I find the multiple error/warning boxes at the top of the default (per-country) Access Restrictions page quite disrespectful to the communities that put some work to update and rely on these pages.
Your decision to make these changes without consulting the global community, and then reverting someone else’s edits, is concerning. If you have conflicts with your own country’s access restrictions, please address them internally.
You’ve made a valid point about the possible low adoption of these defaults by routers, but discouraging readers from using this page isn’t the answer. We should focus on finding ways to improve adoption instead.
I meant the default access restrictions page per country, Thailand refer to this page in its own country guidelines, and I am sure many of the other countries do to.
The UK section of that page is utter bobbins, so a warning is very much justified there… One issue with the UK section is that there are 3 separate legal systems and the rules in each of the three are somewhat different. It also doesn’t link to any of the information that is actually relevant for those three jurisdictions.
Edit (11 days later): To expand on this, I wrote this diary entry in order to think about what the defaults should be for England and Wales. Scotland and Northern Ireland are of course different.
Then the warnings should be placed under the respective countries section, it’s unfair to warn everyone just because some info is conflicting for some countries.
If it was publicly known that this page has so many problems, what would have been justified is an announcement to deprecate this page, and ask respective countries to take their section to their own country guidelines.
Given the problems mentioned earlier, Template:Questioned - OpenStreetMap Wiki may be more appropriate than the current warning. That sends the reader to the talk page, where the issues can be described and debated in more detail – maybe with a link to the relevant forum threads. Or an inline Template:Dubious - OpenStreetMap Wiki if the problems are confined to individual line items.
It seems reasonable to have some sort of warning box on this page given that data consumers don’t actually use the information in these tables. The warning doesn’t need to be so verbose and there don’t need to be two of them.
Fair enough, but was it always expected that data consumers would support this? A summary of defaults for each country is still very useful for mappers who want to manually maintain the tags.
However, (having spent quite some time on writing the section for my country, the Netherlands) I fully support these warning boxes and after another user deleted these warning boxes -without consulting or discussion- with only a changeset-comment that showed that he did not grasp all the caveats contained in the wiki page, I did restore the warning boxes.
The deletion of the warning boxes was justified in the changeset with
“These defaults are extremely important, and the claim about the Netherlands is factually incorrect.”
and the actual text of the wiki proves both statements to be false.
Despite all the warnings in the actual wiki page, mappers still use the values in these tables as an excuse for deleting valid data from the OSM-database (also in mass-edits) “since that value is default” Since normal warnings in the text don’t seem to be sufficient, a warning box seems to be very much in order.
I do agree that these tables can have some value, and the work that was put in it (also by myself) was not for nothing.
Personally I would like something like this in an introductionary box:
If you are the developer of a router that is looking for possible fallback values you might consult these tables for inspiration, but please note the caveats mentioned in both the general and country-specific section, other values might be more suited depending on your goals and areas.
If you are looking for general traffic rules in countries these tables might provide useful references, but please realize that general traffic rules can often be overruled by traffic signs or access signs in the real world
if you are a mapper please consult your local mapping guidelines and documentation, these tables are not meant to define which tags should be omitted or deleted from the OSM-database
Thank you for the clarifications! Apologies for missing this one—I reviewed most of the recent changes but overlooked this particular one.
From my experience, it’s just not realistic to manually keep track of all the possible access tags for every single way. When mappers in my area tried adding a bunch of the common ones, it only ended up confusing new mappers and making a big, inconsistent mess. This also spread to outdoor tracks and trails, where access tags were used to show if a trail was physically passable instead of using the appropriate scale tags (sac_scale, mtb:scale, smoothness…)
There’s got to be a better way. For example, if we can figure out the most common defaults for routers, we could skip those tags. Like, no need to add motocar=no to a highway=path.
I don’t mind adding motorcycle=yes to paths in Thailand since that might not be a standard for routers. But I’m not up for maintaining foot=yes, bicycle=yes, horse=yes, motor_vehicle=yes, etc., on most road classifications if they’re already set as default by routers.
The first step would be to survey the defaults already implemented by each router and identify any discrepancies among them. If routers have formed a consensus that differs from what’s on the wiki, it’s worth having a discussion about that rather than taking either the routing profiles or the wiki page at face value. Note that some routers, particularly Valhalla, rely on a complex system of heuristics that doesn’t necessarily correspond one-for-one with OSM tags.
However, you seem to be arguing in the discussion section that a page which has been around since 2008 is part of an inactive proposal from 2018. I’ve moved the reference to it to the “See Also” section.
A warning at the top may well be justified, but on the grounds of reliability and lack of router implementation, not a dubious claim that it’s part of a subsequent proposal…
As @SomeoneElse points out, the single section for the UK deserves its own warning in its current form. It really needs splitting into separate subsections.
The DWG mailbag gets quite a few complaints** of the form “xyz app, map or router is routing people along somewhere that it should not”. Unfortunately most of the complaints (by a country mile) are about closed source apps and maps that use OSM data in combination with others, and these often don’t publicise the rules they are following, so looking at what is documented won’t help.
