[Talk-GB] boundary mania (was: 'historic' county boundaries added to the database)

[Frederik Ramm]

Hi,

On 08/26/2018 12:46 PM, Colin Smale wrote:
> It has gone all quiet here, and in the mean time smb001 has been making
> steady progress across England. 

I think he shouldn't have done this. He should have argued his case here
and the community should have come to an explicit resolution, rather
than one party creating a "status quo".

Personally, I am very much against mapping historic boundaries in OSM,
mostly because the exemption from the "on the ground" rules that apply
to current administrative borders (they are so important that we make an
exception) don't hold for historic boundaries.

But there's a general problem with boundary relations getting out of
hand. Take this little unnamed waterway here

https://www.openstreetmap.org/way/614127384

which is meanwhile a member of 19 different boundary relations:

* South East England European Parliamant Constituency
* The admin_level=8 boundaries New Forest and East Dorset
* New Forest West UK Parliament Constituency (4152802)
* Alderholt Civil Parish and Damerham Civil Parish
* Cranborne Chase & West Wiltshire Downs AONB (2664452)
* Dorset historic county and Wiltshire historic county
* an administrative region called "South West England" and an
administrative region called "South East England", both admin_level 5
* The Hampshire Constabulary boundary ("boundary=police") which exists
twice (relations 3999378, 8188274) if any proof was needed that this is
getting out of hand even for those who added it
* The Hampshire Fire and Rescue Service boundary ("boundary=fire")
* Hampshire County and Dorset County
* Hampshire Ceremonial County and Dorset Ceremonial County
* A statistical boundary called "Hampshire and Isle of Wight"

I have not analyzed these in detail and I won't make an attempt to tell
the readers of this mailing list which of these make sense to have in
your country. But I have a hunch that, say, the statistical boundary
"Hampshire and Isle of Wight" is not actually defined as a boundary. I
have a hunch that if the boundary of Hampshire were to change, then this
statistical area would also change - because it is *not* defined by
geometry, but just by reference to existing administrative boundaries.

I think we should all think twice before duplicating and triplicating
data in OSM just because there's yet another boundary that includes
Hampshire. We should find a way to reference existing boundaries instead
of copying them.

Practically all of the relations above have version numbers in the
hundreds, version numbers that have again increased when smb1001 did his
historic boundary mapping - of course he hasn't changed anything in the
statistical boundary "Hampshire and Isle of Wight" but still he's listed
as last modifier of this relation just because he has just split up a
way that was part of the Hampshire boundary.

I think if we continue heaping ever more boundary relations onto what we
have, we'll make things less and less understandable, less and less
maintainable.

But that's a general remark, not *specificall* aimed at history county
boundaries.

Bye
Frederik

PS: Of course, public transport relations are an even bigger culprit.
There are a handful of ways in OSM in England that are member of more
then 100 relations, mostly bus routes as far as I can see.

-- 
Frederik Ramm  ##  eMail frederik at remote.org  ##  N49°00'09" E008°23'33"