It should be done if the relation is too large and a super-relation is a replacement, at least that is what the guidelines are saying. Again anyhow, it is just a guess. You have edited the two cycling relations before so perhaps you'd know why they are tagged as they are now.

All in all a networked route (with multiple interconnected highway=cycleway connected into one osm.wiki/Tag%3Aroute%3Dbicycle) may not be the norm outside Asia but it happens for reasons (one of which I think is for saving server resources). It is not breaching the guidelines and the flexibility is there for region-specific situations so as long it is conforming the major guidance applicable across OSM. If we are able to reach a consensus in the mapping community that this should be transformed into a Relation:route_master instead then we could do so.

(For your 2.2, I am well aware of it and thus I have raised in a "strictly speaking" matter should we decide to dwell into such precision on directionality)

i. We would have been able to comprehend your interpretation better had you underpinned that "most" of the walking routes in OSM is located in rural area, or that walking routes should comprise "mostly" rural sections.

Assuming that you have meant the latter one (otherwise my example still stands valid), it would have been the best if the unspoken rules are codified into tagging guidelines/conventions as what other OSM communities have done upon reaching a consensus (say, differences compared to route=hiking). By such, experienced editors such as you and newbies like me could follow better. I am sure some of the discussions in this thread, also seemingly not explicitly detailed in the guidelines such as the naming convention, are also worthy of being listed in the respective guidelines.

ii. for example, two bicycle routes relation/10553685 and relation/10619470 are both grouped region-wise without a clear main direction. Strictly speaking, some of the route=ferry do not contain separate forward & backward relations as route=bus and route=road does. If there has to be a clear to-fro mark for all relations, then we will have to take care for all of these as well.

A random search lands me at relation/1623950, an urban walking route tagged with route=hiking. "Mostly rural" I imagine is not exclusive?

i. The feature of walking route (the right link should be this instead: osm.wiki/Walking_Routes) covers both roads or trails. If there is a consensus among the community that there should be a narrow definition of urban routes, then it would be another case.

ii. Yes directions are encouraged in route feature, but as far as I am aware it is currently mostly found in route=bus and the major highways. If for perfection branches in the footbridge system would have been divided up into several routes, but as said, for simplicity without compromising validity, they have been combined into one relation without a osm.wiki/Relation:route_master.

I have changed it to osm.wiki/Relation:network as a compromise. Again, osm.wiki/Hiking dictates that so as long it is a "shared common knowledge" then it could be marked on a map as a walking route. This is in addition to the signage currently present and visible in the malls and on the footbridges. Furthermore, I believe quoting osm.wiki/Relations_are_not_categories is only weakening your argument because this is not the case, which I have stated already.

On ii., let's just say that "x in y" and "x of y" are generally accepted as names if they are marked and denoted in real life. If not, it should have been given careful consideration because it reassembles some "descriptive names". But it should not have been outright rejected conclusively and completely.

i. In my last comment I am once again stressing that there should not be a stringent rule to determine if a name should be included. I am not questioning osm.wiki/Ground_truth in any sense, please. In this case, both the footbridges and the underpass per se are mapped rightfully under the rule. But as said, if we need a direct observation of a signpost, then the name of Wuhu St. Underpass would not exist.

ii. There are names of sovereignty states appearing on OSM that fulfill the form of "x of y". But yes, thank you for pointing out that they are not the norm, but it doesn't mean they are unreservedly excluded.

iii. As said, it is a walking route that runs crucially from north to south and had gained its notability for the increased accessibility. I don't see this is something that contradicts the osm.wiki/Ground_truth or other guidelines on OSM.

Thank you for making it clear to me and others. Wuhu St. underpass (whether capitalised or not) appears not officially demarcated by a Government Notice. If I was not mistaken it also seems not to be the sole, authoritative name for that covered segment of the Chatham Rd. N. when there is alternative name found on government websites. In any case, I am not trying to challenge if it is right to have Wuhu St. underpass or other street names in the area that got into the same situation (the fact is in my view they are done appropriately). Rather, this could disprove a hard-and-fast golden rule that a name should appear on the OSM map only if it is signposted or with an official extent. Similarly, if "x of y" is rejected without exceptions, wouldn't there be many names removed across OSM database?

And to general consumers of OSM data such as me, highway=* is more recognisable than this walking route (whether it is network=* or route=*) as it would be visible from more rendering applications. I'd also like to put it to you that I am not trying to hold out against reclassifying the type of relation of the footbridge system, and I welcome your comments on this. It is just my limited experience on OSM that guides me to the decision that "route" is the most applicable option.

i. Thank you for letting us know your personal preference on this matter and that "If there's not a signposted or official extent, it should not be added" is not a strict rule applicable everywhere

ii. I would welcome your enlightenment on this - why is "x in y" categorically not accepted and "x of y" the opposite?

i. I am sure we could at least accept names that perhaps are not falling into either category and are established as common names, such as way/151396680.

ii. I am referring to your rejection of names in the form of "x in y" and by that standard all "x in/at/on/of y" would have been rejected as well for being synonymous with what you considered to be a descriptive name (would "Footbridge System of Tsuen Wan" have been accepted in place of "Footbridge System in Tsuen Wan"?).

i. I am well aware that this relation is *not* for the extension project but rather the existing structures erected and serving. Indeed "Footbridge Network in Tsuen Wan" in the gov website did not specify the extent of it, but through media reports and the local knowledge, the meaning of the network had been established to cover elevated walkways & indoor corridors that are interconnected in the area; and that the name of footbridge network here in the osm is intended as the translation for "荃灣行人天橋網絡" which is by its own a common name and official name. However, there is not a widely recognised English name and therefore the official use of it has been adopted.

With your concerns expressed and detailed, I am changing the English name of it to the synonymous "Tsuen Wan footbridge network" and moving the current one into an alternative name. My concerns remain in designating "Footbridge network in Tsuen Wan" as a description - because by plainest evaluation it would have meant that a footbridge at, say for instance, J/O Yeung Uk Road & Texaco Road a member of this relation, which is not what intended.

ii. At least I think we should not categorically consider "x in y" a descriptive naming. In doing so, won't that rule out all "x in/at/on/of y" terms as generic names? I agree we have to be more careful when inserting such terms, but some of them have established their own meaning and interpretation on their own and perhaps should take them in case by case.

The connotation of "x in y" cited in the guideline points to the consensus/idea that they should not be "a loose group of somewhat related items", which is not in the same league - it is not loose nor only "somewhat related" in this case. The editor has cited the example of "Scottish Lochs" to prove his case and that "x in y" is not the exhaustive example.

Hypothetically, it would have made a difference if an article on Wikipedia is created dedicated to it, because it would have meant that subject has its own notability. Also hypothetically, there could be an article of "Ancient Roads in East Anglia" that somewhat relates to historical walk in the region, which could perhaps be recorded in OSM. If it is just ordinary, then likely there would be an article of "List of footways in East Anglia" - and that would have been a case that could be rejected by OSM. But ofc, a hypothetical example is not worthy to be dwelled into over here.

Hope it helps. Cheers.

Quite certain it's not for the Chinese (as the name of an tourist spot). The English name could be reviewed. My view is that it is an established common name for the tourist attraction (see the press cutting at the separate OSM notes)

i. I am a bit surprised that I would need to have the wiki page link pasted here, but here you are (name=*), at the third bullet point in the "Values" section; and one of the many names appearing on various government website (https://www.hyd.gov.hk/en/our_projects/road_projects/6145tb/index.html)

ia. I am well aware that there shouldn't be "descriptive name" (e.g. "public toilet", "small temple") or names that should be descriptions but as I wrote this is not the case.

ii. Exactly, thank you. The guideline is referring to "Category:" / "List of" pages which this is not.

i. "key:name" considers official usage as one of the usual sources of primary names. This gives "Footbridge Network in Tsuen Wan" legitimacy to be inserted as a name. Generally speaking and as indicated in OSM wiki, "key:name" is designed to "expect data consumers to expose in a label", while "key:description" is intended to provide additional information, not as a mere alternative to "key:name". This issue has been raised more than once and I am sure Kovoschiz is aware of such views, whether it is from me or not.

ii. The strictest interpretation of "Relations are not categories" guideline would seemingly not apply in this case. The footbridge network has warranted its own notability as a Wikipedia article (regardless as a walking system or as a construction project for the community). The examples the page quoted (Footways in East Anglia and Scottish Lochs) are not apples to apples comparing to this case.

iii. That gives way to the second issue in question - whether there should be a relation created for the purpose of the footbridges. The page you and I cited confirmed that "[relations] are meant to model a close (and usually local) relation between objects", which I see suitable to describe the footbridge network itself.

iv. The third case in question would be whether it should be categorised into "relation:network" or "relation:route". The prime intention of tagging it a "route" is because it runs (crucially) north to south and (as branches) east to west at both near Tsuen Wan station and near Tseung Wan West station. Indeed it could be divided up into several "routes", but alternatively, it had been grouped into one "relation:route" that, in my view, does no harm and does not deviate from the good use of "relation:route". Had it been otherwise it could be contained in one "relation:network" named as "Footbridge System in Tsuen Wan". It is clear that "relation:network" is used in some places to tag relations, but whether it could contain ways instead of relations is not mentioned. The tagging decision is taken with prudence and I am open to any dissenting considerations that would improve the OSM data and align it even better with the intended usage.

2&4. The article you cited rebutted exactly your points - "A good example for a valid and useful grouping is the "route" relation, where multiple ways are connected to form ... a walking route..."

I am referring to (3). As for the use of key:ref, I don't see any requirements that limit its use to those with signpost but I won't object to removing them if you consider it to be most suitable

Please see the official map for the trail at https://www.roundislandtrail.gov.hk/en/project-overview/project-background

The subway exit is metres away from the corner of the covered walkway and not positioned at the end of the east-west path. The "unrealistic" and "casual" snap was trying to show this insitu situation.
In fact I'm glad that you have reviewed my works which you may faithfully consider to be defective. From which it highlights the collaborative nature of OSM that makes it stands out. I'm sure experienced contributors like you would have appreciated this better than I do.

For the record that the mergers are conscious save for a few that are inadvertent