Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Lost city in Darfur (openstreetmap.org)
229 points by Noelkd on Jan 9, 2014 | hide | past | favorite | 70 comments


I don't think of myself as a luddite in any way, but one can't help but get the feeling that the image of the world brought to us by huge tech companies has gigantic holes in it.

When you spend all of your time inside of that image of the world it's difficult to see the gaps. I love this. It's as if he walked in to a void.


This is more a shift in perception of what online maps are taken for. Online maps were traditionally made for one purpose only, car navigation, not to show a complete picture of the world.

And the only commercially geodata available in digital format, from a few specialized companies, were for car navigation in lucrative markets (like the US and Europe). It wasn't a given that online maps work properly or show more than base coverage outside those regions.

Only recently maps providers started integrating user created content. OSM as a whole is also surprisingly sparse outside those established markets.

Just a reminder, back in 2007 Mexico wasn't even covered by Google.

http://google-latlong.blogspot.de/2007/09/more-of-world-for-...


> Online maps were traditionally made for one purpose only, car navigation, not to show a complete picture of the world.

That tradition didn't last much longer (if any?) than the current state has already lasted.


Microsoft generously provided the imagery he was using.

(Well, probably. Anyway, big tech companies are doing a lot of the work that goes towards satellite and aerial imagery being easy for people to access. So are governments. It just isn't as clear cut as you stated it.)


Of course, of course. I meant 'image of the world' in the broader sense — sort of like the "OK" from Mr. Penumbra's 24 Hour Bookstore.

I didn't mean to imply that they're purposefully ignoring things, but that it's difficult to create a methodology for mapping the world that doesn't ignore features of it — either by accident or as a rule.


Arguably, such a methodology would be unusable anyway:

http://en.wikipedia.org/wiki/Bonini's_paradox


state is talking about something more like fully capturing a limited set of the features that are visible at some given scale.


map != territory is the traditional phrasing.


Couldn't agree more. Its like a Doctor Who episode.

- It shouldn't be there?

- Try telling that to the local population...


The context is "unknown to online mapping projects", not "unknown to humans". He's saying "This thing is missing from the catalog".


Yes, but did you know it was missing?

My comment was to agree with the concept that its easy for me to blindly believe in what the big tech companies publish. Mapping or otherwise.

I love Doctor Who episodes where on the surface everything looks normal, but on closer inspection things are very different (eg Weeping Angels or The Silence).


"one can't help but get the feeling that the image of the world brought to us by huge tech companies has gigantic holes in it."

Compared to the image of the world you get from your chair without those tech companies, I wouldn't describe it as 'holes', but as filling in 99% of the white areas.

Columbus didn't create a huge hole in the wolrd's map


> the image of the world brought to us by huge tech companies has gigantic holes in it

When I was a kid I liked to go buy USGS maps of the area around where I lived. While what's online is certainly not perfect, I find the fact that I can get detailed, free maps of almost the entire world nothing short of amazing!


I'm a Google employee, but don't work on maps. I also don't speak for the company... and so on and so forth.

If you zoom in on Google Maps, you can see that the city is marked as Gereida, so it's not the lost city he's claiming it is. Now, I'm happy to listen to people who think it should surface at higher zoom levels, since there's nothing around it, though I suppose that's a feature request for the maps people.

Cheers, Doug


It's marked as a PoI of type "City Center", potentially from a user adding it. It is not in your data set as a city and does not render as such, which is why it disappears at higher zoom levels. This also explains why people down-thread cannot forward resolve the city's name but searching for it works once a small-enough bounding box is established.

In fact, it's this PoI: https://plus.google.com/108972490756349810804/about?gl=us&hl...

So no, Google does not render this city and it's probably inaccessible via your geocoding tools. It's the equivalent of a restaurant from Google Maps's perspective. It goes without saying I feel weird explaining this to you. :)


I opted to give the city a 5-star review. Maybe now it'll show up easier. :P


    tobyjsullivan 4 hours ago | link
    I opted to give the city a 5-star review. Maybe now it'll show up easier. :P
A+++++ would be internally displaced again [1]

[1] http://www.oxfam.org/en/programs/emergencies/sudan/sudan_ger...


There is a road going through it, but the city is unmarked. The conclusion in the forum is that this city is Gereida, but a search for it doesn't return anything in GM.

https://www.google.com.br/maps/preview#!q=11%C2%B016%E2%80%B...


