I do see one person trying to help make a great map in this thread: Mateusz fixed one of the example problem nodes I linked.
I commented in Changeset: 160609138 | OpenStreetMap - and if they will make further edits before replying I will write to DWG so account will start getting blocks (feel free to do so on your own, if you spot that problem continues - I may be busy with other things).
(report to DWG can be as short as: X ignores changeset comment at {LINK}, they edited after getting it {LINK TO LATER EDIT})
Initial blocks are typically 0-hour ones making mandatory to read warning before editing further.
If someone knows Spanish - can you ask in Spanish there? I wrote in English.
(yes, it does not solve entire systematic problem but at least it will stop damage caused by this specific person or will make them aware that what they do is unhelpful)
can you link challenge? Or have you flagged it at MR already?
Current reaction time to reports is long enough that stopping damage by blocking flagged ones would be highly welcome and would encourage people to bother reporting challenges causing harm.
Currently flagging appears to be mostly placebo.
Reported Challenge #17807 - Invalid turn restriction · Issue #7 · maproulette/challenge-reports · GitHub example listed in my 2023 report was processed in February 2024 after being reported in February 2023.
5 Likes
It seems quite possible that they genuinely believe they are improving OSM by adding POI data. The challenge description tells mappers to add any POIs that are missing. The mapper may well believe the proposed data is from a reliable source. If as you say many of the POIs don’t exist, that assumption may be wrong.
I think a common theme is that many mappers don’t really grasp that there is no “authority” behind OSM. If ID tells them to “upgrade” tags they think that somebody has definitively approved a change in tagging and don’t realise it may be the result of a random interaction of name suggestion index presets that nobody foresaw. If they are presented with POIs to be added they think someone “in power” has already determined that the data is valid. I honestly dont know what to do about this…
4 Likes
be more clear that some heuristic guessed by QA tool may be wrong (so instead of “you should make this change” say “automated tool guessed that this change may be a good idea” and similar)
and MR descriptions should be clear that unclear, confusing or dubious tasks should be skipped
similarly, there should be a clear way to escape/hide/drop cases where automated tool found something unexpected
5 Likes
A post was merged into an existing topic: Discussion about the tagging of intersections of (minor) waterways and highways
I’ve moved the specific posts about the discussion about the intersections of waterways and highways into another topic. Feel free to reach out to me if I’ve missed a post or selected too many!
One other issue that is worth mentioning - notifications via MapRoulette are basically ineffective. I commented on https://maproulette.org/browse/challenges/45132?tab=conversation back in August 2024 via the OSM account https://www.openstreetmap.org/user/SomeoneElse2 which has a valid, working email address. The corporate creator of that “challenge” replied immediately, but I was not notified about their reply at all. This is because Maproulette maintains its own private messaging system rather than just sending a message to the logged-in OSM user, which I would have immediately received by email, and would also have been visible as an unread OSM message in all the apps that check that.
Edit 16/2/2025: Maproulette now (but not when I sent the message in 2024) now allows “contact via OSM”, so that option is now available.
3 Likes
I have reverted this and similar changesets in the same area by the same author in https://www.openstreetmap.org/changeset/162536114.
In order to identify them I have to use a ChangesetMD database to identify the scale of the problem and an overpass query to search; there was no way to inspect the damage from within MapRoulette itself. There was no way to click through from the challenge to the users who have made changes, the areas in which data was changed or the changes that were actually made.
This does not fix all problems created in OSM by this challenge; it does not even fix all errors by this contributor doing this challenge (this node has lost information), but it does at least remove some “fake towns” from Ireland.
2 Likes
More attempts to fix things: I opened
that attempts to make things more clear in iD. Feedback welcome there, especially about new term. I am not a native speaker.
