What's the protocol for correcting ways that have been "corrected" using different data than the original?

A couple of months ago, someone modified a large number of ways based on GPS data. While it is likely that the relocated ways are more “accurately placed”, the older (unmodified) surrounding ways match Esri/Bing imagery. The result is that several ways which should be perfectly straight now have significant jogs in them - and most ways in the area no longer align with Esri/Bing imagery.

The changes that were made encompass over 50 ways and over 1000 nodes. I’m not sure what the area looked like before these changes were made.

Part of me wants to just go in and re-align everything to the map imagery. However, that seems somewhat counterproductive. Also, I tend to not be very tactful. My initial reaction is to get upset and say something like, “You really screwed up this area with your so-called ‘fixes’”. In similar situations involving only a couple of ways and a small number of nodes, I’ve just made the changes that I thought were necessary. I could do that is this case too, but is that the best way forward?

I guess what I’m asking isn’t how to deal with this particular situation, but how to handle such situations in general.

Just try to contact the person via the changeset and for sure start “nice”.

A good question would be to ask what is the accuracy of the GPS data he used and get that also for the Esri/Bing imagery and share that. Do not yet propose solutions, it is better to ask the other side on how to get it solved but that can also be in a follow up phase.


You can’t always trust the images from Bing, Esri etc.!! Sometimes such images can be several years old! So, before saying the GPS data is wrong, you must confirm the images you see in your editor are updated.

1 Like

This appears to be an appropriate method to begin the process of resolving the issue. I’ll probably have to spend a couple of days trying to compose an introductory message that isn’t accusatory.

I didn’t mean to imply that the GPS data might be wrong - only that the use of GPS data for a portion of various ways has introduced significant jogs in ways that, in reality, are perfectly straight.

You raise a very interesting point. Let me throw a hypothetical out there based on that point. Let’s say a very long, straight street has been mapped into OSM using Esri. Now someone comes along who isn’t concerned about the entire street but only with a small length (perhaps 3 blocks). They meticulously gather GPS data for that short length of street and modify OSM based on the data they have collected.

As a result of “improving” the OSM data, the street is no longer perfectly straight but has an offset jog of 4-5 meters in the middle. While that 3 block range might be more accurately located in OSM, the question becomes: Which is more important - to have a small area that is accurately placed but creates a discontinuity with the larger surrounding area, or to have an overall area consistently mapped even if it isn’t entirely accurate because the map imagery was offset?

I don’t trust GPS/GPX data for a minute. There’s claim of the modern navigation/mobile phone devices have a 1-3 meter offset error, but what I see is easily 5-10 meters and more in hill/mountain and densely buildup areas it’s worse.

1 Like

Near tall buildings there is usually a systematic error, pushing the position determined by GPS away from the building. This could explain the “significant jogs”.

1 Like

That would suggest an improvement to OSM data, if they really are more accurately placed now.

I’m guessing that’s at the limit of the area that;s been fixed? Or is there something else going on (like alignment with GPS inaccuracies, as suggested elsewhere)?

Perhaps if you link to the area we can have a look at the GPS traces there and also the available imagery?

1 Like

Ideally the imagery should be shifted to match the GPS data and the entire street should be edited. Barring that, simply “feather” the jog.

Incidently, ESRI or Bing imagery of today may not align with ESRI or Bing imagery of next year, so this isn’t just an issue of GPS vs imagery.

1 Like