Proposal:Deprecate crossing=zebra in favor of crossing:markings

Deprecate crossing=zebra in favor of crossing:markings
Proposal status:	Approved (active)
Proposed by:	Bauer33333
Tagging:	crossing=uncontrolled + crossing:markings=zebra Data item
Statistics:
Draft started:	2024-05-30
RFC start:	2024-06-28
Vote start:	2025-05-09
Vote end:	2025-05-24

Proposal

This proposal aims to replace most cases of crossing=zebra with the more popular combination of crossing=uncontrolled + crossing:markings=zebra and as appropriate with crossing_ref=zebra.

Rationale

Tl;dr:

ID users added crossing=zebra to all kinds of marked crossings over a span of about five years. In a random sample area in the US this led to only 5% of all crossings tagged with crossing=zebra in 2022 to actually have zebra markings. There exist better tags now without this issue and that also don't have the issue of mixing crossing type and crossing markings in one tag. Just using the crossing tag fails even in its home country UK to represent the different zebra types.

Long version:

Crossings in OSM were originally described by their british animal names. Applying this UK-centric tagging scheme in other countries led to quite some struggle and not really fitting tag selections. Mappers were confused if eg. crossing=zebra describes the markings on the ground, the right of way situation, the lack of traffic lights or something else. A proposal approved in 2008 deprecated all the crossing animals including the zebra. Instead crossing=traffic_signals and crossing=uncontrolled was introduced to describe the presence of traffic lights with the crossing=* tag only. To handle the country specific configurations and implied additional features crossing_ref=* was introduced. The value would be the region specific name of such a crossing setup, such as zebra crossings in the UK with their implied features.

While crossing_ref=* started to get some traction, probably not as much as it could due to the lack of documentation of the implied features. The deprecation of the animal values in crossing=* did not and they continued to be used. After an edit war on the crossing=* page they were documented as UK only shorthands, initially with a warning about the deprecation which got lost over the years. The lack of deprecation and multiple systems in use at the same time lead to confused mappers, sometimes adding multiple values to a crossing, like traffic signals and zebra. Due to a lack of better tags crossing_ref=* was also used to describe marking types instead of crossing types.

Then in 2014 iD added a preset for crossing=zebra. This lead to a massive increase in the amount of crossing=zebra: https://taghistory.raifer.tech/#***/crossing/zebra The issue with that was that iD called the preset not "zebra crossing" but "marked crosswalk", based on the legal status and not the marking type since all UK marked crossings have zebra stripes. Outside of the UK it was used on effectively everything that would be tagged as crossing:markings=yes today. I've found crossing=zebras from that era even on signalized crossings or ones that were only marked by a row of cobblestones. While some countries where the majority of crossings were in fact zebras could keep that mass mistagging somewhat under control it was a lost case in areas with multiple marking types.

After this went on for five years ID tried to fix the situation by giving a deprecation warning in the editor that suggested to retag crossing=zebra with crossing=marked. This got a lot of criticism, on the one hand because crossing=marked had almost no users compared to crossing=uncontrolled and on the other hand because mappers considered the zebra information lost. But that info was pretty much lost already, with over 90% of the crossing=zebra worldwide added during the mistagging period completely ruining the original UK meaning of the tag. ID removed the undiscussed deprecation about a year later and only hid the preset that contains crossing=zebra instead.

In 2022 crossing:markings=* got approved and usage of it skyrocketed almost immediately, as seen in the graph on the right. The new tag allowed to clean up crossing_ref=* from the cases where it was used to describe the crossing markings only, leading to a massive global drop. It has a clear definition of being for crossing types now. The new tag gained massive traction, as of May 2025 the just about one and half a year old crossing:markings=zebra had already about twice the uses of the ten year old crossing=zebra with no visible trend of slowing down. Meanwhile crossing=zebra has seen almost no growth in recent years. In regions without a lot of crossing=zebra the amount seems to decrease, while it rises in areas with almost no crossing=zebra, what is an indicator for mostly new mappers picking that tag from the tag suggestions, since it seems to describe a true zebra crossing better at a first glance if you don't know that with the amount of skunking it means basically the same as crossing:markings=yes in some areas. With the low amount of crossing=zebra tags there have also been single editors doing noticeable shifts in the numbers. Compared to the continuing growth of other crossing tags the zebra one remains overall flat, especially in 2024.

