The painted markings are a different, lower meaning (they imply "___ gestattet"). See the Swiss tagging guidelines for the correspondence between tags and signs: osm.wiki/Switzerland/Map_Features#Cycle_and_Foot_Ways

These particular sand-coloured FGSOs really were intended by the city as bike "infrastructure", specifically where they couldn't fit real bike lanes because the street was too narrow and it was a bus route so they couldn't make it one-way. See e.g. the article here: https://www.provelozuerich.ch/wp-content/uploads/2016/02/vj_regio_16_5_ZH.pdf (page 3).

That said, I see the point that this intent isn't very verifiable as FGSOs are used so randomly by communes.

According to the wiki, (cycleway=shared%20lane?uselang=en) cycleway=shared_lane is exactly for this type of legally-not-binding paint-on-road "infrastructure" that just reminds car divers bikes might exist. In the US it's used for "sharrows" which is where they just paint some bicycles on the right hand side of the road without any demarcation. Since the yellow stripes on these two roads _are_ supposed to be for bicycles AFAIK, but are clearly not bike lanes (neither legally not practically since they are too narrow), I thought this tagging was appropriate - it's a marking telling cyclists to share the main car lane.

> every Swiss street is allowed for bicycles by default anyway

Yes but not every Swiss street has a Tiefbauamt-painted indicator. This tag is for the indicator, not the access (that would be bicycle=yes, which nobody tags on roads because as you said it's the default).

> Tagging it as such would make it look "cycle friendly", which it's definitely not.

It exists on the ground, so there must be some way of tagging it. We shouldn't not tag things that exist in reality because we don't like those things...

I don't feel that strongly about, though, it if you think it should be reverted.

I recently tagged the Stadt Zürich Velovorzugsrouten (e.g. relation/18164970) for similar reasons :)

I think the state tag on OSM has a slightly different meaning to what the canton's dataset might say: something in the default state on OSM is supposed to be findable on the ground. So for a route, that means you should be able to find _some_ route wayfinding signs/paint/etc. But at least in some of these cases, there really is no bicycle wayfinding at all, as far as I can recall from my last ride through.

Although one shouldn't map for the renderer, for what it's worth both CyclOSM and OpenCycleMap render state=proposed cycle routes, just with dashed rendering. Which IMO is appropriate: a user might be rudely surprised if they see a route on the map, assume they can follow it, and then find it not signed in real life.

There is one annoying thing, which is routes can only have one state value for the whole route. You can't mark just the sections that are not signed, for example. For the VVR I decided to err on the side of caution and tag them as state=proposed if they had significant portions not signed. The lcn=yes tagging on ways is in this respect more flexible.

Some of these bike routes are not signed at all (e.g. relation/18248732), and therefore probably don't meet the OSM criteria for verifiability. In particular the reference numbers should be at most unsigned_ref=*, not ref=*, since they aren't signed anywhere. Also, perhaps the routes themselves should have state=proposed until/unless they get signage.

> Looking at the other signage along Bahnhofstrasse, I think one might be supposed to go on the sidewalk, but not on the street?

No, there would need to be a blue pedestrian sign with a white "Velos gestattet" subsign on the footpath for this to be the case. Otherwise the law is that all footpaths within 2m of a road are strictly for pedestrians, and bikes should ride on the road. For paths further away from the road or not parallel to a road, they are usually signed as roads with motor traffic banned, in which case bikes are allowed; or as roads with all traffic banned, in which case they aren't.

Your panoramax photo is consistent with that, since it uses "no entry" instead of "no vehicles", which is only used in Switzerland for one-way roads.

I believe the allowance is only for northbound traffic, which is why you didnt see any signs coming from the north. But as the road is tagged oneway=yes here, the prior tagging should have been correct and already implied no access from the north.

This edit (and the one I made immediately before it) are based on changed published on the city website, where they announced the loosening of the restrictions on this section of Bahnhofstrasse: https://www.stadt-zuerich.ch/site/velo/de/index/erfolgreich-umgesetzt.html (and search for "Bahnhofstrasse (2. Massnahme)"). There is a chance I implemented the described change incorrectly, though. A picture of the sign is on the webpage.

Hi, this edit was just wrong. The Langstrasse tunnel has bike lanes now. Please don't answer questions in StreetComplete without visiting in person and checking the current state.

Having skimmed the thread you started, I agree that the specific tagging of restriction:bicycle=give_way is pretty bad, and would welcome something better. But I agree with the person who said this: "Main problem here, is that while I dislike relations it may be tricky to guess to which road this red turn right applies, maybe relation may be needed
for a full clarity." They are right that in general it's often unclear which way(s) are the ones for which the traffic light can be ignored, especially since in fully-tagged intersections, the traffic_signals node comes before the intersection of the ways.