I’m giving a talk later today at the Salt River Project on the future of geospatial technology and though why not show what Google has been up to with their mapping product. So I did what anyone would do, zoom into Google Maps and see how it looks. When I did so I saw something that made me do a double take:
As you can see, there is a lake with roads across the top. I know that there isn’t a lake there, but a parking lot and a building. Of course the Google Maps aerial confirms this.
So what the heck happened? How could Google put a lake in the middle of a desert? Enter Legend City. Before my time in AZ, there was a small amusement park in Tempe. It of course failed and SRP bought the land and put one of their buildings on it. The park closed in 1983 and I’m sure the lake dried up or was drained in weeks.
Map of Legend City - See the lake on the left side of the map?
So I know you are thinking the same thing I was. Let us take a look at the USGS Quad map for the area.
So there it is, Google Maps are no worse than 30 year old USGS Quad maps. What makes it even more fun is you can search Google for Legend City and get everything but a phone number.
Google best clean this stuff up. I’ve gone ahead and “reported” the problem so hopefully it will be removed here soon, but how can one build routing and other online apps on an API that has data which is so inconsistent? I guess it is up to the community to fix Google’s maps for them.
35 Comments
Wow…Great find…I’m glad I’m not the only one finding surprises like this in Google Maps. Thanks goodness for aerials!
Wow, I used to live in Phoenix back in the late 60s/early 70s and used to go there. Never would I think that I’d see it in a Google Maps.
Google would appear to have some really weird datasets they’ve mashed together.
It’s not just Google but all of the data providers have outdated features in their GIS data left over from starting with old USGS data. Thomas Brothers still thinks there’s a pond in Santa Clarita that’s a parking lot. Google maps shows the same thing.
At least yours has historical consistency. This happened when they started using “their” data (the parcels are actually Charlotte County’s data, uncredited of course).
http://maps.google.com/maps?f=q&hl=en&geocode=&q=&mrt=all&ie=UTF8&hq=&ll=26.97731,-82.091861&spn=0.023559,0.035191&z=15
@Caitlin: In this case Google had it right before, but now its not. Pardon the routing.
Here you can see Edison Shell and Edison Days Inn placed in the midst of a residential neighborhood in the next town over:
http://maps.google.com/?ie=UTF8&hq=&hnear=New+Brunswick,+Middlesex,+New+Jersey+08901&ll=40.491299,-74.429455&spn=0.002309,0.004823&t=h&z=18&iwloc=lyrftr:h,2288301044450562741,40.490431,-74.429578
A person I know inside the google maps team recently discussed some of the work they are doing, not a “spatially trained” person, but some of the vibe was that they were playing with a lot of new data and systems and methods but really didn’t know a heck of a lot about what they were doing with it all, nor much clue about update issues and legal issues around quality of data. Amusing.
A great example of why Google should take advantage of some of the recent advances in automated feature extraction. Errors like this could be readily identified using an object-based expert system to extract water from recent imagery like the 2007 color infrared NAIP.
Would you rather that Google wait till everything is perfect before making the maps available? What’s strange is that they don’t have a big a$$ caveat slapping you in the face every time you look at a map rendered through their servers. I guess the little “Terms of Use” in the corner cover it.
Your link now shows “Place closed” Someone get a stopwatch on this?
I marked it closed this morning and it showed that right away. So lets just call it 8:30am MST.
Neato! It’s not a mistake it’s magic! By the way is the old Firebird stadium still out there near the SRP?
Yes, it was refurbished to be the Spring Training home of the Oakland A’s: http://en.wikipedia.org/wiki/Phoenix_Municipal_Stadium
The Firebirds moved to Scottsdale Stadium in 1992 and the Giants Spring Training went with them: http://en.wikipedia.org/wiki/Scottsdale_Stadium
Now of course the Firebirds are no more, moved to Tucson (became the Sidewinders) when the Diamondback’s came and eventually left to become the Reno Aces.
As I’ve mentioned to friends in other venues…SOURCE the data. It’s not difficult Google! Reminds me of how I needed to take a screenshot from a few years ago when Google (or TeleAtlas at that time) had a non-existent bridge to Cozumel from the Mexican mainland.
Cheers, -Matt
James, are you getting paid for your services?
I wouldn’t freely help a commerical company.
Amen, Brother!
shame on google for having an error in their map? it’s a rich set of spatial data, it’s gonna have errors. all in all they’ve done a good job. could it be better? sure. spatial data can always be better. the fact there’s an error is not really news. report the error, fix the error, move on.
I couldn’t disagree more, they had it right with TA and NAVTEQ.
it’s just, i can go dig into most databases and find errors. if you’re saying that the google layers have significantly more errors than ta or nav tech I suppose it’s a valid criticism of their recent switch, but to flag a single error or a set of errors in a national dataset and cry foul seems a stretch.
To me the question is not what the state of their dataset is today, but what it will be a few years from now. maybe they can do it better themselves, maybe they can’t–time will tell.
Tomorrow? Really? Tomorrow I could with the lottery and start paying for someone to drive me around.
I believe that Google is counting on quite a few people freely helping their commercial venture.
If you were to ‘monetize’ the notation process, what would such submissions be worth? I’d go with a half a penny or so per submission, given the volume of input might flow their way if they opened that box.