Evolution of crossing:markings=zebra, crossing_ref=zebra and crossing=zebra.

crossing=marked + crossing:markings=zebra (bright blue) vs crossing=uncontrolled + crossing:markings=zebra (dark blue) vs crossing=zebra (green)

For example in Hamilton County, Ohio only 7 crossings have been retagged to crossing:markings=zebra since late 2022 while 129 got an value describing an other kind of marking.

So we end up with a tag which meaning differs depending on the area as well as the time it was added to an object.

Most data consumers can't handle that and internally replace everything that contains zebra as a value with zebra crossing, even if it was not intended to be one by the mapper back then or are just replacing it with crossing:markings=yes. Additionally it can also not distinguish between legal and informal zebras which exist eg. in the UK.

Therefore I'm advocating to deprecate crossing=zebra and use the appropriate value of crossing:markings=* to describe the crossing markings of crossings and crossing_ref=* to describe the legal type of crossings instead.

The various taggings currently in use.

Additionally this would reduce the amounts of different tagging schemes for one thing, there are currently three different keys with a zebra value and these can be combined with various multiple different crossing=* values as well as seen in the flowchart on the side.

Some countries have already taken this step without issues, eg. France or Russia.

Tagging

mapping new zebra crossings

Some people have voiced concerns about two tags needing more effort than using one tag. The same one click action can be achieved using presets. The presets would need to be named appropriately (eg. uncontrolled zebra crossing) to avoid confusion with light controlled zebra marked crossings. Some editors like Rapid already offer presets with marking styles

Countries with a legal zebra crossing

In regions with a legally defined zebra crossing type the editors would offer a zebra preset containing crossing=uncontrolled + crossing:markings=zebra + crossing_ref=zebra

Countries without a zebra crossing type

In regions without a legally defined zebra crossing type the editors would offer a zebra preset containing crossing=uncontrolled + crossing:markings=zebra

.

Dealing with already existing crossing=zebras

In the areas where crossing=zebra has been used to describe all kinds of marked crossings, no matter if they even feature anything remotely similar to a zebra crossing, automatically adding crossing:markings=zebra would cause more harm than good there. In areas where crossing=zebra has been used on true zebra crossings only retagging them automatically with crossing:markings=yes would lead to a loss of the zebra information. So a global automated edit is impossible.

Avoiding this issue would be just a deprecation warning without an automated replacement leading to a manual review. In case an automatic retagging is wanted instead there would be a need for a list of areas where crossing:markings=zebra is preferred and where crossing:markings=yes is a better way to go. From what I've seen generally areas that use American English suffer more from the misnamed preset than other language areas. For those areas a MapRoulette challenge or similar would be nice as well to figure out which crossing=zebras actually have zebra markings and which have other markings.

The cases where the crossing values conflict have often seen multiple values separated by a semicolon and would need to be manually reviewed as well.

Which crossing=* value to choose?

Instead of crossing=uncontrolled crossing=marked could be used instead, but the former has seen a lot more usage recently: https://taghistory.raifer.tech/?#***/crossing/uncontrolled&***/crossing/marked. Note that while in normal English crossing markings are usually a kind of crossing control, the OSM definition only sees crossings with traffic lights as controlled crossings.

Examples

Here is an overpass turbo query to find crossings currently tagged in the proposed way: https://overpass-turbo.eu/s/1MfQ

Rendering

✓ Straßenraumkarte Neukölln uses all three zebra tags and converts them to one internally, so they won't see any effect.

✓ OSM2World uses crossing:markings=*. crossing=zebra and crossing_ref=zebra serve only as a fallback.

