Proposed automated edit - removal of crossing:markings=yes tags introduced in undiscussed automated edits

In April 2025, @Numbergod (an armchair mapper apparently based in New Zealand) made a number of essentially automated/mechanical edits in the UK, affecting hundreds of objects per changeset. The changeset comments were generally the unhelpful and vague “Fixed various issues.”, but the character of each changeset was essentially to accept every tag “upgrade” suggestion made by the Rapid editor without question, including examples where it was obvious that the suggestion was wrong. Apart from one changeset immediately following Numbergod blocked by trigpoint | OpenStreetMap , they have failed to respond to changeset comments.

I would be quite happy to see every recent change which they have made in the UK reverted, because checking ~100k potentially tainted objects to validate a misguided box-ticking exercise masquerading as QA is a wildly disproportionate level of effort.

In the mean time, I propose to remove every instance of crossing:markings=yes added by this user in the UK as the result of automated edits which added nothing useful whatsoever to the OSM database.

8 Likes

I agree with your solution to revert all their edits.

4 Likes

would it complicate or make harder to revert other edits they made?

2 Likes

Another one of their changes was of a Chinese restaurant to a pub purely based on the name.

It’s pretty clear that “Mr Brain was not at home” at the time they were clicking at least some of the buttons. :slight_smile:

2 Likes

Possibly. Apart from there being protocol to follow regarding automated edits, that’s part of the reason I’ve initiated this discussion.

1 Like

Also, the rate at which they must have been clicking to accept the proposed tag “upgrades” leads me to wonder whether you can get a water-cooled mouse to avoid it overheating from friction :slight_smile:

2 Likes

I do have a monitor set so that I will be able to block them again if they edit before they make sensible comments on the changeset comments made.

I do believe there should be a complete revert of their UK edits as a minimum. Their edit rate shows that they did not put brain in gear before accepting rapid suggestions. The shear volume makes it next to impossible to find all errors such as Chinese restaurant to pub.

Rob’s suggestion that these types of edit should be rate limited is a good one, not just for new mappers. This mapper started in 2008 so really should know better.

5 Likes

@Numbergod also amended three Underground Stations with false data.

I’ve reverted them but happy for their changesets to be fully reverted.

I’m unaware of this particular editor, but should some of the blame (especially for new contributors) be levelled at software which offers supposed validation suggestions?
I feel any ‘does this need correcting’ or ‘recommend adding this tag’ swings the needle of suggestion firmly into the zone of implying a needs to be changed.

3 Likes

If not already noted anywhere above, did anyone notice that ID Editor is part and parcel of the crossing:marking=yes perpetuation? Well, just to verify, deleted all tags from a crossing that has the proper crossing:markings=zebra and selected ‘marked crossing’ from the available presets, and this is what ID inserted as tags:

There should not be, yet they do.

1 Like

If you delete a more specific tag and then choose a less specific preset, iD won’t restore the more specific tag. Similarly, if I delete cuisine=italian from a restaurant and then choose the Restaurant preset, it’ll restore the amenity=restaurant tag but not the cuisine=italian tag. The situation is a little different in Rapid because it has more specific presets, one for each crossing marking pattern, such as “Marked Crossing (Zebra)”.

1 Like

My point is, if you start fresh with a crossing and land on this preset in ID, large chance is the meaningless ‘yes’ will be injected into OSM by the unsuspecting mapper, and while selecting crossings wherever I map, this is the finding instead of zebra, dots, surface, etc. Don’t blame the guy who did the yes tagged who takes ID for gospel… They are not!

1 Like

The keyword yes on its own is indeed quite meaningless, but crossing:markings=yes means a crossing is marked. From this user’s changeset comments, mappers are raising doubt about whether each of these crossings actually is marked, whether the tag is being applied accurately. In English-speaking countries, crossing=uncontrolled is ambiguous because an “uncontrolled crossing” doesn’t necessarily mean it’s marked. (Some mappers routinely used crossing=uncontrolled for unmarked crossings by intuition.) So anyone bopping the suggested fix button repeatedly needs to exercise due care that they might not have to in a more homogeneous environment.

Anyways, all of that is hashed out in a different global thread. This proposal is more tightly scoped to a mapper whose edits are being questioned for other reasons besides crossings.

1 Like

I couldn’t persuade either Rapid or iD to suggest changing these underground stations to the (incorrect) Elizabeth Line which was added. Automated edits in the name of QA (Quality Assassination?) are bad enough, but adding their own misunderstandings is even worse. I’d love to see them trying to fit a mainline train into a Northern Line tunnel.

Maybe it got fixed since then? Sounds like it is coming from NSI.

I don’t think it is from NSI. No commit message in the last 2 weeks mentions Elizabeth Line, London Underground or TfL.

Another automated edit by this user set operator:type=private on Great Ormond Street Hospital, which was from NSI (fixed in Update hospital.json - NHS trusts are operator:type=public, not private by rskedgell · Pull Request #10909 · osmlab/name-suggestion-index · GitHub ). This would be obviously wrong to anyone in, or with adequate knowledge of, the UK.

1 Like

Either way it sounds like literally all edits they made should be reverted.

3 Likes

This user is now doing the same in New Zealand, per latest changesets: Changesets by Numbergod | OpenStreetMap

1 Like

I too would be happy to see these changesets reverted and the user blocked until they make contact.

Does this take an explicit request to the DWG?

2 Likes

I think they’re based in Christchurch, NZ - How did you contribute to OpenStreetMap?

For completeness, we know about this and have been discussing it internally. I’m not the DWG person dealing with it, but my impression is that a fuller revert than has currently happened will be required, but that will need a bit of careful planning.

From a personal point of view it’d be great to have thoughts on how we can catch situations like this where someone has been many good-faith changes that actually weren’t valid partly because the editor they were using was a bit “economical with the truth”. This mapper was clearly trying to improve OSM; it’s a real shame that they were misled into making the edits that they made.

5 Likes