It certainly will be better but I'm not holding my breath about Apple maps surpassing Google Maps any time soon. Google bought a Rasmussen brothers' company (what eventually became Google Maps) in 2004. That's eight years of product development, data collection, storage and refining -- all at Google's pace. Will Apple find some shortcut to compress this time in one-two year? Unlikely.
It's very easy to underestimate the amount of pure, repetitive effort required to build products of this sort and how hard it is to "disrupt" the industries with players like Google.
I don't think it will be easy for them to catch up (it would be absurd to say that they can have something better than Google Maps in such a short time).
What bothers me a lot, is that most people blame the data errors to the algorithms or ability of the team to create good software. You can have the best coders in the world, and if they have data that is not good, they can only algorithmically clean it so much.
Google has several vehicles that they drive around and collect all kinds of information themselves. Sure you can hire Navteq (or in Apple's case, TomTom aka TeleAtlas) to go around and do it for you. Guess what? When you license products from either of those companies that means you are relying on their editors. That means no fancy object recognition geotagged image that you can send to a support vector machine to flag an area as needed revision... the product you get from that licensing is the vector line with the attributes already attached to them. So that is what you work with.
Google has an entire army in India, using Google-made custom editing tools that do these things. Unless Apple builds the same, they will always be behind. Can Apple do it? Sure! They have the cash and the talent. Will they? It is up to them to decide if investing 400 million into their mapping infrastructure is worth it. I think it is.
At the same time, Apple has a carefully cultivated history of not shipping half-baked products. They famously worked on tablets for years before they built one that they were happy enough with to ship.
Now, for reasons that the end user surely does not care about, they have taken a product that was missing a key feature or two (navigation, street view) and replaced it with something that is much less reliable even for its fundamental task.
Yes, making a really good mapping application is hard, and requires many people on the ground, around the world. So either you need to make the commitment to do what it takes to deliver that app, or you should outsource the whole thing, and use Nokia Maps or Bing Maps or whatever. The third alternative -- doing it yourself and shipping it to end users before it's ready and without having done all the due diligence -- is likely to be embarrassing.
For instance, in the map of the Shibuya station area in [1], besides the general paucity of detail, note that only roads are considered worth mentioning, although the vast majority of people don't arrive by road. That may fly in Cupertino, but it certainly doesn't fly in Japan. This is not the middle of nowhere—it's one of the most popular places in the largest city in the world, and indeed, is iconic even in the U.S.
The impression one gets is that Apple is way out of its depth here, and given the importance of mapping for smartphone users, that seems a pretty major screwup.
[1] http://www.japanmobiletech.com/2012/09/ios-6-maps-fail-in-ja...
"Sources tell The Verge that Apple began work on the iOS 6 maps system nearly five years ago"
http://www.theverge.com/2012/9/20/3363914/wrong-turn-apple-i...
Wow. Just wow. Now we see how strategically brilliant (and utterly necessary) Android actually was.
Given the relative sizes of patent portfolios prior to Google's acquisition of Motorola Mobility, I wouldn't bet on Google should the IP wars shift to mapping...of course I wouldn't bet on Apple either.
http://theamazingios6maps.tumblr.com/post/31923886066/are-yo...
There are a lot of problems with road map and point data (at least in the UK) which are nothing to do with the points you've raised above, though those are fair criticisms too - however the satellite data is the least of their worries, I'm sure they could buy in a better cleaned up set of satelite tiles if they're willing to spend enough money, and people expect some problems with that, but basic search and mapping is not as complete as it should be.
I love OSM and I use it every day - but I know what it does, what it is good for and what it is not meant to do.
Getting a list of towns for the entire world is easy. Solving the issues in the listed by the OP's tumbler requires data that OSM does not have.
For most people the 3D is an amusing toy, not really useful. I'd be happy without it frankly.
OSM doesn't do elevation
See 3D above
OSM doesn't do geocoding
They have data for searching on street addresses and towns, and the data is available separately for postcodes etc (TomTom for example will have sold a solution for this to Apple, so they could have gone down that route). Google has this down, as you'd expect given their search expertise, and Apple would have done better to pay them at least for access to this service if nothing else as it's a complicated area.
OSM doesn't have a cleaned up dataset of Points of Interest
Yelp does, and that's who Apple are using for POIs
One other point you didn't mention is that OSM doesn't do satellite data (their stuff is from Yahoo isn't it?), so that is a big area Apple would have had to sort out from another provider.
While you're right that there is no one-stop shop for mapping services, and it's a really complex area (i.e. just choosing OSM would not solve all these probs, as you point out) I do think Apple have missed a chance to form a symbiotic relationship with OSM and enhance the quality of their maps with an open process rather than relying on the often out of date commercial data available and having to either fork the commercial data from this point on, or merge changes. Dealing with all these fixes will suck up a lot of time they could spend on improving the app itself.
The complaints about the "zigzagging" are about the 3D viewer, hence, my comment about 3D.
>They have data for searching on street addresses and towns, and the data is available separately for postcodes etc (TomTom for example will have sold a solution for this to Apple, so they could have gone down that route). Google has this down, as you'd expect given their search expertise, and Apple would have done better to pay them at least for access to this service if nothing else as it's a complicated area.
To have good geocoding, you need to have line segments with correct ranges and topologically accurate (for example splits in the right intersection taking in consideration overpass/underpass and grade changes - the 3D stuff you don't care about - along with no dangling edges). If you think OSM fits this bill in a world-wide dataset you are incorrect. Go ahead and download the entire dataset and see for yourself planet.openstreetmap.org.
> Yelp does, and that's who Apple are using for POIs
Actually, yelp only has coverage in the US and in the few outside countries they are in. Most of the world is not covered by Yelp's POIs, so sorry, but "Yelp" is not a worldwide solution for POIs - which is what Apple needs.