✓ The OpenSidewalksScheme converts crossing=zebra to crossing:markings=yes, it already requires crossing:markings=zebra for it to see a crossing as having zebra markings.

✓ OSMAnd is listed as a tag user on taginfo, but I was unable to find any mention of crossing=zebra in their code, only lots of crossing:markings=zebra. Seeing them not listed at that tag their taginfo file is probably outdated and they deprecated crossing=zebra.

OSMCarto does not render crossings in any special way.

Current editor status

	adding crossing=zebra	adding crossing=uncontrolled/crossing=marked + crossing:markings=zebra	encountering a crossing=zebra
Rapid	The legacy crossing=* value is displayed in the "Crossing Type" field in the sidebar, and a reasonable default value is set by all of Rapid's crossing presets, but the user may change it. When editing any crossing tags, Rapid's validator will check that the legacy crossing=* value is consistent with the newer crossing:markings=* and crossing:signals=* tags.	There is a preset for crossing=marked + crossing:markings=zebra	Adding crossing:markings=zebra is recommended. crossing=zebra is left untouched.
iD (Go Map!!)	Search for the preset of a marked crossing. Replace crossing=uncontrolled with crossing=zebra. Due to the global amount of crossing=zebra it is listed as the fifth value option but not highlighted in blue as a recommended tag.	Search for the preset of a marked crossing. Replace crossing:markings=yes with crossing=zebra. ID offers a drop down with icons of the different marking styles.	An otherwise hidden preset is shown. No action recommended
StreetComplete	crossing=* is not added by StreetComplete.	crossing=* is not added by StreetComplete. Answering a quest about crossing markings will add crossing:markings=yes.	It will not ask a question about crossing markings and leave crossing=* as it is.
Vespucci	crossing=zebra needs to be typed manually, not part of a preset.	After selecting the crossing preset crossing=uncontrolled can be selected with a checkbox, crossing:markings=zebra from a dropdown.	"zebra" appears as a new unlocalised bullet point on the checklist, no further actions.
JOSM	Taginfo lists no use of crossing=zebra for any of the know preset projects, so I assume that the tag is already unsupported.

Features/Pages affected

Wiki pages affected:

Tag:crossing=zebra

Key:crossing

External discussions

Forum thread: https://community.openstreetmap.org/t/rfc-feature-proposal-deprecate-crossing-zebra-in-favor-of-crossing-markings/115293

UK community forum discussion: https://community.openstreetmap.org/t/zebra-crossing-deprecation-proposal/115300

Mailing list: https://lists.openstreetmap.org/pipermail/tagging/2024-June/067963.html

Dutch discussion about zebras in general, resulted in eg. crossing:markings=zebra being added on all zebra crossings in the Netherlands: https://community.openstreetmap.org/t/zebra-crossing-in-nederland-ndw/121871

Historic discussions:

Note that these discussions are older than crossing:markings=*.

2 meaning for crossing=zebra: https://www.mail-archive.com/tagging@openstreetmap.org/msg39997.html

automated edits seem to remove crossing=zebra drastically: https://www.mail-archive.com/tagging@openstreetmap.org/msg55098.html

See also

These are not related to what is proposed here but may be interesting for people who care about crossings:

https://wiki.openstreetmap.org/wiki/Key:crossing:signals

https://wiki.openstreetmap.org/wiki/Proposal:Highway_crossing_cleanup

https://wiki.openstreetmap.org/wiki/Proposal:Crossing_signalization

Comments

Please comment on the discussion page.

Voting

Voting closed

Voting on this proposal has been closed.

It was approved with 40 votes for, 12 votes against and 3 abstentions.

It has been emphasized that global mass edits should not happen under any circumstances and implementation of this proposal needs to be done carefully. Multiple comments also stated that there are general issues with crossing=* itself, some calling for deprecating it completely or the deprecation of other values.

