[OSM-talk] The world’s best addressable map
Jóhannes Birgir Jensson
joi at betra.is
Thu Oct 23 17:55:22 UTC 2014
Thanks for creating this project in the first place Steve.
In an interview earlier this year you said pretty much the same thing. I
believe this is a vital step forward in the First World countries where
other competing solutions, both local and global, are not delivering enough.
However I disagree that this might be THE key issue. Where we are
lacking even more at the moment is the world we knew previously as Third
World but as Hans Rosling has demonstrated that name does not mean
Let me take Botswana as an example, a project I've been trying to get
rolling for the past year as Mapping Botswana. They do not have many
addresses at all, in the newest areas they are creating US-style
cul-de-sacs and putting addresses and streetnames but in the capital
itself, Gaborone, the older areas are designed as plots, arbitrarily
sized areas that encompass many buildings so your address will be Plot
525 and then you need to look for the correct sign to find the correct
building. This plot data is currently not under a license we can work with.
Elsewhere the rural villages are a mishmash of roads and paths and
arbitrarily placed buildings mostly, with no street names or anything.
The only thing we could use there is a census number that each
residential building is supposed to be assigned. I doubt they were
intended as addresses but we are looking into it.
The offline apps and tools are vital here - in Africa, Asia and Latin
America where mobile networks are still slow, still unreliable for
coverage and data usage often pricey. Not to mention the map coverage is
often limited to a name on commercial maps. This is where OSM makes a
I live in a country where Google has already StreetViewed most of it, a
local service ja.is has imported all official address data (including
bad data) and they also made their own version of StreetView called
360°. In Iceland we are up against corporations who are doing their
utmost to make a good map and so we try even harder to be better. But we
don't kid ourselves, if Iceland were deleted from OSM then there would
still be good online maps from these other providers. The offline
feature starts to give OSM an edge - they are something we should strive
to make better, something done by many app makers, some of the good and
some of the poor.
If however Botswana were deleted then there are small parts of Botswana
who would still enjoy pretty decent coverage on Bing and Google but all
the rural areas, villages and hamlets, are not there. The HOTOSM
projects and related ones are what is giving OSM the bite, in my view.
Delete the ebola-affected areas from OSM and you set back local efforts
and local knowledge with devastating results. Delete New York City data
and you can still get around on Google or Bing or whatever, although you
are missing out on many great improvements that have been made.
This is my view of OSM, it matters most where there are no other
Þann 23.10.2014 16:56, skrifaði steve at asklater.com:
> I'm adding OSMF-talk since it concerns what I outlined in the
> original “vision statement” email.
> I was perhaps too specific and jumped ahead saying “world’s best
> addressable map”.
> What I really mean is the “world's most complete open map”. There are
> three pieces to a modern map. There's the display piece, the routing
> and the geocoding.
> We won the display piece. It looks great. We are ok at routing. Not
> perfect or great, but ok. We're really lacking on the addressing. If
> we can get addressing even to the “ok” stage then a lot more people
> will use OSM, which means more editors, more community and more data.
> This is because the main use for maps today by the public is to get
> somewhere, and we can't help with that without all three pieces. Right
> now we have 2/3.
> I jumped ahead because I see this every day, and I understand not
> everybody does. I think all the other things are good too, even every
> tree in OSM! I just know that if we had to pick one thing to focus on
> it would be addressing, as it will get all the other things to happen
> faster too. But that doesn't mean you can't add trees in to OSM at the
> same time, just that the shortest path to getting more of everything
> is to get more of addressing.
> Also let's be clear - addressing isn't easy. It's complicated and
> hard. But that's a good goal to have, and OSM was complicated and hard
> in the first place.
> *From:* Oleksiy Muzalyev <mailto:oleksiy.muzalyev at bluewin.ch>
> *Sent:* Thursday, October 23, 2014 2:53 AM
> *To:* Jóhannes Birgir Jensson <mailto:joi at betra.is>,
> talk at openstreetmap.org <mailto:talk at openstreetmap.org>
> I agree that addresses is a complicated field. There are different
> historical systems, there are cities where even many streets are
> without names, etc. There is a lot of space for innovation, certainly.
> What I meant is that it is not obligatory to map a city or a town
> addressable from one end to another, one house after another, or wait
> until a municipal government releases into public domain its database
> of addresses (which may be not without errors or omissions too).
> If there are, say, 10% of buildings where 90% of the population lives,
> studies and works, it makes sense to map them addressable first. Often
> these are large modern buildings with clear addresses.
> And it is much easier to return into the same area for the second
> time, when there are already at least some large buildings with
> numbers, much easier to orientate oneself.
> I see from your example that in the city of Reykjavik almost every
> building has a number, so you have a more advanced set of priorities.
> Best regards,
> On 23.10.2014 10:39, Jóhannes Birgir Jensson wrote:
> I like addresses but they don't behave like you would think. For
> example we have a part of a street that has each individual flat
> as its own address number. We first used the number;number;number;
> approach but I'm now in favor of naming the house what it says on
> the front (the range 37-51) and then put address nodes on the
> building so it appears in search, with roughly the position
> accounting for where in the house the apartment is. In this case
> the numbers closest to the street are at the bottom floor (the
> stadium approach I favor). I'm in favor of moving this same method
> over to the other houses.
> As for being able to search within a specific town or area then I
> think we should look again at relations and super-relations. You
> could group streets relations into a neighborhood relation and
> then into a town or municipality relation etc. This of course
> works very differently based on country but for Iceland I can't
> see us hitting any limits.
> Regards on behalf of the Icelandic Local Chapter applicant,
> Þann 22.10.2014 18:28, skrifaði Clifford Snow:
> On Wed, Oct 22, 2014 at 5:04 AM, Oleksiy Muzalyev
> <oleksiy.muzalyev at bluewin.ch
> <mailto:oleksiy.muzalyev at bluewin.ch>> wrote:
> It is not necessary to put down a number on each building.
> It is possible to use /addr:interpolation/ (/odd, even/,
> or /all/).
> We put down a number on the first building, then on the
> last, connect them in JOSM, and add /addr:interpolation:
> all /. For example here:
> http://osm.org/go/0CFn0AZ_d--?m= . It is also very useful
> on a street with many small houses. And it is searchable.
> For example if there is number 15 and number 27 on the map
> for a street, and they are connected with
> /addr:interpolation: odd, /and//if one searches number 21,
> the map will show the number 21 all right.
> Then, there is another approach. We first map addressable
> large building, where a lot of people live or work. Kind
> of of going after the low-hanging fruit.
> talk mailing list
> talk at openstreetmap.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the talk