It's nice to promote open data, etc. but lets be honest- it's all about the money. And Google charging for Google Maps usage is going to result in a lot of people doing this.
I'd advise that everyone at least explores the possibilities out there, though. I'm making a mobile app right now that uses roadmaps, and have ended up generating my own map tiles using TileMill and open data. It's surprisingly easy to do, and makes things like offline caching (not possible with the Google Maps license) possible.
Well it was more specifically that their service could not support the cost. Sounds like they would have paid if it had been less or their business was much more profitable. It is a big problem with this type of service where many uses are not directly revenue generating. The traditional model has been to try to get the sales team to price at what the customer will pay, rather than have list prices, but Google by the sound of this has not got a good sales team in place, they are not culturally a sales organization. They should perhaps try some sort of auction based pricing that works for this type of data service in the adwords model so it can become self service.
OSM is of course an excellent product for people who want to do more than the off the shelf google functionality.
Absolutely. Enough articles are written on the topic of abstracting layers of your application away from each other (data access, view, service, etc) but rarely do we hear about the importance of abstracting your application from 3rd party libraries and services.
And with all the cloud stuff nowadays it's actually very, very important. each could go down any day, start charging more any day, or just not be as good as you hoped.. and then a 3 line changes to switch service might just be the difference between dead company and profitable company.
According to the article, the Google "sales force" seems to leave much to desire. They don't keep appointments, they can't explain their product, they don't understand their client, and they overprice their product.
It does not surprise me in the least that savvy users will find other arrangements.
Well we just went through the same process. And the guy we talked to at google was extremely helpful and knowledgeable. But in the end that did not matter, because even after a 80% discount of the list price the final google maps offer was still 350% more expensive than our local alternative(a country-specific provider). So come january 1st, we are changing map provider. Fortunately it was rather easy to switch. Its been great having been able to use google maps all these years for free, and for that we are grateful.
It takes one to I know one... Google was founded by phd's not sales people- it is. Very different to hire and manage a sales force compared to an engineering team... Sales people tend to be less honest and often comeback to bit - look at groupon
They also don't type on computers during interviews, which I doubt is something specific to engineering. My interviewers took notes, copied down what I wrote on the whiteboard, and then (apparently) later entered all of this into the computer system.
As for the questions, they certainly don't ask any that are in the article. Steve Yegge's article covers the exact category of questions they ask:
I don't remember the exact questions I was asked, and signed about 80 pieces of paper saying I would Never Speak Of Them, but they are of the style where you must know X and then use pieces of X to solve completely unrelated problems. (X could be, for example, quicksort.) Graphs, algorithms, and data structures are essential, but nobody is going to ask you "implement a binary tree"; that is way too easy to study for. (I'm guessing that discrete math questions are also common; but because of the position I was interviewing for, they replaced those with testing and refactoring questsions. Which was good, because I was never good at discrete math.) I had not seen any of the questions they asked before, and two of them were interesting enough to keep me thinking about them for several days. (I was disappointed to discover, while implementing my solution to one, how easy it is to brute-force problems involving English words. There just aren't very many words. But the idea is what counts, and brute-force is not going to get you hired :)
Since I've been hired by Google, people have pointed me at a lot of "popular media" articles or videos about Google, and not many have matched my experience, but then again, I haven't actually started work yet.
(The linked article also talks about Bank of America intervewing, which doesn't have a standardized process. I know because I worked there and asked different questions, 100% chosen by me, every time I interviewed someone. So while one person at Bank of America may have asked "What animal are you?", that's not some sort of company policy. So it's a bit disingenuous of the article to name-drop Bank of America there.)
I would imagine it's because brain teasers test mostly one thing: If you've heard the brain teaser before. (Or, alternatively, how much of a smart aleck you can be. "Why are manhole covers round?" - "Because manholes are round. If they were square, they wouldn't fit")
There's much to be done in the web mapping yet, Google Maps initially brought a renaissance to the field but the problem is that now everyone try to beat Google by emulating Google Maps (just like some companies loves to emulate Apple products). For example, Google buys the majority of surveying data that it uses in building it's maps, everyone can buys and use the same data.