• I have comments but abstain from voting on this proposal. crossing=* tag in general is effectively unusuable. crossing=* tag tries to be a smart shortcut and ends as confusing mess (see crossing=uncontrolled / crossing=unmarked). People are very confident what implies what and when given values should be used. Sadly, they disagree with each other. I do not see point in replacing one crossing=* with something that still includes crossing=* key --Mateusz Konieczny (talk) 21:41, 9 May 2025 (UTC)
• I approve this proposal. --AntMadeira (talk) 22:06, 9 May 2025 (UTC)

• I approve this proposal. --Steglitz (talk) 22:30, 9 May 2025 (UTC) I think this is a step in the right direction. I'm not a fan of having marking types being mixed in with crossing types (e.g., crossing=unmarked). Per our discussion in the Tagging chat, crossing=unmarked could also be removed in favor of crossing:markings=no, but that's for another proposal. I think this proposal moves us closer to a type + marking type model, which I think is a better way.
• I approve this proposal. --WalkerB (talk) 02:45, 10 May 2025 (UTC)
• I approve this proposal. --Dknelson9876 (talk) 01:03, 10 May 2025 (UTC)
• I approve this proposal. --Carnildo (talk) 04:44, 10 May 2025 (UTC)
• I approve this proposal. --Dimitar155 (talk) 04:58, 10 May 2025 (UTC)
• I approve this proposal. --501ghost (talk) 05:39, 10 May 2025 (UTC)
• I approve this proposal. --AmOosm (talk) 05:57, 10 May 2025 (UTC)
• I oppose this proposal. despite I'm very much in favour of deprecate crossing=zebra, the suggested remplacement crossing:marking=zebra is wrong because crossing:marking=zebra means that there is only one series of markingss, information which is not present in crossing=zebra (it could be zebra:double, zebra:paired,, zebra:bicolour, zebra:rainbow) Marc marc (talk) 06:26, 10 May 2025 (UTC)
• I approve this proposal. A thousand times yes! --Riiga (talk) 07:04, 10 May 2025 (UTC)
• I oppose this proposal. Firstly, there is no need to deprecate crossing=zebra. Many crossing=zebra's are in fact zebra crossings, as defined locally. I am all for crossing:markings tags, but I am dead against crossing:markings=yes, which this proposal advocates. If you know there are markings, you know which markings, so tag that. Thirdly, the proposal suggests to replace crossing=zebra with crossing=uncontrolled. As it happens, a lot of zebra's are controlled. --Peter Elderson (talk) 08:27, 10 May 2025 (UTC)
• I oppose this proposal. The top level crossing key should be used for that what gives the most information, just like crossing=traffic_signals, a zebra belongs on top level. --Emvee (talk) 09:21, 10 May 2025 (UTC)
• I oppose this proposal. Since I'm voting on all parts of the proposal, I'm going to oppose it because I think the crossing=* key should be deprecated as a whole and this includes the values crossing=uncontrolled and crossing=traffic_signals also. Please change the tagging in your proposal to include crossing:signals=* and I'll happily vote yes. --Pavvv (talk) 10:03, 10 May 2025 (UTC)
• I oppose this proposal. uncontrolled is a confusing name and marked is more generic so requires an additional tag to say which markings, while zebra is concise and implies a crossing marked with zebra stripes, without traffic signals. Who has priority and who must yield may depend on the jurisdiction, but this isn’t solved by the suggested replacement either. Maybe it depends on the area, your proposal seems to suggest you have been looking at the usage in the US, maybe the scope of the proposal could be limited to the US. —Dieterdreist (talk) 12:58, 10 May 2025 (UTC)
• I oppose this proposal. "no", for many of the reasons stated above, and also because where I am "what makes a zebra crossing" is more than just stripy markings on the road. I'd suggest looking at the usages at https://taginfo.openstreetmap.org/search?q=zebra#values and asking "if tag X is set, does that, on its own, mean that this crossing is a zebra crossing" (and in most cases I genuinely don't know the answer). To echo Mateusz, it'd be better to clarify what the other tags actually do mean and just ignore the one that people think means different things. —Preceding unsigned comment added by SomeoneElse (talk • contribs) 14:14, 10 May 2025
