[Talk-GB] OSM UK address project: tags - Updated guidance

[Rob Nickerson]

Evening.

Just to clarify, as there does seem to be some confusion today:

The majority of addresses will be simple housenumber/name and street name.
It is the cases where a child+parent relationship is needed that is causing
the trouble. Options discussed are:

1. Child = addr:place, Parent = addr:street -> discounted because these
tags shouldn’t appear together.

2. Child = addr:terrace, Parent = sometag -> Already being used by some
mappers in the UK. Only works if the child feature is a terrace therefore
is not feature agnostic. Some mappers have suggested deprecating use of
addr:terrace in the UK.

3. Child = addr:street, Parent = addr:parentstreet -> Already being used by
some mappers in the UK. Only works if the child feature is a street
therefore is not feature agnostic. Note that some mappers have suggested
using this even if the child feature is not a street but others have
disagreed and said addr:street should only be used for highway=* objects.

4. Child = addr:substreet, Parent = addr:parentstreet -> The child tag is
proposed as a feature agnostic tag that overcomes the concerns raised about
the feature specific tags in 2 and 3. Unfortunately introduces new UK tags.

5. Child = addr:place, Parent = addr:parentstreet -> We know that currently
the addr:place tag is very misused in the UK (and globally) as mappers mix
it up with addr:suburb.


Of these, 4 and 5 seem to be the ones being most discussed. I do understand
the idea behind option 5 as it is very close to my original proposal (just
with a different parent tag) and only introduces one new UK specific tag.
However, it is not without its issues. Primarily we have seen that
addr:place is very badly misused in the UK with a long way to go to clean
up the tag [A]. Given the UI of some editing tools, I am not sure we will
ever properly clear addr:place up – after all what proportion of
contributors are reading these lengthy talk-gb posts or the wiki pages!
Furthermore it would introduce a weird nuance to the definition of place;
namely “addr:place is meant to be used when the address does not reference
a street at all, except when the street is a parent street and you tag it
with addr:parentstreet”.

Consequently I believe that option 4 is a much more pragmatic solution. We
will know exactly what people mean when they use this tag, which is not the
case for addr:place. In a way, this reminds me very much of the time when
we were struggling with Public Rights of Way mapping. There were those that
argued for an approach based on the original meaning / intent of a tag, but
at the end of the day, the only way to pragmatically solve the issue was to
introduce the designation=public_footpath (and so on) tags. These removed
all ambiguity of the other inconsistently used / interpreted tags.

I appreciate that this runs counter to Sarah’s note that “Data users will
thank you if you go for one or the other, not both” but I cannot help but
think that some data users will also thank us if we use a tag that is
consistently used. I suspect that some use cases will require this
consistency more than others.

We have a little time before the OSM UK address editor is ready so it is
always possible to revisit this decision if we make big inroads into
cleaning up the use of addr:place. (I will try my best to fix any wrong
addr:place tags I have introduced).

It’s never nice to end on a compromise, so it is worth reflecting on some
of the wins we have made along the way:

    1. We’ve learned a lot about addr:place and some mappers have already
started correcting their edits to use addr:suburb in cases where this is
now understood to be the right tag.
    2. We’ve (re-)discovered the meaning of addr:city which should result
in less use of addr:village, addr:town (both not globally accepted tags).
    3. We’ve identified a whole range of tags to avoid in the UK
(addr:locality, addr:site, addr:subdistrict, addr:district, addr:province,
addr:state, addr:country) and some to possibly avoid.
    4. We’ve become aware of the nominatim QA tool which should keep us on
track in the future and help us to fix our errors of the past.

Finally I want to apologise to anyone who feels they have come out on the
wrong side of this compromise. Like I said in December “we need to be
realistic that there is no silver bullet that solves everything without
breaking anything”. If you feel that you’ve been caught in the crossfire of
my non-silver bullet then I am sorry. Hopefully we can all chuckle about
this when we have mapped all 27 million residential addresses thereby
making one of the UK’s hold out closed datasets open to everyone :-)

Thanks
*Rob*

[A]
https://nominatim.org/qa/#map=8.06/51.79/-2.08&layer=addr_place_and_street
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-gb/attachments/20220109/6f09640a/attachment.htm>