OpenStreetMaps has its own data but it is provided by a community that does not have as much momentum as Wikipedia, even New York City data is pretty much incomplete
Part of the problem is that people really do not know anything about cartography or how a dynamic maps must behave, even worse the majority of people do not even know how a web mapping service is implemented.
> For example, Google buys the majority of surveying data that it uses in building it's maps, everyone can buys and use the same data.
This is no longer true. Google spent millions of dollars paying people to drive cars around collecting data for them. The majority of their data nowadays is their own (in most of Europe, Canada, and the US at least). Recently, they've even convinced their users to work on their data for them with the Map Maker product (all while holding the data behind a non-commercial, non-compete license and charging 10's of thousands of dollars for the right to use it).
I read the post and it's mentioned many times that OSM is as good as Google Maps. Others argue "its just about the money". Why would someone want to pay for a service that you have to pay for and its just as good as the alternatives?
I think the key factor here is something Google does very, very well: UI. There is just something about google maps that makes it more appealing to the eye. And that's crucial to create a great service. I saw a post long ago (i couldn't find :/) that compared the graphic style of Maps, Bing and others. Google just nails layout, ease of understanding and other factors. I believe it is this key factor that makes the difference.
"I read the post and it's mentioned many times that OSM is as good as Google Maps."
And it's also mentioned repeatedly that they were perfectly willing to continue with Google Maps until they had a bad sales experience and were estimated a figure that they couldn't pay.
When you say that and you put the charging issue third, it makes everything else look like a rationalization.
ETA: And as I go to OpenStreetMaps and see it struggle to serve me a page showing my house (and showing exactly the same wrong place for my house that Google Maps does, because it's using the exact same data), it doesn't much make a case for anything but "well, at least it's free".
Yes, it loaded glacially when I tried it, and there was a link begging me to donate for a hardware upgrade. It showed precisely the same wrong location, given a search address, that Google Maps shows - which is pretty impressive if they're somehow not using the same data, there. As for fixing it, fix it yourself. I'm not a cartographer, and I'm not actually interested in this project or in being hectored by someone to pitch in on a project I don't give a damn about.
Definitely agree. This is why we (Stormpulse) not only avoided Google Maps but spent 1,000's of hours designing the UI. I would avoid mapstraction for the same reason: we see mapping as a core competency of our business. YMMV of course.
I just checked my home address on OSM. My whole subdivision and the major 4 lane road that goes past it was all build 6+ years ago. None of it exists in OSM. This is a well populated suburb of Atlanta.
Yep, 80% of my town is unmapped. The same goes for some other towns in neighboring areas that I've checked. Compared to Google maps it's just no good. I have no idea how or on what basis the blog author concluded that OSM is just as good or better then Google maps. Also, I didn't see the option for satellite view.
Well, and my town is mapped perfectly. The difference is that with openstreetmaps, you can simply go and improve what could be better. Of course it doesn't change anything about the quality of the maps, just that you have a choice, as opposed to google maps with which you could possibly be stuck with an incorrect map.
And regarding satellite view, how would you even imagine volunteers making satellite images? Basically anyone can get a GPS receiver and make maps, but I have yet to see an amateur hobby satellite in orbit. (I guess you could use R/C planes with GPS and cameras, but that would be incredibly time-consuming and not ideal at large scales.) However, even if you could do that, it's just not goal of OpenStreetMap. You can't use satellite images for routing, there is no legend, etc.
And I think that they don't allow external satellite images because of their philosophy on freedom - similarly to you having to explicitly allow installing non-free packages in Debian.
> And regarding satellite view, how would you even imagine volunteers making satellite images?
Across the world, I can't answer. Here in Western Australia, one of the largest single states in the world, there's been digital mapping for two decades now, predating the rise of Google maps.
Data from MODIS satellites is downloaded direct from the sats as they pass overhead and streamed to those with an interest. The better quality data comes from the aerial surveys performed each and every summer over populated areas which produce several thousand digital scans of wet negatives (they may have recently moved to high res digital in the planes, I've been out of the details for a few years).
Google didn't develop their mapping tech initially, they purchased the starting technology from Keyhole, a US based company (iirc), it's worth noting that several international companies had equal or better offerings that were considered including our locally developed technology from ermapper which is now owned by Lecia.
As for volunteer contributions it's been our experience that if you provide data access and annotation tools then many competent people that care about ground accuracy (forestry services, land owners, search & rescue, exploration crews, etc) will happily add value.
