Local weighting for local mappers?

Posted by scruss on 3 November 2011 in English (English)

Having seen some of my carefully GPS walked/biked local trails being modified by distant mappers relying on out-of-date Yahoo/Bing mapping, I wonder if we should consider weighting edits based on how far they are from a mapper's home base?

How would you feel if your edits were ranked by the inverse square of the distance from your base?

Comment from 42429 on 3 November 2011 at 15:01

Sorry, but GPS tracks are usually about 10 metres off, especially when walking, depending on weather conditions. GPS tracks are only more accurate if they trace a recently built street which does not appear on Yahoo/Bing images. That's why I don't rely on GPS tracks where accurate aerial images are available.

Where did you experience edits from non-local mappers?

Yours, FK270673

Comment from scruss on 3 November 2011 at 15:13

But current aerial maps aren't available. They're 3-5 years out of date around here. With a WAAS-enabled handheld with a good fix, I'm +/- 3m.

Comment from Richard on 3 November 2011 at 15:45

Interesting thought.

I don't think the distance thing would work. It's possible to know lots about an area where you don't live. For example, I pretty much live in two places at the moment, but my home location can only be one of them. Similarly, I've been doing a bunch of mapping work in a place I often go on holiday.

But we do need to think of some way of restraining the global-change monkeys (to which Dasher's post, next to yours, also refers).

Comment from Vincent de Phily on 3 November 2011 at 15:47

Without going into the "tracks VS imagery precision" debate, I don't see how you would implement this technically:
* You can't just "own" a way an disalow modifications by another mapper... And weight-averaging positions of multiple edits wont work either (what about added/deleted nodes ?). Remember there is no "commit approval" process.
* What if the local mapper _is_ wrong ? He used a smartphone GPS in a narrow lane with high buildings ?
* What is a mapper's "base" anyway ? I live in Ireland 80% of the time, and in France the other 20%. I map at work, at home, and on holidays.

Concerning your original problem, I think the best you can do is to put details in the "source" tag. If you've gps-traced a road with a good device multiple times over the course of a few days, say so in the source tag. Bing mappers sould set "source=Bing" anyway, so they should see the previous value and think twice before editing.

Another thing you can do is to upload your traces. A lot of mappers will download traces too when they download the OSM data. It's especially usefull when there are many traces for a road, because it reduces errors. Good armchair mappers are carefull to calibrate the imagery offset using these tracks (or any other hints) before editing.

Comment from scruss on 3 November 2011 at 16:57

Did I say I had all the answers to this?

Comment from Zulu on 3 November 2011 at 19:14

No you didn't, but your proposal simply won't work.
I understand what you feel, but I can only share my own experience -- when I was in similar situation, I wrote a message to the mapper reverted the map to the Bing-shot state explaining that the area was rebuilt since than and my knowledge is more accurate.

Comment from AndrewBuck on 3 November 2011 at 20:26

Adding source=survey is also a common thing to do with ground verified stuff. That should also help keep people from changing it if the ground is different than in the photo.

Ultimately the best source of information depends on a lot more on whether it is just imagery vs GPS. Ground truth is the only source of much information and good GPS traces give you positions well, but detailed road and building shapes are gotten better from properly aligned imagery (if available).


Login to leave a comment