Comment by pooriaazimi | Hacker News Reader

Comment by pooriaazimi | Hacker News Reader

Parent Full thread

It wasn't an approval problem. I bet Apple would include Google maps in a heartbeat if it was free (and ad-free), but they had to pay for the privilege of having Google Maps.

The old app was Apple's maps app, that just happened to use Google APIs. And of course, to use Google maps data without letting Google show any ads, you have to pay Google. And if you have 400M iDevices, the number you have to pay Google is probably in order of >$50-100M per year (or something like that - I have no idea; just wanted to say that it's a lot). And even after paying that much money, Apple still didn't have turn-by-turn navigation or vector maps (the rumor is that Google wanted more user data and Apple refused).

So, Apple decided that instead of giving Google $500M in the next few years, to spend that money on developing their own mapping technology and be masters of their own destiny. Sadly, the result is laughably worse than Google's data, and probably will be comparable to Google's 2011 sometime in 2014.

> Sadly, the result is laughably worse than Google's data, and probably will be comparable to Google's 2011 sometime in 2014.

It will improve sooner than that. Apple has been rapidly making deals with data providers in each country. They never had those in place at launch which was why it was so bad.

As someone who likes Apple and its platform(s) and has invested heavily in them, I'd like to believe that, but something tells me it won't.

I'm quoting John Siracusa, because I think what he said makes a lot of sense: Google does brute force (i.e., driving cars all over the world and having thousands of real people auditing the maps data), but Apple wants to get its data using "clever" algorithms. Clever algorithms is what caused [1], and countless other issues.

What I said above is debatable, but the next statement is (I think) absolutely non-negotiable: Until we see Apple Maps (or one of its partners') cars driving around the world (the whole world, mind you - not just US and EU), its data will be demonstrably worse than Google's. It might be good and usable, but still will be worse than Google's :)

[1]: http://www.macrumors.com/2012/12/11/apple-fixes-australian-m...

Where do you think the data providers get their data from ?

They have cars as well. And they've had them for DECADES before Google. And it's not like Google is driving around continuously updating their data. For example Street View in many parts of my city is over 3 years old and still missing many newer suburbs.

So no. Google does not have a brute force approach. They have just had longer to perfect their algorithms.

So why didn't Google just buy data from them?

They did, but restrictive licencing terms drove them to start collecting their own, the process is still not complete.

As mapping professionals have observed, it isn't as simple as just having more and better data sources. Putting together a solid mapping solution from many disparate data sources can be harder than gathering more data yourself (i.e. there's a reason all those Google Street View cars are out there).

See the frequently referenced: http://blog.telemapics.com/?p=399

It's also worth noting that Apple's local POI search has been a particular pain point. Thinking it will be easy to match or outdo Google here (when Google Maps often pops up the right POI as a suggestion when I've typed as few as three characters) is foolhardy.

Again. Google Street View cars are NOT going around continuously updating data. It is their existing data providers that are doing this for them. I know this for a fact. And sorry but Google doesn't own any POI data either. They license it from third party providers. For example in Australia it ALL comes from Yellow Pages.

If you do a POI search e.g. Sydney restaurants in Google Maps you can scroll down to see the attribution e.g. "Business listings distributed by Yellow Pages". Likewise for most other countries.

But hey I will grant you that Google Maps has a better autocomplete.

I didn't say Google did everything themselves. That would be ludicrous and, AFAIK, no one does that. I said that gathering more data yourself can be easier than integrating disparate data sources. In other words, direct data gathering is an important part of the story and fixings things isn't as simple as "MOAR DATA"!

> And sorry but Google doesn't own any POI data either.

Some of Google's recent acquisitions would beg to differ:

http://www.wired.com/business/2011/09/google-buys-zagat/

http://money.cnn.com/2012/08/13/technology/google-frommers/i...

And, unless you know differently, I'm pretty sure Google owns Google Places...

Oh, by the way, I wouldn't trivialize what Google's doing with autocomplete. The way Google works on reliably guessing user intent from limited, often noisy input makes a huge difference on mobile. To take one example, search interfaces that can't tolerate misspellings are irritating on the desktop and downright painful to use on a phone or tablet.

There are lots of data sources other than Street View, satellite and aerial imagery being big ones obviously, but also user provided data, Business Photos, phone calls, etc.

The Yellow Pages attribution you see doesn't necessarily mean all data came from Yellow Pages, just that some may have.

> Again. Google Street View cars are NOT going around continuously updating data.

I think you might be a little bit dated in your understanding. Google any random city on maps.google.com. Look at the fine print at the bottom: "Map data ©2012 Google"

I remember what that used to say a 4-5 years back. Copyright NAVTEQ. They have been quite busy.

Great, so we'll just sprinkle the magic data provider fairy dust on their maps, and oh wait, yeah, it's significantly more complicated than that.

Great, so we'll just sprinkle the magic data provider fairy dust on their maps, and oh wait, yeah, it's significantly more complicated than that.

Look, don't get me wrong, i'm a big fan of being able to catch up. I was part of a small group of folks that took GCC's compiler technology from late 80's to state of the art in the mid 2000's by cherry picking the best of the best from research of the past 20 years. It worked. In 5-6 years, we were able to get within 1-2% of the best compilers at normal opt levels.

But it still took 5-6 years. Getting good takes time. There is no magic fairy dust. The secret to getting the last 25% of the way there in anything is hard work and lots of tuning. There is no magic bullet that will suddenly fix everything.

Reply on news.ycombinator.com