I'm not arguing against open data model or users' contributions. On the contrary, I think it's great. I merely wanted to point out that at least for now it's not good enough. I'm also missing route planning/directions, however I may have missed it while browsing the site.
Something that people don't get (as it's a tricky concept) is that OSM isn't primarily the online mapping you see at openstreetmap.org. Just like Linux or GNU isn't about a single distro or use case, OSM is a community and tools for building a variety of different user facing tools from a shared digital mapping infrastructure.
Also, the best of OSM is in places like Germany where in some areas they're mapping down to the level of individual trees and lampost locations as well as where animals are located within zoos.
Apparently there is an Android app that will let you upload gps tracks and do other editing. I haven't really played with it yet, but I'm also in Atlanta so if OSM is really that bad here I might start helping.
That's awesome aeroevan. This is the whole reason I wrote the post - to encourage others to dive in and start contributing, be it on data gathering and/or on the technology side. The community is great and technically geo is a very interesting space. Please dive in.
On the other hand, google maps has been showing a road right through our neighbours' living room for the last ten years, while OSM correctly shows all roads and footpaths. Different maps have different errors, look what works for you.
Google Maps was first published in 2005, so it's been doing nothing for the "last 10 years"... And during that time they've changed base map data providers several times in most countries (Navteq -> Tele Atlas -> Google), making it incredibly unlikely that such errors would have persisted for the whole time.
Have you "reported a problem"? The "tile" under my property was misaligned and the road that it showed on the map was about 50' skewed from the satellite view. I did the report a problem thing and they fixed it in 12 hours. I'd let it be like that for years, figuring no one would care...
I might have gotten very lucky there, but it can't hurt to report it, right?
The last time I checked OSM was based on US Census TIGER data. The TIGER data format is extremely quirky and importing the full dataset is non-trivial. I'd imagine they only do it periodically. Add to that the lag between new construction and appearance in TIGER and it's easy to see why OSM has not yet caught up.
The good thing is that, unlike Google Maps, you have complete power to update OSM with the correct information for your locale. Get mapping!
Another warning about mapstraction... I use it on my site and am not that happy with it. It doesn't support (seemingly) exotic features like z-index for markers and other things I ran into which I can't remember off the top of my head. When asked on the mailing list, devs stated they want to keep the library small. Sounds great but I've got an application to write that needs more than lowest common denom. So you end up hacking the google objects underneath directly to access many features (which you'll think are standard in 2011).
So now I have two libraries to support and users to download, much less documentation to rely on, and am not completely abstracted.
If the ability to switch map providers is critical, would it not be worth the time and effort to cleanly extend your abstraction layer to cover the missing pieces rather than give up and call the specific backend directly?
There's a difference between "our profit margins will be reduced" and "our profits will disappear" and it sounds like this company was in the latter case. Profit may not always be king, but staying in business surely is.
In addition to the map tiles, OSM has an unbeatable advantage over any proprietery solution - you can take advantage of raw data and do whatever you can think of. For example, you can write your own route finding solution, tweak it whatever you like and potentially find new unexplored ways to existing problems (which I did for social based routing project in Eastern Europe). Needles to say - none of that is possible with the proprietary providers like Google.
Other than the Google logo in the bottom left hand corner, Google has done a great job of taking the back seat on development. You can pretty much make a map that looks like it's completely void of the Google family.
When creating a mapping application, it's extremely important to stay away from that. There's no point to creating an application that's just a styled Google map. There has to be a point to having your application centered around a map. I think a drawback of using Google maps (or any of the mapping services) is that it's so familiar to people. A lot of people might assume that the application your building is just a Google growth. Fortunately, Google has a good deal of features built into the API to keep that from happening. You can pretty much design what your map will look like down to the most seemingly superfluous of details.
Mapping applications are so awesome because of how interactive and driven by visual exploration they are. I think there will be a lot more of them doing things that we don't expect to see in the future.
One last note: Any TileMill developers out there? What would your advice be for someone who is making a mapping application/considering moving over from gmap?
That's a shame. Addresses can be very hard. OSM is an global phenomenon, and trying to figure out all the dozen different ways you can encode a latitude & longitude into a string as an address (and vice versa).
