[Tagging] Adding housnumber the lazy way.

Alan Mintz Alan_Mintz+OSM at Earthlink.Net
Tue Dec 22 12:01:07 GMT 2009

At 2009-12-22 02:07, Erik Johansson wrote:
>On Mon, Dec 21, 2009 at 10:25 PM, Alan Mintz
><Alan_Mintz+OSM at earthlink.net> wrote:
> > At 2009-12-21 11:01, Roy Wallace wrote:
> >>... If you don't know where the other end of the
> >>street is, you can't use an addr:interpolation way, so it seems to me
> >>that you are just tagging a sign.
> >>
> >>Is there already a tagging scheme for this? If not, propose one - but
> >>(as others have said) don't use existing tags in a way they are not
> >>intended for. (btw, please don't follow up with "but I want it
> >>rendered..." :P)
> >
> > I've been "tagging the sign" from survey photos, with address nodes to
> > which I add the tag pseudo=yes. When you get info for adjoining
> > intersections, they could be used to construct a true picture of the range
> > of possible addresses.
>Shouldn't that be psuedo_position=yes, or some thing describing that
>you don't know the accuracy of the node you have entered?

It's not really the position - the address itself is not real. It is the 
beginning (or end if you like) of the range of addresses known to start on 
the nearby corner.

>Here is one of your nodes:

And the pic from which I tagged it: 

Using the knowledge that even numbers are on the south side of the street 
in this city[1] and the position of the sign (on the north side of the 
intersection) from the GPS track, I marked the pseudo-address 698 on the 
south side of the street, just west of the intersection, as being the most 
easterly possible address on that block. Similarly, the starting addresses 
on the other three corners of 13th St are marked 699, 700, and 701.

If the next intersection west along 13th St (5th Ave) were surveyed, it 
would likely read 500 (left) and 600 (right), at which point that corner 
could be tagged and we would know for certain that the range of possible 
addresses is 600-698 on the south side and 601-699 on the north side.

> > This area shows the results of a survey of both pseudo-addresses (from
> > street signs) and actual ones (from mailboxes): 
> http://osm.org/go/TaBihQXG4-
> >
> > (Yes, I need to discuss/document this. I suppose this is the "discuss" part
> > :) )
>I'm inclined to mark the position as inaccurate and some tag to be
>able to put an interval there as well.. The current scheme with
>drawing a way to interpolate is too much work and cumbersome, for me

I agree it's cumbersome. The interval is not definite - only that it be at 
least 2 because of the spec of being odd on the north/even on the south. 
There was some discussion a couple weeks ago that the Karlsruhe schema 
implied that all addresses within the given range were actually present, a 
scheme that would not be realistic anywhere I have travelled. As I think I 
wrote back then, most places seem to use a relative position within the 
block to assign the actual address. Even within fairly uniform tracts, 
there will be non-standard intervals between some adjacent properties to 
account for slightly smaller/larger lots, rounding, different driveway 
positions, etc.

It does seem that these address nodes will need to be associated with the 
road they sit beside _somehow_ in order to make it useful with reasonable 
efficiency for any sort of navigation.

et seq

Alan Mintz <Alan_Mintz+OSM at Earthlink.net>

More information about the Tagging mailing list