• I approve this proposal. I’ve said a bigger piece about how two four wrongs don’t make a right and we should leave crossing=uncontrolled out of the proposal. But mostly I feel like we should finally move on. Whatever its proponents intend, crossing=zebra effectively doesn’t convey much information in practice. crossing_ref=zebra was approved as part of the infamous proposal that gave us crossing=uncontrolled, so really this proposal is nothing more than a restatement of fact. If nevertheless some regions want to cling to this tag anyways, let them cling to it, as with so many other deprecated tags. – Minh Nguyễn ^💬 16:42, 10 May 2025 (UTC)
• I approve this proposal. --Computae (talk) 17:05, 10 May 2025 (UTC)
• I approve this proposal. I actually agree with everyone above that crossing=uncontrolled is a terrible and confusing tag name, but in my opinion the core part of the proposal (“deprecate crossing=zebra in favor of crossing:markings”) is definitely a step in the right direction out of the current mess. Next proposal: deprecate crossing=uncontrolled? --Willem1 (talk) 17:59, 10 May 2025 (UTC)
• I approve this proposal. — Koreller (talk) 20:34, 10 May 2025 (UTC)
• I approve this proposal. --Perenniallylate (talk) 21:55, 10 May 2025 (UTC)
• I approve this proposal. As many others have already said: A small step in the right direction. I hope the momentum can be used to overcome crossing itself/to continue renewing the whole crossing concept. (And yes, I too will never use crossing=uncontrolled :) --Supaplex030 (talk) 22:05, 10 May 2025 (UTC)
• I approve this proposal. --Akadouri (talk) 23:03, 10 May 2025 (UTC)
• I approve this proposal. If all crossings marked as such has indeed crossing:markings=zebra and are actually crossing=uncontrolled, then by all means. Just try to double check or just make sure people know that crossing_ref=zebra means it was not checked and for instance delete this tag after confirmation that markings and crossing tags are valid --Yog Sot (talk) 00:54, 11 May 2025 (UTC)
• I approve this proposal. In the Netherlands some zebra crossings have traffic signals, so crossing:markings=zebra can be conveniently combined with either crossing=uncontrolled or crossing=traffic_signals. --JeroenvanderGun (talk) 12:02, 11 May 2025 (UTC)
• I have comments but abstain from voting on this proposal. I'd like to support this effort, and I was intending to vote yes. But there is a pretty big problem: This proposal implies that crossing_ref=zebra refers to the legal status. However, the wiki page for crossing_ref=zebra does not say that, which has (and will continue to) lead to confusion. And this proposal has no plans to clarify the definition on the crossing_ref=zebra page :/ --Kylenz 12:13, 11 May 2025 (UTC)
• I approve this proposal. --Wolfy1339 (talk) 13:57, 11 May 2025 (UTC)
• I approve this proposal. --Endim8 (talk) 15:23, 11 May 2025 (UTC)
• I approve this proposal. --Wolfgang8 (talk) 18:26, 11 May 2025 (UTC)
• I approve this proposal. The exact replacements will differ by region, and existing tagging is a mess. That's a job for the wiki. This proposal enables sensible tagging by removing a confusing duplicate. --Jofban (talk) 18:56, 11 May 2025 (UTC)
• I have comments but abstain from voting on this proposal. I have mixed feelings about this --chris66 (talk) 08:11, 12 May 2025 (UTC)
• I oppose this proposal. A very good description of the problems, but not a proposal which I feel to ready to vote on --voschix (talk) 15:48, 12 May 2025 (UTC)