I just recently assembled a tile server using mapnik, OSM data and modtile. Data is importing on our new geocoding server, against using OSM data and Nominatim. We will be switching our routing service over to pgrouting, but I have to wait for a massive hardware budget to make sure it can equal gmaps in routing performance.
Going completely over to OSM takes alot of work when building your own hardware stack, but our alternative (with relatively low traffic), is paying google maps $40k next year, and who know what the following...
This seems rather disingenuous. A map that is missing major areas is not only not well mapped, it clearly "does not work".
This sounds to me like the age-old argument us techies make to laymen. Use open source! If there are problems you can fix them yourself! Except these people are not programmers, nor do they have any interest in being one.
I'm not a cartographer, I have no interest in being one. I certainly am not signing up to integrate OSM into what I'm building and be on the hook for user complaints about missing map data, and have to do that myself.
Not to mention, contributing to OSM in the strictly legal way is not at all trivial - for legal reasons you simply cannot consult any other map, so basically you'd have to go out, walk/drive/bike around your area with paper and pen in hand and "map" it yourself. That or trust you have perfect memory w.r.t. your city. The bar for contributing back is high, can we blame people when they'd rather just have a map that worked?
> This seems rather disingenuous. A map that is missing major areas is not only not well mapped, it clearly "does not work".
It is very rare for OSM to be completely bare in a major metropolitan area (especially in the US). It could be that the addresses are gone, but at the very least the road network will be there from the TIGER import.
> I'm not a cartographer, I have no interest in being one. I certainly am not signing up to integrate OSM into what I'm building and be on the hook for user complaints about missing map data, and have to do that myself.
You don't have to be a cartographer to add data to OSM. You sign up, point and click or drag and drop. It takes minutes and you don't have to have any particular knowledge before hand (especially cartography). Having said that, I completely understand not wanting to be on the hook for user complaints. That's why Google charges thousands of dollars. Some people (the writer of the blog post in particular) aren't interested in paying that.
> Not to mention, contributing to OSM in the strictly legal way is not at all trivial - for legal reasons you simply cannot consult any other map, so basically you'd have to go out, walk/drive/bike around your area with paper and pen in hand and "map" it yourself. That or trust you have perfect memory w.r.t. your city. The bar for contributing back is high, can we blame people when they'd rather just have a map that worked?
This is not true. Mapping with paper and pencil on foot or bike or car is a great way to get outdoors and thousands of people use this technique to add to OSM every day, but you can also use any one of the dozens of aerial imagery backgrounds available to help you create data from memory.
> "It is very rare for OSM to be completely bare in a major metropolitan area"
I don't think anyone claims OSM is simply missing entire cities - but entire developments in cities are definitely missing in many cases.
> "You don't have to be a cartographer to add data to OSM. You sign up, point and click or drag and drop."
Not true. OSM gives you a handy satellite image reference, so you can certainly fill in streets - but unless you're from the area, you are poorly qualified to decide what type of street it is, and you certainly don't know the street names. In other words, the common contributor is empowered with little more ability than to trace satellite imagery, with no knowledge of place names, street names, or landmarks that are necessary to build a real map. In fact, the only places that the average contributor is qualified to map is the area directly around themselves.
This may work for a casual contributor who wants to see their neighborhood on OSM, but it's entirely unrealistic for a developer who wants to support his/her users.
Consult another map? That violates the terms for contributing to OSM - for good reason, they want to steer absolutely clear of any allegations of plagiarism or copyright infringement.
So if someone complains that my app has missing map data, unless the user is located near me or one of the areas I'm intimately familiar with, they're SOL until someone else fixes it.
> "Mapping with paper and pencil on foot or bike or car is a great way to get outdoors and thousands of people use this technique to add to OSM every day"
This makes no sense at all. I get the spirit of the project, and I fully support it. In fact, I have contributed to OSM. But expecting this out of most developers, or even a significant portion, is lunacy.
This is like telling your IT manager "we should use this very incomplete library. The code is open, so any missing functionality or bugs we can fix ourselves! This will be a great benefit to our coders, so they can crack their knuckles on some problems they don't usually get to work on."
Which is entirely true, but misses the point by a mile - developers may be interested in contributing to OSM (I know I am), but how many are willing to ship this in a production environment, where the goal is to build a great product, not baby-sit an incomplete data provider?
