From previous discussion here I realize that there is a lot of missing data from the roads in my local area that should be addressed. To help fill in the (very large) missing pieces I want to utilize Map Roulette for both my own organization and hopefully to bring others into it as well. I very much like the idea of gamifying crowdsourced efforts.
The goal is to create very small tasks that will build on each other, while needing no local knowledge or research further than what is available in ID. So far I have the concept of
Determine road surfaces
Determine if a paved road has line markings.
Determine the number of lanes for roads with line markings.
Each of these steps will be broken down into a MapRoulette challenge, building on the results from the previous challenges. Each will be limited to single counties to help make reasonable goals.
I’d like feedback before putting anything into the public. A preview of the first challenge is unlisted here
I appreciate your approach of having participants tag tasks with specific reasons for Can’t Complete. Having managed MapRoulette challenges in the past, I’ve found the reasons mappers provide to be very useful signals for subsequent cleanup tasks.
We should have more “how to recognize” tips like this on the wiki too. The GeoGuessr player community has put together a wealth of resources like that, but it’s very Google Maps–centric, whereas the imagery we work with can have different characteristics sometimes.
I noticed that service=driveway is considered out of scope for the project. In my experience, many driveways are misclassified as major service roads (lacking a service=* tag), largely due to the TIGER import and Amazon Logistics’ earliest mapping efforts. Would any of your tagging suggestions or identification tips be impacted by a service road misclassified in this manner?
Yes I have seen the same pattern in this area. I don’t think labeling driveway surfaces would be much use to anyone, but I can not think of any negative impacts for doing it because of a missing service tag.
Ill have to put some additional thought into if I want to add to the task, make a different challenge or something else to solve that particular issue.
I think this is a great idea and wish you best of luck, especially if you can crack the nut on how to get many mappers helping on this. I recently ran I similar challenge, as part of a larger project to just add surfaces on the rural roads of Minnesota. If you don’t mind I’ll share my experience. First, I’ve been working on this by myself as my main OSM work for about 3-4 months. I thought I might get some engagement from others by starting a project on the wiki and announcing it on the OSMUS slack channel. Then I added a maproulette challenge as another, easier point of entry. Over the month that I had the MR challenge up (it had 1000 tasks), I got 10 tasks completed by others, while I completed 12 counties (each with 200-1500 tasks).
What do I think I’ve learned from this that you might learn from? First, I really like your tight framing of what to do, i.e. just change these tags or give a reason you couldn’t. I asked people to change the surface and potentially change highway tags (to driveways) and to fix awful geometry, so I think your challenge is clearer. Second, I think the MR interface to iD and JOSM is just too clunky for people to make 100s of fixes, so, unless you can recruit people, it will be very slow going. If there was a way to MR tasks where you could select from a menu in the MR interface (like you can with cooperative challenges) I think it would be much easier for a casual mapper to plow through a bunch of these really simple tasks.
What would I do in the future? I think something like this would be better done with the Tasking Manager where a mapper would change all the roads in a smaller (say square mile) area. I think you could map more efficiently if you can fix 10s of roads before leaving the editor and progress would be much easier to see if you’re clicking off sub areas, as opposed to individual roads.
I don’t want this to sound critical, because I’ve definitely not figured out how to do this very well, and I’d be happy to share other experiences with you if you have any questions.
A couple more thoughts. I think triaging by highway type could help a lot:
secondary and primary are probably all paved, so separate that out
tertiary is likely majority paved and there is strong connectedness among them-- a paved tertiary way is highly likely to be attached to another paved tertiary way and vice versa for unpaved
Unclassified and residential are probably pretty similar in their surfaces.
One special class in MN was residential w/o names. 70% of those were just driveways that had been labeled as residentials in Tiger. In your case, you may want to skip those, or save them for later.
If you do decide to do it on your own and use JOSM, the “todo” plugin in JOSM is super helpful. Also, the overpass-turbo query that I have on the wiki page will exclude ways in admin_level=8 (mostly towns/cities) so that you don’t avoid working on all the in town highways.
I’m ending up using Task Manager near me and focusing on small areas at a time. I created a Jython script for JOSM that helps sort out some of the name_1, name_2, and name swap craziness. NOTE: this only applies to a specific county as it is written, but the pattern is likely to be useful for other counties. I specifically have it work with only a single highlighted road, since I have to zoom in to validate geometry anyway. It defaults to an asphalt surface.
If I spot a gravel or concrete road, I set surface before running the script. JOSM Script to modify tags · GitHub
Based on the feedback I have gotten so far I have decided to do two things:
Exclude towns. These would be better done with the Tasking Manager and done by area instead of by way. I filtered them by categorizing anything not a “Township” as excluded. This has worked fairly well, even if it includes some higher density areas that run up next to town borders.
Exclude service roads. I spent some time doing this task by hand and found about half of the roads needing tags were service roads. Most of these roads only serve a single destination and won’t be included in the next steps anyway. Perhaps this can be revisited at a later date.
These two modifications changed my test challenge from 6379 tasks to 2002.
Thanks again for all the feedback, scripts and queries. They have all been helpful.
.