• I approve this proposal. I'm in favour of deprecating crossing=zebra. However, when re-tagging, mappers and editor developers should not automatically change crossing=zebra to crossing=uncontrolled + crossing:markings=zebra, as crossing=zebra has also been used to mean crossing=traffic_signals + crossing:markings=zebra! --Dafadllyn (talk) 16:27, 12 May 2025 (UTC)
• I approve this proposal. Do we need a tag for every type of marking? No. crossing:markings=* offer all the eventualities we might need to cover and does so clearer and with the same amount and greater precision. At some point we need to move forward. --JPinAR (talk) 22:09, 12 May 2025 (UTC)
• I approve this proposal. Finally, a step in the right direction that actually has momentum. It's not perfect, for reasons well explained by others, and it's time to start moving away from crossing=*. --Lumikeiju (talk) 02:00, 13 May 2025 (UTC)
• I approve this proposal. --Langläufer (talk) 14:08, 13 May 2025 (UTC)
• I approve this proposal. I think it is a step in the right direction towards un-overloading crossing=*. --Popball (talk) 14:11, 13 May 2025 (UTC)
• I approve this proposal. It's a step in the right direction. We should deprecate the crossing=* tag anyway, as the original 2008 spec was not designed in a way that produces data useful for pedestrian routing. --Bhousel (talk) 14:33, 13 May 2025 (UTC)
• I approve this proposal. This tag is ambiguous to the point of hopelessness. However, global tools should firmly avoid making changes without asking for an on—the—ground survey (which is likely to be needed in some places anyway even with local knowledge), otherwise they will contaminate another tag. --Andrew (talk) 17:12, 13 May 2025 (UTC)
• I oppose this proposal. The tag crossing=uncontrolled is absolutely confusing and should not be featured in any way. --drolbr
• I approve this proposal. --JakubS 13:24, 16 May 2025 (UTC)
• I approve this proposal. crossing=* is a bad tag, this is a good start for deprecating the other bad keys (unmarked, uncontrolled, etc) --DwarfNebula (talk) 14:01, 16 May 2025 (UTC)
• I approve this proposal. --Bauer33333 (talk) 04:34, 17 May 2025 (UTC)
• I approve this proposal. A very necessary next step even if further refinements are needed. HellMap (talk) 09:50, 20 May 2025 (UTC)
• I approve this proposal. --Aradrin (talk) 07:44, 21 May 2025 (UTC)
• I approve this proposal. --Cquest (talk) 07:48, 21 May 2025 (UTC)
• I approve this proposal. Thanks God, finally! Would have included other crossing values as well, but that's a start. --Lejun (talk) 07:50, 21 May 2025 (UTC)
• I approve this proposal. I'd like to see all other values of the thoroughly skunked crossing=* key except crossing=no deprecated, hopefully this is a nail in its coffin --Rskedgell (talk) 10:01, 21 May 2025 (UTC)
• I approve this proposal. --Tordans (talk) 10:18, 21 May 2025 (UTC)
• I oppose this proposal. Zebra is a crossing type, not a marking, and change as fundamental as this will just create mass retagging edit wars TomH (talk) 14:48, 21 May 2025 (UTC)
• I oppose this proposal. This again? No, of course not. --InsertUser (talk) 23:08, 21 May 2025 (UTC)
• I approve this proposal.--syntex (talk) 22:37, 22 May 2025 (UTC)
• {{Vote|no}} While I don't doubt the good intentions here, what is suggested here (such as it is) would likely make things worse rather than better. A far better approach would be to concentrate on getting other tags describing particular elements of a crossing correct. SomeoneElse (talk) 12:52, 23 May 2025 (UTC)

  This is a double-vote. – Minh Nguyễn ^💬 08:12, 25 May 2025 (UTC)

• I approve this proposal. Nadjita (talk) 14:10, 23 May 2025 (UTC)
• I oppose this proposal. I think that there are not enough details for all the different variations of Pedestrian crossings nor the differing country uses --TonyS (talk) 15:25, 23 May 2025 (UTC)
• I oppose this proposal. This well though out attempt to provide a solution to a long standing issue But, I believe this deals with one aspect but will make others worse. --JassKurn (talk) 19:13, 24 May 2025 (UTC)