[Talk-GB] Non-intuitive addresses

[Mark Goodge]

On 12/02/2022 21:17, Andrew Hain wrote:
> I remain unconvinced that the road is called Boat Lane at all. When OS 
> Opendata was first released a quite unfamiliar spelling of a road near 
> where I was brought up was copied into OSM; in this case the discrepancy 
> with the addresses is a red flag against the accuracy of OS products.

It's definitely Boat Lane in this case; that's the name in the NSG. And, 
although we can't use the NSG as a source for the name, we can use the 
LineString in the Open USRN dataset and one thing we can see from that 
is that the line is a continuous single string. This implies that it 
also has a single name for its entire length (as if it did not, it would 
have multiple USRNs). So we can also infer that the name of the entire 
street is the name which appears on the name signs. Which is Boat Lane. 
And that also agrees with OS OpenNames. So, even without a surreptitious 
look at the NSG, we have sufficient confidence there.

The name of a street in OS open data does, sometimes, differ from the 
NSG. I'm aware of one such locally as well. But OSM practice is to 
follow the open data in such cases, and there's really no alternative 
unless there's good on-the-ground evidence to go for something else.

Mark