Google sued over collapsed bridge death


The Openstreetmap website still shows the bridge intact at lower zoom levels,
map (2)

One might say well people should zoom in always to make sure bridges are really there.
But one might be operating offline with just the zoom level available already on the data one is traveling with.
Or one might not be in the habit of zooming in to be sure about everything.
Yes, routing programs would detect the problem, but not people’s eyes.

Looking at Google Maps, Google has now taken the extraordinary step of ripping the entire road off their map, including blue StreetView traces. (But one nonetheless can click on supposedly empty areas, revealing two different years of StreetView.)

The extraordinary move was perhaps to prevent the same phenomenon happing in their renderer.

Anyway, how to solve OpenStreetMap’s Website from showing the bridge still intact at lower zoom levels?

  1. Rip out the road entirely (editing for the renderer… making the gap wider), or
  2. Fix the rendering, yes, even though carto doesn’t totally represent OSM, but that’s what’s on the Website. Or,
  3. Click on some other layer, than the Default. Alas, all the other six working out of six layers had the same problem. And Default is what the general public will be seeing by default…

Never mind human life. What about getting sued one day when something bad happens that could be prevented? Certainly Google has more No Warranty statements than OSM, and they still got sued anyway.

IF YOU ACT NOW and remove the bridge that isn’t, put noexit tags both ends of the roads going into the deep, the lower zoom 2KM, 5KM and down will see this removed tonight, Friday about 22:30 UTC. A Ctrl+F5 will be required to clean and refresh local caches. That is for Carto Standard !

OSM editors removed the bridge months ago. The problem is with Carto, joining gaps that aren’t “wide enough”.


I don’t see how a map can be responsible for bridge collapse death?? It make no sense to me.


You’ll need to hop into your time machine first, because it was edited in OSM some months ago.

1 Like

Er what? That makes no sense to me at all. Can you give an example of a router (and it is following a router that caused the problem here) that will route over a gap, such as this one?

It doesn’t make sense to me either. What happened to looking where you are going?


also, the maintainer of the road infrastructure is responsible to mark the road as closed.


Lol, At lower zoom you can walk across the Calais Channel and unfortunately I misplaced my crystal time machine ball to look if all is fine back then.

My suggestion is to map the bridge with a lifecycle prefix.

As @jidanni said:

Looking at it in iD, the gap could made a bit wider:

Plus noexit=yes at the end nodes as @SekeRob said,

currently tagged with barrier=yes

problem was with a router, but arguably a gap in a road should be well visible in the rendering, those line-ends should ideally be butt (hopefully this word will not result in an admonishment)

Are there actually barriers? On one side I can’t see one; at the other side it looks like a tree blocks the gap.

I can’t see barriers on the imagery too … barrier=ditch wouldn’t be completely false, but across the street not along the street.

If there is no actual barrier, I’m thinking of noexit and maybe a hazard tag.
Wouldn’t be bad if the owners of the house moved some scrap, sand or heavy stuff on the road. Or some crime scene tape.

Given recent events, I wouldn’t assume that what you use on historical images reflects what is there currently.

Press reports elsewhere have suggested that barriers were erected but removed by vandals. No doubt more details will emerge in due course.

1 Like

not doable with the current tech stack (it would result in gaps also within continuous road)

Overall such generalization is really tricky to get working reliably.

Let’s take Bing,

As you zoom out does the bridge magically join together? No.
It does on the OSM website, and it’s not due to stale map tiles.
(OK it did, until somebody today widened the gap. But it is just going to happen somewhere else. Like:

(Worse in Bing.) How are we to keep people from looking at the map and thinking they can use it? (I am talking about eyeballs. Not AI or routing programs.

Gates, barriers disappear at various zoom levels. One needs to remove a long enough segment to ensure the renderer won’t “heal” the gap.)

Just like things that are joined in reality (or in the editor) should not split apart as one zooms in in the renderer, things that are not joined in the editor should not become joined upon zooming out in the renderer.

Or, if the algorithm cannot be corrected, then such critical areas should have tiles that say “This tile temporarily not available, due to public safety concerns. Sincerely, OSM Foundation.”

What about rivers that are dangerous to cross, maybe on foot?
They vanish at certain zoom levels, should they always be rendered?

Yes, you can see type 3 barricades and “Road Closed” (R11-2) signs on either end of the gap in 2014 Bing Streetside imagery. Barricades such as these can be very easy to remove.

After the accident, some concrete jersey barriers were installed.

Regardless of whether navigation applications are at fault in a situation like this, we all have a positive role to play in promoting traffic safety. In the U.S., we’ve been playing whack-a-mole for years with washed-out bridges across creeks and ravines, fallen trees, and bypassed fords that come in through TIGER and persist a long time until a local notices. We’ve also been mapping other poorly marked hazards like unsignalized level crossings, sometimes in response to tragic accidents that we hear about in the news.

Of course people should look where they’re going and the roads should be safer, but I think maintaining vigilance over our map data is more constructive than blaming the victim or relying on the authorities to build everything correctly to spec ahead of time.

1 Like