In England and Wales we get quite a few complaints because ways without any access tags are assumed by some apps to be open to all - some apps suggest cycling on a highway=path with no other tags is likely to be OK. It isn’t - arguably it’s the mapper’s fault for not making access rights explicit, arguably the wiki page editors’ fault for the section of the page mentioned above, and arguably it is the app author’s fault for thinking that the road rules they are familiar with (in e.g. Germany) apply in England and Wales too, or that the OSM wiki is somehow authoritative.
However expecting app authors to know all the rules and exceptions - to Allemansrätten in Sweden, to the Scottish Outdoor Access Code in Scotland, to obscure (to me) forest track width rules in some German provinces, is setting a high bar to clear, and you’re not going to do that in a simple wiki table either.
Personally, when mapping I’ll try and make access rules explicit, and when addressing issues reported to the DWG will try and do the same. That way the router authors have no excuse
** More than a few complaints are misplaced - OSM is correct, the router is correct, but the complainant would just like reality to be different. I’m ignoring these “nimby complaints” in this answer.
Right, I only suggested a potential starting point for folks who want to salvage this concept of wiki-defined default access restrictions, but it has a very uphill climb ahead of it. The body of existing data consumers is definitely a limiting factor. Even open-source routing profiles get deployed and forgotten about.
And, uh, wait until someone walks into the room who thinks motorroad=yes should imply certain access restrictions too.
I would be a happy camper if we could even just get the major default routing profiles to fall back to half-decent speeds that aren’t overfitted to pre-pandemic downtown San Francisco when maxspeed=* is unset, based on highway classifications there that didn’t even follow state or national tagging guidelines (true story).
I’ve consulted that table now and then for cycle.travel. I don’t automatically parse it (that way lies madness), but certainly I’ve eyeballed it several times, in particular to look up which countries allow bikes on highway=trunk.
The most concerning thing, besides the bits that are factually wrong (no proposal, this was just documentation), is the preposterous claim “The standard practice is to tag the access rules explicitly …”
Now I realize that the explicit tagging fraction believes it has won because it has a big stick and wields it indiscriminately, but in reality, not only do most have no idea what the actual laws are for their country in detail (just consider HGV categories and other non-automobile transport means), nor does OSM have vehicle/transport mode modelling that is fine grained enough to express such details and neither is OSM tagging expressive enough to model the more complicated rules that likely every country has.
Regardless of if a routing engine dev/operator wants to go in to any detail in modelling or not, the only thing that makes sense from a pure practical pov is to map the exceptions to the rule and not to try to to recreate dozen of pages of legislation every time you want to map a road.
PS: I would note that for whatever reason, it was found necessary to write half a treaty on why the Dutch table is wrong, instead of fixing the table values and adding comments where things don’t really work for OSM, like essentially everybody else has done.
The reference on the wiki to the default-proposal was made in 2010 (fourteen years ago) by user @FrViPofm (the author of the default-proposal) , my first edit on this page is from 2022 (see page history) .
why do you claim to link to the proposal to be dubious if the author has made it himself and it has been on the page for 14 years?
In the proposal access-defaults are also linked to and the default-wiki page has led to confusion that has caused the most problems with these tables:
the notion that the values here should not be tagged on the element (which for some is even today a reason to remove them from the OSM-database)
For a correct understanding of the way to interpret these tables it is crucial that it is clear that that specific notion of not-tagging (ad deleting) all the values from the tables was never a part of an accepted proposal, although some mappers treat it as such (for reasons that are beyond me), shout “that value is needless since it is default , see the default-access page”, and go on to do mass tag-deletions of valid data that often go unnoticed.
Moving this notion from the end of the introduction (where it was from 2021, fine by me) to the very bottom below all the tables where no one will read it does not do this justice, it is really a step backwards your arguments mentioned here seem -how can I put this friendly- not overly researched.
The point is the page we are discussing has a -reference- to the Proposal:Defaults - OpenStreetMap Wiki it -isn’t- and never was “the proposal” (which had some momentum for a while, but was always IMHO too unwieldly).
I’m glad you included the “essentially” there. The page throws up its hands in the air when it comes to the U.S., and for good reason: this country’s regulations have almost nothing to do with highway classification in OSM (or else service=driveway2 would be a thing by now).
You seem to be missing the point here:
there is no “fixing the table” if it is not clear what question these tables should be answering.
As it was pointed out eloquently by the developer of the much appreciated Brouter :
***** Attention *****
This proposal here creates big confusion by mixing two completely independent questions:
(1) what is the legal access if an OSM way has no access tags,
(2) what is the legal access if a (real world) way is not signposted.
While there are still some mappers insisting that both should be similar, since many years mapping praxis established OSM default rules (independent of national regulations).
So please read the national tables on this page with extrem caution and with the fact in mind that it’s not clear what question they answer.
And if you are missing the default access for “track” and “service” in the top (“default”) table in this page, please do not look in “your” national table, but look in the map for examples.
The values answering question 1 can be very different than the values answering question 2 , as the Dutch examples show
And for question 2 there is no single best answer to “fix”, but it is dependent on both area within a country (different operators putting up way more or less signs placing actual restrictions) and a more optimistic vs a more careful approach when assuming actual access when data in OSM is missing)
