facebook
Remove the "fix all" QA feature, as this effectively performs an undiscussed automated edit
Recent low quality OSM edits of many thousands of objects in the UK were made by a user blindly accepting all Rapid's QA suggestions, including some which were obviously wrong. These edits have now been reverted by DWG, but the functionality to perform what is clearly an undiscussed automated/mechanical edit should not exist in the Rapid editor at all.
https://wiki.openstreetmap.org/wiki/Automated_Edits_code_of_conduct
See this comment and the rest of the tread on OSM Community: https://community.openstreetmap.org/t/proposed-automated-edit-removal-of-crossing-markings-yes-tags-introduced-in-undiscussed-automated-edits/129614/48
But JOSM has similar autofix options that can be misused. I doubt you would make a suggestion like this to the JOSM team.
I think you should just ban the user and revert their work if it turns out they are using the fix all tool irresponsibly. I find that feature helpful if I have my screen centered on a reasonable view and I can see what I’m doing.
I also think nearly all of the suggestions made by the validator are correct. But if you have ideas on how to reduce false positives , we could definitely discuss those too.
But JOSM has similar autofix options that can be misused. I doubt you would make a suggestion like this to the JOSM team.
I certainly would if I were aware of any such options. When JOSM's validator suggests a fix, it still requires manual intervention, there isn't a default fix you can simply accept. I think you're conflating the ability of a JOSM user to perform a search and replace operation (which is subject to the code of conduct) with the ability of a Rapid user to accept suggested "upgrades" without reviewing them.
It doesn't matter that most suggestions are correct if an armchair mapper in New Zealand is, for example, automatically changing a Chinese restaurant in London called "Young's" to a pub run by the former brewery "Young's".
When JOSM's validator suggests a fix, it still requires manual intervention, there isn't a default fix you can simply accept.
for some reports there is magic "fix" button that can be applied without individual review - but it is only for safest ones, not all
When Rapid was first introduced, with it's AI/machine-learning features, we were told that since Rapid requires mappers to manually approve these AI suggestions, it doesn't count as a data import. If they've changed the software to do this, then it's become a different thing.
If I’m not mistaken, the validator is distinct from the MapWithAI (now Rapid Assist) feature, right? I haven’t seen an option to import an entire dataset with one button; that would kind of defeat the purpose of using Rapid…
