OpenStreetMap
Over the past few years, I’ve been a regular user of ITO World’s very useful map analysis and comparison tools. One tool (not unique, but well done) compares OS Street View data with OSM highway names and produces both a completeness report, and also a set of map tiles which show differences.
This diff layer is very useful in JOSM to spot errors in both map databases, be it simple typos or show areas for physical survey. In my area North of Newcastle, this has typically shown up schools being closed and turned into small development plots for housing which I’ve worked into my cycle training runs to make a physical survey.
As I know my local patch well, these areas are often already tracked as landuse=brownfield, of highway=construction so it’s just a case of adding in the street names once the developer has bothered to actually put the name plates up!
One such diff intrigued me - close to where my old University digs used to be (Ethel Williams Halls were demolished years ago!), a very small stub off a residential close started changing name.
Penfold Close became Whitby Crescent which ITO dutifully reported on. I though this to be a plausible omission, so added the name change and gave the credit to OS using the source:name=OS_OpenData_Locator tag. As this is about 12 miles away, and the road section in question about 20m long, I didn’t undertake a physical survey - slapped wrists all round!
Recently, it changed again to Whitbay Crescent, which seemed strange - had a mapper in OS Towers lost a key from their keyboard? Well, as this seemed so strange and my Winter fitness improving, I cycled out from Cramlington to find out.
Off a small residential road, a t-shaped stub is surrounded by 5 blocks of semi-detached homes. Only one has a name plate showing Penfold Close, but the door numbers show odd and even consistently showing they are in the same street.
So, what is going on with OS Street View changing names of a cul-de-sac all of 20m long? Well, I suspect it could be a Copyright Easter Egg!
The suspicious mind in me wonders if OS has been making small but insignificant changes in OSSV open data to track if and how fast they appear in OSM and other databases? I don’t believe this raises any copyright issues (I added the source tag to credit their information, as requested by the licence interpretation), but the feeling of possibly being tracked is both creepy, and reassuring.
Creepy - no one likes being instrumented and put in an experiment.
Reassuring - If my paranoia is correct, could it be that OS take OSM seriously to the point that our Open Street Maps are being watched keenly and closely by intelligences greater than our own*?
With apologies to HG Wells!
Discussion
Comment from EdLoach on 16 March 2015 at 08:39
If it is a Copyright Easter Egg it is more likely to catch users that aren’t attributing rather than those such as OpenStreetMap that are. It does suggest that all ways with a source:name=OS* tag should really be surveyed at some point though.
Comment from Warin61 on 16 March 2015 at 22:46
That would indicate that OS data is not too trustworthy! Well spotted.
I’d then only take OS data if OSM did not have data on the object. And naturally attribute the source.
If there is a discrepancy between OS and OSM then flag OSM data for checking .. but don’t change it untill OS data is confirmed. If the OSM data is correct then add a note or use the tag check_date=2011-11-12 (example only of 12 Nov 2011).
Comment from SomeoneElse on 17 March 2015 at 13:20
It’s not the first and it won’t be the last “comedy misspelling” (1) in the OS’s data (which, let’s not forget, came originally from some rather lowly local council employee). Rather than a deliberate “Copyright Easter Egg” I suspect they’ve just made a mistake. As Bernard Ingham said in another context, it’s much more likely to be a cock-up than a conspiracy.
(1) https://twitter.com/sk53onosm/status/323712730248933376
Comment from Pieren on 18 March 2015 at 14:57
Easter Egg and OpenData are not compatible.
Comment from chillly on 18 March 2015 at 17:56
What is really odd about the variable quality of the OS OpenData is that the high-quality, paid-for OS datasets (that we can’t use in OSM) don’t have any of these mistakes. You would expect that the OpenData is derived from the quality database, so how do these mistakes emerge? Then there’s the occasional variation between the OS Locator and OS StreetView names. There must be a manual process to introduce these mistakes, or a manual intervention to deliberately introduce them.
@pieren since the OS licence requires attribution (which OSM satisfies), easter eggs could be useful for OS to find unattributed use.