. Thank you, Rocky.

AGRAWAL: Thank you.

FLATOW: Rocky Agrawal is principal analyst for reDesign mobile in San Francisco. Transcript provided by NPR, Copyright NPR.

Most Popular