I crafted overpass turbo that identifies fords last edited by this specific mapper and fixed some of them (feel free to help with fixing more, I will try to look at more but no idea when I will manage)
1 Like
maybe there shouldn’t be challenges for waterways crossing with highways, because it is common that it cannot remotely be determined whether there is a bridge or a culvert or a ford, especially for ditches and streams, and the challenges seem to attract guesswork
4 Likes
In the particular case I am complaining about in Niagara Region, the Maproulette task was (as far as I can tell) to add “unmapped roads” (driveways, parking aisles, tracks, etc), and the culvert/ford mistake was instead encouraged by iD (hence Mateusz’s GitHub issue above)
I don’t know what the Maproulette task was for “improve water geometry” that introduced at least one incorrect ford in Connecticut (Changeset: 90431281 | OpenStreetMap)
More generally I stand by my statement that Maproulette tasks that drop mappers into places unfamiliar to them and ask them to make changes based on Maproulette and iD recommendations and perhaps aerial imagery amount to undiscussed organized mass editing (as complained in the very first message in this thread)
editing to add: and those challenges shouldn’t be done/allowed
editing to add: There have also been Maproulette tasks that consisted entirely of “apply iD presets to shops” (for example the one linked in the first message in the thread) which also IMO amounts to organized editing and shouldn’t be allowed without prescribed organized editing steps
2 Likes
I suspect that it’s the mappers themselves rather than Maproulette that choose to “fix things while they are there” and (as you pointed out above) if a mapper accepts iD’s brand recommendations somewhere that they are unfamiliar with they’re basically just guessing.
In the last 24 hours there have been ~1500 MapRoulette changesets
changesets=> select count(*) FROM osm_changeset WHERE created_at > '2025-02-15 12:46:00' and tags -> 'comment' LIKE '%maproulette%';
count
-------
1492
(1 row)
A quick look at some of those (first locally, and then internationally) finds a number that prompt more questions (for example, one fairly local to me). @mvexel who is expected to do the QA for these ~1500 changes per day and how is it expected that they will find these changes local to them? With a ChangesetMD database I’ve been doing this to look at local ones:
changesets=> SELECT id, user_name, tags -> 'comment' FROM osm_changeset c, (SELECT ST_SetSRID(ST_MakeEnvelope(-11.0,48.0,2.5,61.1),4326) AS geom) s WHERE ST_CoveredBy(c.geom, s.geom) and created_at > '2025-02-15 12:52:00' and tags -> 'comment' LIKE '%maproulette%';
With OsmCha you might be able to use something like this (after adjusting the area of course) but I’ve always found OsmCha a bit hit and miss. Is there some QA built-in to MapRoulette?
1 Like
Are all other editors and their developers similarly expected to be responsible for those that misuse them?
Would you ask, “Who is expected to do the QA for the tens of thousands of changesets made daily by iD users?”
Wouldn’t the answer be the same for any edit - history tab on osm.org, OSMCha filter, etc?
11 Likes
iD doesn’t host challenges and give mappers points for solving tasks
Sure it does. The yellow “Issues” it prompts for fixing add changeset tags which count how many a given user has resolved or ignored.
1 Like
Please stop the whataboutism. It is insincere, and not suitable to advance the discussion.
3 Likes
In terms of catching inadvertent edits by unsuspecting users, yes (those and a whole lot more, some of which I’ve mentioned above). However, I think you underestimate the efforts that editor authors go to to try and make sure that people using their editors make correct edits. See for example here (for iD) and the numerous discussions around StreetComplete (including in the language channel on OSM US’ slack to try and make sure that nuances of English aren’t missed).
The situation with MapRoulette is slightly different because mappers (incliuding mappers new to OSM) are being literally being told what to do - we need to make sure that what they’re being told to do makes sense.
Edit: And to provide an example of how challenges can be broken up to make them more specific, see my earlier post about a MapRoulette challenge last year that was causing users to make incorrect edits.
5 Likes
Oh well if you put it like that then I definitely do believe there should be also some accountability and oversight of what iD suggests to “fix” and how. Otherwise what’s to stop incorrect changes being put into the map?
Do you believe that editor maintainers shouldn’t be responsible for their tools’ “suggestions”? How would that work? If I think that a hairdresser isn’t a shop because you don’t buy anything there and so my editor will tell its users to upgrade shop=hairdresser to amenity=hairdresser, am I not responsible because it’s the users who apply the suggestion? What if I host a challenge inviting others to make this change throughout Canada?
1 Like
In software, I don’t think we tend to place blame for how you use a tool on the tool makers themselves. Major OS’s have spell check, that doesn’t mean they’re responsible if you don’t validate its output. For a more recent example, AI is notoriously wrong but we are all coming to understand it’s on the user to verify information is right even if the tool is saying otherwise. How is MR or id any different when suggesting edits? Trust, but verify.
2 Likes
The thing is, MapRoulette doesn’t make the suggestions. The guys who create the challenges do that. I think its unfair to attribute this to MapRoulette itself without thinking of at least something how to prevent this. MapRoulette is just a tool.
So, one way i can think of is: There could be a mandatory field with a link to a forum-thread that just accepts a properly formatted link. If there is no discussion, its not allowed. Its a least something to force challenge-creators to think about what they are doing.
5 Likes