[Talk-GB] Non-intuitive addresses

[Colin Smale]

> On 02/12/2022 8:27 PM Robert Whittaker (OSM lists) <robert.whittaker+osm at gmail.com> wrote:
> 
>  
> On Sat, 12 Feb 2022 at 18:54, Colin Smale <colin.smale at xs4all.nl> wrote:
> > On 02/12/2022 6:32 PM Paul Berry <pmberry2007 at gmail.com> wrote:
> >> I'd map that as:
> >>
> >> addr:street=Abbots Walk
> >> addr:parentstreet=Boat Lane
> 
> Me too, I think.
> 
> > But that would be patently incorrect, as Boat Lane is not part of the postal address.
> 
> It may not be part of Royal Mail's postal address for that property,
> but we're not allowed to use that information in OSM. Based on how UK
> addresses usually work, and the fact that there is no actual street
> named Abbots Walk, I think it's logical to assume that Abbots Walk is
> somehow subordinate to Boat Lane. Hence Boat Lane should be included
> in the address. I don't think it really matters whether or not Boat
> Lane is part of Royal Mail's version of the address. If it makes sense
> to add it to help users, and to reflect the reality on the ground,
> then I think that's fine.

However logical it seems, it would not be correct. That it may be helpful on the ground for navigation/orientation purposes, is irrelevant, as we map postal addresses (as used by RM etc. to deliver post to the correct letter box) and not physical addresses. Use of "addr:substreet" in this way is "tagging for the renderer" in the sense that it is not correct, but is chosen to aid a particular use case of the data consumers. The UK's somewhat opaque system of addressing is not a problem that OSM is going to fix. Reality is that there are multiple ways of "addressing" a property.

You mention "Royal Mail's version of the address" - in practice theirs is the only version of the postal address. Of course DHL, Amazon etc are in theory free to invent their own system of addressing if they want. I suppose what3words and Google's Plus-codes could be considered alternative postal addresses, if you could find a delivery service prepared to accept them as a destination.

The statement that "there is no actual street named Abbots Walk" is presuming a specific source of which streets exist, and which don't. The council for example may well have a different view on that from RM. The point is, it is irrelevant what other sources say, it's the correct postal address for these houses. I was actually thinking of addr:street:not=Boat Lane ...

How about knocking on the door of one of these houses and asking the resident how they write their address? Is there any other evidence, anything at all, that "Boat Lane" is in normal usage as part of the postal address by anybody at all? After all, that's the kind of question we would ask about the value of name=* where we are looking for the name that represents common usage.