One interesting thing is that the search for "Gereida" does work, but only if I'm already over Sudan. I guess it needs more context to do the search, but "Gereida, Sudan" isn't sufficient.

edit: it does work anywhwere if you search for "Gereida near Buram, Sudan", but I don't think anyone would ever search for that.



It is named on the map view (non-satellite) for the 3 highest zoom levels. I also doubted dmayle's allegation, but did find the label after a minute of searching. I agree with the suggestion that the label should show up on lower zoom levels.


Hmm, I'm using the new Maps "preview" interface, looks like the city only shows up in the old version.

edit: the marker appears and disappears in a seemingly random way, and goes away completely after zooming in satellite mode.


Was this true as of when they had the discussion? Note that the discussion took place in December, and they subsequently added the city to OpenStreetMaps and to Wikipedia. Someone may have added the point of interest to Google Maps since then, or Google Maps may have automatically pulled that information from OSM or Wikipedia.


This made me look up my local area on OSM. The streets are there, but most of the residential streets are not named. Time to do some editing tonight!


It is clearly visible in the satellite views on Google Maps, but the map view only shows the main road going through it. I wonder how many other such places could be found just by browsing the web? This seems to be what this guy is doing. If you look at older posts of his, he is working on automatic road following in satellite imagery.


On Google Maps many places have incomplete map data. If you get a bit farther out of Accra, Ghana, quite a few streets will just not be there. That was quite obvious when my sister was there for a few months two years ago but it seems to have gotten somewhat better. Accra is not in the middle of nowhere, it’s a major and important West African city (and 2.3 million people live there).

Maps are imperfect images of reality and keeping them current costs a lot of effort. Depending on who is likely to use them maps will be more detailed in certain places – especially maps that aspire to provide global coverage.

In Accra its missing roads in the suburbs, next to the place I grew up (a small town in Bavaria) its a few (but not all) footpaths in the park missing and a bad match of satellite image and map data (with streets being a few meters off in different directions, a problem that strangely doesn’t seem as prevalent in Accra).


My friend Dr Bayliss did just that - 'discovered' an African rainforest, Mt Mabu on google earth, interesting read/chap http://www.theguardian.com/science/grrlscientist/2013/nov/08...


That's pretty cool actually.


I live in a suburb of Sydney, Australia and in this area most of the residential streets on OSM are unnamed. So my assumption would be that there is a LOT of metadata waiting to be collected in most (all?) cities across the entire world.


More than likely the roads were traced from satellite/aerial imagery and there's no source for road names. If you're in that area you might want to use the Edit button to add the road names (it's really really easy now). If you don't want to sign up for an account, you can use the "Add Note" feature in the lower right and do it anonymously.


See my other comment in this thread. I'll be adding in a ton of names this weekend.


In the last thread comment he's created a Wiki article for the place including a couple of interesting links:

https://en.wikipedia.org/wiki/Gereida


I'm not sure unmapped (on google/OSM) cities are rare at all. My daughter was born in a town in Ethiopia and it's not on any map that I've been able to find while it is listed (without lat/lng sadly) in wikipedia.


If there was only one unmapped city it would be considered rare yet there would still be thousands and thousands of people who could share your anecdote. Adding yours to this thread (albeit without any information about it) brings the number to 2 - for all I know there could be hundreds of them, but there could also only be those 2.


For much of the world no-one has ever done large scale mapping. The Soviet Union and USA produced military mapping and gazetteers for targeting, but it often lacks good on-the-ground knowledge. In rural areas it can be difficult to even get a consistent place name from people who navigate from memory.


Someone in the linked comments talks about adding multiple other cities near this one.


if you can, please add it to osm and add the latlong to Wikipedia!


I'm possibly going to sound dumb here, but, he said he was mapping the area and found it... How was he doing this?

Looking at aerial data/pictures and couldn't find it in a matching map or something else?


Yeah, from the way it was written it seemed to me that he was actually there, taking gps measurements. So I was thinking 'why doesn't he just ask somebody there' until I was a few paragraphs in.


go to osm and hit the edit button! you will see the raw map data displayed over aerial imagery. I guess that's what he used.


the airstrip was built by my friend martijn (https://github.com/MartijnR) when he worked as a construction expert with the Red Cross in Darfur. He had to "hire somebody to chase away goats before plane landing".


relevant: In case anybody is interested. The NGA keeps a list of placenames. I don't know if this place is on here, but it's a good resource.

http://earth-info.nga.mil/gns/html/namefiles.htm


the place has also been in geonames for a while.

http://www.geonames.org/8504621/gereida.html#tab-c

[Addition]

A list of this type is traditionally called a Gazetteer

http://en.wikipedia.org/wiki/Gazetteer

The Gazetteer of Sudan: Names Approved by the United States Board on Geographic Names

as well as the file linked under the parent of this post lists this place as Qureida/Quraydah as a well, with Sa‘dūn as the name of the nearest populated place

http://books.google.de/books?ei=PkzPUsKSE8LOtQbc5IGYCw&id=DK...