> This is like telling your IT manager "we should use this very incomplete library. The code is open, so any missing functionality or bugs we can fix ourselves! This will be a great benefit to our coders, so they can crack their knuckles on some problems they don't usually get to work on."
There are plenty of organizations that do this. Google spends thousands of developer hours and millions of dollars contributing to open source projects. And it's not just for the developers: they recognize a better resource and use it. When their needs outgrow the library's capabilities they add to it.
There are at least a few companies doing this exact same thing for OSM data, too. For example, MapQuest and Bing both use OSM data in some of their products and contribute data and tools/source back to OSM.
> ...developers may be interested in contributing to OSM (I know I am), but how many are willing to ship this in a production environment, where the goal is to build a great product, not baby-sit an incomplete data provider?
As I've previously mentioned, OSM is certainly inadequate in many places (and the community is working really hard to fix that -- thanks for contributing!), but in some places it's _MUCH_ better than Google. It's certainly a business decision: do you want to pay Google $10k+/month (where it was $0/month not 2 months ago) to run your site or do you want to use OSM? If your business's primary area of interest is Europe (especially Germany), then you probably want to use OSM. If you want to cover the entire world or don't want to deal with the lack-luster-but-improving data in the US, then use Google.
I'm not trying to do the common "patches welcome" snark, but I'm trying to help. The poster might have some actual problem (eg "osm can't do directions") that I might know the answer to ("cloudmade do directions based on osm data").
Although osm contributing can be a lot of work, its very easy for lots of people to add some data, esp. about where they live or grow up. Is there a coffers shop, school, or path near where you grewbup that's not on the map? Every little contribution helps. People are often proud to see their home correctly represented on the internet.
Many of your complaints are mostly valid if there is a totally accurate alternative map available, but its not the clear cut. In many places osm is of better quality than Google maps, and considerably cheaper than the national mapping/ordnance survey agencies. There is no perfect solution.
I live in South Africa and I haven't found a place (yet) that isn't perfectly mapped. I checked various far-flung African countries for a customer who was looking to move from Google Maps and they all seemed fine. So, the map does 'work' for me, just not in the U.S. This chicken and egg isn't going to spring out fully formed, so either U.S. mappers get out there, or this situation will continue.
Obviously developers won't start driving around in cars to map things out, but I wonder why the map-oriented communities haven't supported OSM? And surely map-oriented developers could start an initiative to highlight less-well mapped areas?
Now if only I could use Twilio and Google Payments here. We can't even buy music from iTunes, despite everyone having iPods and iPhones for almost as long as they've been around, and Apple stores in most major malls.
I really like the Open Street Map project. I think it's shameful of Google to create Map Maker, which essentially copies the functionality of Open Street Map, however Google don't release the user contributed data under a reusable license, although I believe they do in some poorer countries.
The article has a very ingratiating tone towards Google. The summarized version is: Google started charging for Maps usage, their sales team sucks and their pricing is too high. All of that could have been said without the unctuous praise of Google's work on geo technology.
Should a startup which uses so much of Google Maps that it needs to pay be complaining that about poor salesmanship when it can't afford to pay? It sounds to me that both companies have monetization issues.
we're more of a small business than a startup. We've been going for 6 years and are nicely profitable. The issue is the prices quoted were more of the level for large corporate multinationals - hundreds of thousands of dollars/year just for our UK service (one of eight that we operate).
Read the post again, you are missing the point. The price itself was the last drop. The overall experience they got, for a company that has been featured on Google Maps promotional publications, was abysmal.
I totally agree that the sales process was bad. My post was responding to politician's insinuation that because they couldn't afford to buy the data they had no right to complain about the embarrassing salesmanship.
It's simply interesting that their use of the service was such that they were asked to pay, yet when they couldn't afford it, they complained about the way in which they were asked.
Suppose your government provides electricity for free below a certain threshold. Suppose you exceed that threshold. Suppose a rude person calls you up and asks you to pay. Suppose you can't afford it.
Should you then complain that they were rude and how it was their rudeness that caused you to switch providers? Or do you simply explain that you found a cheaper provider that aligned with your long-term goals? Was the point of the post to try to give Google customer service a black eye? Because, to me, all it did was betray pettiness.