> One other point you didn't mention is that OSM doesn't do satellite data (their stuff is from Yahoo isn't it?), so that is a big area Apple would have had to sort out from another provider.
Yahoo, nor google, nor anybody else have their own satellite providers. That data usually comes from GeoEye, Digital Globe or any other satellite image provider. But yes, you are right, "OSM" has nothing to do with raster data as a whole (except for the elevation-derived contours they have in some places).
Fixing OSM for anything else than displaying maps is a huge undertaking. Then, when you are done, you basically have to share the data with everybody else (including Microsoft, Apple, Yahoo, Mapquest etc etc). OSM is the GPL of databases, so for some use cases, it is not the best choice, for other ones, it is.
The post I replied to claim "Apple just pick OSM", and my answer remains... it would not have solved any of the issues that people are complaining about.
It didn't say 'Apple just pick OSM' (which implies picking OSM would solve all their problems), and as I've pointed out, there are problems with data which it would solve. Here are some examples:
Puna, Peru, roads in the water http://www.openstreetmap.org/?lat=-15.8559322357178&lon=...
Cayman Islands, no road data http://www.openstreetmap.org/?lat=19.2949&lon=-81.343...
These are serious problems with data which Apple will have a nightmare fixing as there are just so many holes in their data outside the US, and they're now waiting for mapping companies like tomtom for updates, or they'll have to fork and then try to merge with tomtom/etc data later again.
For many (I'd say most) users of the maps app, the order of priorities are something like:
1. Accuracy of map data
2. Coverage of map data
3. Accuracy of map searches (streets, postcodes, POIs)
4. Routing (this is hard problem though)
5. Satellite imagery (for some this would be above routing)
6. Streetview (for some this would be higher)
7. 3D stuff
OSM would help with the first three points (along with other sources of course, just as say tomtom needs supplemented), and would also help in future as other people would do the hard mapping and updating work for apple, which will otherwise prove to be a mammoth task.
Everything else is far less important than getting the basic mapping data right first, including 3D renders of major cities, and even satellite data (nice to have, but less important than basic search and display of accurate mapping).
If they used OSM data and imported (say) every month, they could just tell people to visit OSM to make corrections, and problem areas would gradually get better. As it is they're going to have to sift through hundreds of thousands of reports of missing roads etc which just say 'this is broken, fix it', and they're likely to just be submerged under that workload. Anyway, I don't think we're saying such different things - I'm not saying OSM would cure all Apple's problems or even most of them, just that I thought it would be the best choice for them for the most important facet of their map product, the basic vector data, which is sorely lacking at present and unlikely to rapidly improve.
OSM data is licensed under ODbL. It has nothing common with those "license terrorists".
True OSM doesn't have that, but there are some public domain datasets available for that, e.g. SRTM.
OSM doesn't do geocoding.
True. OSM is just a database, you'd have to make your own "address ↔ position" algorithm. Which is sorta hard.
OSM doesn't do routing.
OSM is just the database. You can (and people have) make a routing engine on top of that (e.g. http://map.project-osrm.org/ http://maps.cloudmade.com/ ). You can do very specific types of routing (e.g. I figured out how to cross Dublin without passing a pub http://www.kindle-maps.com/blog/how-to-walk-across-dublin-wi... ).
OSM doesn't do 3D.
True.
OSM doesn't have a cleaned up dataset of Points of Interest.
Yes and no. OSM is a database and has lots of Points of Interest, e.g. here's a Starbucks in London ( http://www.openstreetmap.org/api/0.6/node/488443269 ), so it's trivial for someone making a map app to make a database of PoIs that an application can look up.
The only time Apple helps open source is when they hire the most important people in really important projects, and then they graciously allow them to continue contributing. CUPS, LLVM, it's happened a couple times. They take good care of these guys, but I always feel like their contributions to the projects after Apple are being written with one eye watching over their shoulder to make sure they dont get sued for talking to the public about some super secret project.
I'd rather have seen more focus on getting the basics right, even if it meant they had to leave out (for example) satellite imagery in the first iteration. As it is they have broken satellite data, broken street mapping, and incomplete place names.
If they had chosen to back an open project like this they could have had a team of people worldwide fixing their maps for them, rather than trying to collate and correct thousands of bug reports from around the world and push them out in revisions. I wouldn't like to be on their mapping team over the next few months :)
http://www.theverge.com/2012/6/11/3078987/apple-tomtom-opens...
http://gspa21.ls.apple.com/html/attribution.html
It's not clear exactly where they get all the data, and what is used where, but simple searches like Paddington Station are not working (Paddington does, but not Paddington Station), and I'm sure they're not using the OSM vector data (in the UK at least) as it differs and is less complete than OSM in rural areas for example.
It would be interesting to know just how they've put their datasets together and what they used from where, but it looks like they use tomtom map data in the UK (along with Yelp places, and perhaps Royal Mail addresses/postcodes).
It's not surprising that densely populated areas in Europe/UK are complete and accurate in OSM.
The main advantage would be that OSM has all the adding/editing/updating stuff down - that is a major operation and if Apple are trying to do it all themselves while merging inconsistent data from several sources and taking care of customer complaints they'll find themselves falling farther and farther behind people like Google who have devoted years and years of effort and a lot of money to this just to get where they are now.
Perhaps Apple considered OSM as a basis and decided it was just too limited in areas that were important to them - it would be fascinating to find out about their internal tools for this and where the data came from but I suppose we'll never know.
They seem to have tried out OSM in the photos app (with a horrible skin), but to have used purchased data for the street maps. Perhaps they felt it wasn't good enough in some locations they tested?
In my experience OSM is superior to Google maps in many locations (even in parts of central London), and it is of course continually improving.