This is interesting. If you go to the USGS to find “the authoritative” map of the area, their map-finder tool is Google Maps. Chuckle.
They list two map products for the area: One is a 2007 ‘beta’ product comprised of an aerial photo, the ‘national transportation dataset’ – whatever that is, and place names from GNIS. The authoritative US Government map for the area was released in 1982, and was compiled from aerial photos acquired in 1978 and ‘not field checked’. This map still shows the lake…as it were.
So what’s better? Mediocre update rates and woeful products from ‘the authoritative source’ or the ability to interact and suggest improvements with an aggregator like Google. Who, by the way, shows every sign of listening to the consumer and using their considerable reach to make their content a dynamic product.
And what´s about the other online map services like Bing or Yahoo? Didn´t you check if they have the same error? Maybe people is finding errors in Google Maps because they don´t use the others. And BTW, I don´t know of any crowdsource error correction system in other providers, like Google has.
Google had it right 3 weeks ago before they switched to their own mapping sources. So they go from a 90% solution to a 70% solution and expect people to freely give them edits to make it 85%? How about they spend some of their cash on doing it right in the first place?
On the face of it, this appears to be a gross miscalculation/mismanagement on Google’s part. But then again maybe not. I wonder how many behavioral psychologists are in Google’s Ph.D. stable. Maybe the switch is the intended result of a thorough socio-behavioral analysis, correctly predicting the sheepish behavior of Google’s users.
See my comment below. Bing gets my address right. What is stunning about this error is that it is zip code based – zip 95762 has never been in Folsom, CA, or Sacramento County. And yes, they used to get this right.
From a thread at dslreports, it looks like google is using the 1979 USGS topo quads to build the “Terrain” layer (makes sense); and in some cases misconverted pink shaded urban areas (from the 1979 series) as pink shaded lakes. They then used the Terrain layer to create the base for the streets map. End result, a pink urban area on a 1979 topo becomes a lake on the 2009 google street layer.
To be clear – the ‘terrain’ used in the 1982 map produced by the USGS was compiled from aerial photographs taken in 1951. That’s 58 years ago. The revisions which show the lake were compiled from aerial photos taken in 1978. The actual map was edited and produced as an interim product in 1982. So that’s 31-year-old content that was published 27 years ago.
Now, is it a bad thing that Google is putting this ancient content up for public scrutiny and making it available for crowd-sourced improvement? I think not.
I think the real travesty we’re seeing here is not that Google is posting bad maps, but that the US government is so utterly pitiful at staying current with their mandate.
This thread really doesn’t show how bad Google maps are – it confirms the validity of their crowdsourcing tools and is a testament to their best intentions.
[I know, I know...don't get me started about their lack of sharing the content...that's the subject of another rant and another thread, one that's been flogged well on these pages in other posts.]
I live in El Dorado Hills, CA, zip 95762, in El Dorado County. Google maps shows Folsom, CA 95762, which is in Sacramento County next to us. Ouch. This is true even if I search for my street address and zip.
Accuracy is overated! (IMHO)
Google maps apparently has some even bigger errors, like creating a fictional city: http://www.npr.org/blogs/thetwo-way/2009/11/looking_for_argleton_so_are_th.html?ft=1&f=103943429
Yeah. My theory is that Argleton comes from data-mining – see http://www.systemeD.net/blog/?post=131 and also Mike Dobson’s recent interview at SearchEngineLand. This is shaping up to be really interesting.
Merely want to say your article is brilliant. The lucidity in your post is simply impressive and i can take for granted you are an expert on this subject. Well with your permission allow me to grab your rss feed to keep up to date with future post. Thanks a million and please keep up the ac complished work
As someone working in the mapping and GIS field in general, I think errors like this can slip by easily. While it’s true that having engineers and geographers run some basic QA/QCs to identify what doesn’t make sense (and they probably do to an extent), errors will always slip by seeing how much of the world an application like Google Maps is trying to map and present as a tool.
Also, I’ve noticed that water features is not a big focus for Google Maps. There are there as a reference more than anything. Their main focus are streets along with high resolution aerial imagery in most metropolitan areas around the world and now in many rural areas too. And even then, errors in streets occur but aren’t their streets from TeleAtlas or NAVTEQ, a different company? Also, at a touch of a button we have aerial imagery that anyone can use to confirm that lake, which is probably another reason they are not too worried about fixing, since you have a “2nd” map to confirm. Who would not love a perfect digital, interactive map?
Ed, they used to use TeleAtlas, but they dumped it for less accurate data. The streets are wrong in many areas and navigation is really unreliable. It is getting better, but it was working before. You may have a second map to confirm while you are sitting at your desk, but on mobile phones it becomes very difficult.
I agree, yes road data can be wrong, but with Google selling this as a navigable solution, I have to question their logic.
2 Trackbacks
[...] Blog posts of late from the geo crowd that point to errors in Google Maps (see Peter Batty, James Fee and Maitri Erwin). The mistakes that these observant industry watchdogs point out are a missing [...]
[...] been noticing two diverging themes lately when it comes to geo-stuff on the Web. First was all the buzz around Google ditching TeleAtlas and using their own data plus open data to support their base map. [...]