[Talk-us] Newbie-safe semi-automated edits? example: turning_circle

Mark Gray mark-osmus at hspf.com
Sat Aug 15 21:51:40 BST 2009


Imagine that a newly arriving websurfer finds OSM for the first
time. It sounds exciting, but there is really a lot to get through
before you can make a contribution. A surfer's attention span
might not be that long. Wouldn't it be nice if we had something
really easy they could do to help without them having to learn
anything about how to edit the map?

I recently discovered the highway=turning_circle tag and started
adding it to the map. I find that it is a bit tedious using
Potlatch if adding turning circles to an area is my main goal. 
It takes several clicks and a few keystrokes for each one, then
panning around to find the next one. Since this is such a common
feature, I started thinking of how adding these might be made
easier and realized that what I was thinking of could be helpful
for many other editing tasks.

I am imagining the following steps:

1. Run a script that finds all road dead end nodes in an area
  (replace this step with a script that finds anything that a
  person could make a quick multiple-choice decision about.)

2. Make a small picture from Yahoo satellite centered on each one

3. Show a person each image and ask: Does it look like a turning circle?

4. Collect the user's response with one keystroke and move quickly
   to the next picture.

5. Collate the responses and queue the edits to be made

For me, this would make some simple tasks much easier even though
I am comfortable with Potlatch and JOSM, but I think it is more
exciting thinking about making this work on a public web page.

There are several things to consider:

- We would certainly want more than one inexperienced websurfer to agree
  before accepting an edit.

- It might be good to let people sign in with their account, but
  it might be nice to let people participate (perhaps with reduced
  trust/additional verification) without even signing in.

- We would need an illustrated description of the task
  (Here is an example of a turning circle, here is a plain dead end,
   press T if the image below is a turning circle, D if it is a
   dead end, C if you can't tell, B to go back to the last one.
   Here is a link to the main map at this location...)

- It would be good to be able to load the next images while the
  user looks at one so the new image can be shown as quickly as
  possible. 

- Would javascript or Flash be good for this? 
  I have never used either one, but I am not afraid to try. 

This is only an idea so far. I have not written any code. Does
anyone have suggestions or some example code that does any part of
this? I need to figure out how to:

1. find (lat,lon) of all dead ends of roads, probably from a .osm file
   (I have not processed .osm files before, someone with more
   experience probably already knows how to do this?)
2. get a Yahoo satellite image centered on that point
   (I have done this but not in a browser-based language)
3. respond to a keypress (a button click could work, but it would be
   nice to respond to either one)

-- 
Mark Gray

P.S. I don't want to distract you all too much with another
project, but this idea reminds me of Distributed Proofreaders:
http://www.pgdp.net/
They have had a lot of success with making it easy for people to
help proofread texts for Project Gutenberg, one page at a time.




More information about the Talk-us mailing list