It is a good starting point, but can be wildly inaccurate in places.


Look it up in Google Earth with the USHMM layers on. Lots of "destroyed villages" are marked in the surrounding area. Interesting how this city survived while so many others nearby didn't.

Also, the imagery shown is from 2006. Something more recent would be helpful.


Bing reports November 2012.

(The capture date is reported in an http header)


Last year "Rostov na donu", a piffling little village with the Mil (helicopter) factory and several million people living in it had only two roads, according to google.

OSM had it, though.


Not a problem for locals who probably still use 2GIS or Yandex maps.


Still? 2GIS is so much superior to Google Maps, it's not even funny.


That's why they still use it! :)


I'm surprised there are towns this large not on maps. It seems like it would be trivial for a large company like Google or Microsoft to write software that scans satellite imagery and cross checks it with current maps to see if anything large is missing. In fact, I would be surprised if they don't already do that. Although I could see this area of the world not being very high priority.


After reading the helpful wikipedia article posted elsewhere in this thread, it seems like the "city" is (or became) a refugee camp. So on old maps (the paper kind digitized by UT--and probably known to Google) it was just a wayside or a small market with a few houses. Geopolitical events led 10's of thousands of people to amass and settle there in a short time. Few places change that fast, but it's not unknown: mining booms and subsequent ghost towns (in a ghost town, the satellite image makes it look larger than its population really is).

This is further reminder that maps are not ever complete, nor intemporal. The human world (roads, buildings) change quite rapidly in both growth and decline.


So you find areas that are missing, then what? Fly somebody there to check out the street names? Remote sensing is a big field, of course they know about naive cross-check techniques like this, the question is what do you do when you find error. And the answer is: you focus on improving the thing that are most relevant to your business first, and war-torn African countries are rather low on that list.


Seems like you could apply some image classification techniques to automatically identify uncharted cities given access to all the satellite data.


Something like "here's a list of changes (sub cat: possible settlements)"


Weird. Expected a spider web, not a grid..


The satelite view is a bit less organized:

http://maps.google.com/maps?q=11.28+25.14&um=1


For more perspective here's a photo of the village - which appears to be more of a camp:

http://explore.org/photos/1725/gereida-sudan-village

In the satellite image I think it's safe to assume each white block is a tent and each dark outline is a fence. There are many dark outlines with no tents showing where the old population has moved out.

The wikipedia article explains the sudden changes in population of the settlement. It would probably explain the grid layout too.

https://en.wikipedia.org/wiki/Gereida


It looks like the north-west part is the actual town and to the east and south is where it expanded as a camp. Of course, now there is more town than (what I assume was the original) town.


+1 for mentioning this. Here is Khartoum the capital of Sudan.

https://maps.google.com.br/maps?t=h&ll=11.274876300000004%2C...

You can see that almost all of it is a nice grid. Makes me wonder why. It's quite amazing despite their poverty, they were able to organize the buildings.


Is the guy writing this thread physically in Sudan mapping this place or is he doing it from GIS imagery?


I ran into him the other day. He is not in Sudan... he is in Berkeley CA, USA :)


From the little I know about the OSM project, you have to physically be in a place to map it out.

Especially a place like this, where most of the aerial photos/maps of the area are decades old.



GPS coordinates for anyone who wants to take a look: 11.2811075, 25.1414136


Here's a cool lat/lng pair: 40.2526, 58.4394. Check it out on a satellite view.


Let me google that for you: http://en.wikipedia.org/wiki/Door_to_Hell


Why would Gereida, a city with a population of 120,000 be considered a lost city just because it is not on an electronic map?The question 'Anyone care to name it?' sounds really patronizing.As if the people currently residing there could not possibly have a name for where they live.


He didn't ask people to make up a name for it, he wrote, "anyone care to find a name for the city?"

As in, can somebody do the research to find out what the city is called--by the people who live there.


Oh,sorry.I did not read that correctly




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: