Wiki page for Key:industrial has been completely rewritten. Was this discussed or voted on?

https://wiki.openstreetmap.org/wiki/Key:industrial

This page was completely rewritten by user Hiausirg in September. It defines over 100 values, most of which have <100 uses, and many <10. I find many values excessively specific and of questionable use. They also moved industrial=oil (40% of industrial=* uses) to the scary-looking PossibleSynonym template list, implying it shouldn’t be used. industrial=factory, which accounts for another 6%, was also listed as deprecated, which I’ve just reverted. I remember this has been discussed before and was deemed controversial.

Inventing new tags is one thing, but essentially redefining an entire key used a quarter million times is a whole different thing. I think all of these changes should be undone and require a proposal with RFC and vote before rewriting the wiki (unless this was already done and I missed it). Thoughts?

Inventing new tags is one thing, but essentially redefining an entire key used a quarter million times is a whole different thing. I think all of these changes should be undone and require a proposal with RFC and vote before rewriting the wiki (unless this was already done and I missed it). Thoughts?

IMHO establishing “industrial” tags for features is not desirable, this kind of thing should have “man_made” as key. Subtagging landuse to represent individual features specifically is not where I believe we should go.

This user has on a number of occasions had problems working with the rest of the community. See also here and here. I don’t think that they should be rewriting wiki pages from scratch without discussion.

For the avoidance of doubt, this isn’t a “bad faith” thing - no-one doubts that they mean well. It’s just that the results are not always beneficial.

1 Like

The largest edit to the page seems to be this one. Hiausirg even went as far as to remove the “questioned” template in this edit, which has been on the page for a long time for good reasons. I strongly disagree with the changes and the fact that they were not discussed beforehand, and they should be reverted in full.

1 Like

I just want to know if industrial=????????? is proper British English spelling. Here on the other side of the pond, we tend to mix a few ¿ in for balance.

1 Like

Technically the content was debated in length, but was implemented directly without broad discussions outside a wiki section (or consensus, or voting on a proposal) Talk:Key:industrial - OpenStreetMap Wiki
There has been an attempted revert by the author @zluuzki Key:industrial: Difference between revisions - OpenStreetMap Wiki

I have suggested that Hiausirg discusses it here.

1 Like

so whats the issue with my changes now?

the changes have been discussed on the articles talk page over multiple months. So this answers the title of the page, could have asked me directly about that

I did not redefine the key, I gave it a proper definition: How it is used and how it makes the most sense. The old version of the page is a giant mess, contradicting everywhere, and being a pain to work with. You would knew that if you would actually map industry.

Old/now again current definiton: Mapping industrial can describe the land landuse=industrial, Buildings or facilities man_made=works + product=, or railway industrial use usage=industrial.* What is this??? Three completely different things. This “definition” makes this key entirely useless for data consumers.
An older definition from the first version of this wiki page: You can describing the type of industry using the tag industrial=* Okay, understandable!
And now my definition: Use industrial= to mark an industrial facility. Not much different from the orginal definition, isnt it? This gets more obvious when comparing the rest of the page.
So did I really change the definition, or did other people do that by mixing in other stuff?

Where in the world does this nonsense idea come from that “industrial” is a subtag of “landuse=industrial”? It was never intended as such, and it dosen’t make sense, in any way.

Of course. This user is a real jerk! That’s the first thing you need to know when this user is involved in anything. Should probably rename myself to reflect that somehow, so that no one is forced to post that everywhere whenever i’m involved anywhere.
But seriously: instead of any actual arguments for anything, this is the first thing which you need to post, each time I had some disagreements somewhere about something entirely unrelated to this. What exactly is the point of this other than to discredit me/my arguments? (let’s just ignore the “without discussion” lie)

That’s like the most insignificant change out of everything?

Agree. There were very good reasons for the template, because the page was/now is again a giant mess … and I tried to fix those issues. Well now they’re back

care to explain?

“the fact” … pretty confident for such a blatant lie? You didn’t look into the topic for more that 2 mins, -otherwise you would have at least looked at the talk page- all you had to know was that I was involved, and because of that you strongy disagree with it. Congratulations!

Now we’re getting to the real important issues … the fact that this link is more clicked than the actual edits the thread is about makes me lose faith in humanity

A thread is created, and it takes 6 days until the person who is responsible for the stuff discussed within it is notified of the threads existence… sad…makes me wonder if there is a real motivation to actually fix the issues


Now moving on, what should be done? i’d say restoring to my version, because clarifying the definiton and adding many examples are not significant changes like redefining keys
What exactly are issues with my version, and how can we fix those issues? Reverting to the old version and abandoning it is not a solution. (@Wynndale)
(Note that before this has been discussed over multiple months, I worked on it over a month, and tagging industrial sites is one of my main editing activities/intrests. So i’m not lust looking for some quick mass edits I can do, I actually know what I’m talking about)
Again. I didn’t redifine the key, I clarified the definition and added a list of examples, which cover de facto almost every possible type of industry. If there aren’t any examples, people will just free-type any stuff again, which is against what the key is supposed to be. Also, there have already been multiple attempts at doing exactly this type of cleanup in the past, to fix this mess of a page, but they have been abandoned for unknown reasons.
Before you comment on anything make sure to actually read my version and understand what is about, and don’t judge based upon two sentences you saw somewhere yeh itz sh1t. Thanks.

Those two values are actually used significantly (100k and 15k uses) and are not includes in my list of examples. The reasons are that they are duplicates/too specific. =oil does just mean “this object is somehow related to oil industry”. It could be a refinery, a oil well site, a pipeline valve station, a pumping station, a oil tank farm, an oil port … each feature already has its own, specific tags. There is no need for an overly broad tag for all those things.
industrial=factory is an exact duplicate of the man_made=works tag (which has 150k uses).
Definition industrial=factory: A factory is an industrial-scale production site that brings together a large number of different operations and produces products largely with the help of machines, production workers and plant management.
Definition man_made=works: The term works is used to denote an industrial production plant, also known as a factory.

man_made=works should describe the fact that it is a factory, and industrial=* then describes which type of factory/industry in a way, which can be systematically queried for, but also in a way, which is not too specific and not too broad - e.g. “man_made=works, industrial=industrial_machinery, product=paper_production_machines”.
The first two keys/values can be systematically queried for, the product key is then free type to what exactly is produced. This also works for non-factory industrial establishments, e.g. industrial=scrap_yard. Not a factory, no “product” produced, but still an industrial establishment, which can be tagged using this way. (But note that all this is already explained on my version of the industrial key page)

The problem, in my opinion, is not that you made an effort to fix things. I actually appreciate that a lot. But such far-reaching changes to the definitions of tags should be discussed in a formal tagging proposal, and you failed to do so.

I also don’t appreciate your tone. Come back when you can talk without making everything you say sound like you’re challenging your classmate to a fistfight on the parking lot.

How exactly are my “changes” to the definiton “far-reaching”?
I just compared above: The changes I made to the definition are, when compared to the original version, only a slightly different wording. The usage of the key is basically the same. Compared to the current definition, the difference is larger, but the definition changes between the current and orginal definitions are the ones which have actually not been discussed anywhere, and are way more far-reaching than my changes.
The only “far-reaching example” I can think of you thinking of is the 100k usage industrial=oil, over 1/3 of all industrial=*. However, that is also in the old/current version of the definition a badly worded value, and better documented values such as industrial=refinery or =wellsite exist already. So that’s nothing where I just changed the definition for.

I don’t appreciate people who are unwilling to put some basic effort into researching the topic they’re talking about, before they just spread easily avoidable misinformation, as if that’s nothing. What you say is more important, than how you say it, IMO. But you’re probably right, less being less aggressive is still better.

Also, you said you “strongly disagree” with my changes. I am open to find solutions for those problems.

This sounds much more agreeable. I’ll do my research and get back to you.

1 Like

I briefly did my research and I’ve come up with a list of more specific things I do not like about your changes to key:industrial and a single thing that I do like:

List of changes I do not like:

  • The list of required tags for industrial=*
  • The explicit prescription (rather than a description) of industrial=* as a primary key without broad community consensus / proposal
  • The prescriptive statement “Ideally, the large industrial polygon should be split up”
  • The prescriptive statement “It should be used in combination with all {{key|industrial}} tags”
  • The statement that product=* should be as detailed as possible
  • Giving a meaning to previously undocumented tags without documented research or a proposal
  • The large overlap with man_made=works that you introduced without a formal proposal

Thing I do like:

  • The added description about the difference between craft and industrial

My conclusion is that I stand by my earlier statements.

1 Like

For the record, I’d like to summarize what’s happened in the last couple days.

On the 14th, I reverted the changes to key:industrial with the comment “revert to pre Sept 17 rewrite due to redefining tag without proposal”. I also left a message on Hiausirg’s wiki profile talk page, saying “due to redefining the tag without proposal or discussion”. I made a couple mistakes here. I should have left out the “or discussion”. OSM has so (too) many ways of communication that I should have been aware of the possibility that it was discussed somewhere I hadn’t thought to look. I also meant to invite Hiausirg to the forum to discuss the matter, but I forgot. This was especially unfortunate because I unchecked the “watch this page” option, so I wouldn’t get notified of replies.

When my edit to key:industrial was reverted, I didn’t see the response on Hiausirg’s talk page. At this point, I started wondering who is in charge of the wiki and what to do when there is an editing conflict. I clicked the help link in the sidebar, then the disputes link on the help page. The disputes page is all about map data and doesn’t mention the wiki, but not knowing who else to contact, I emailed the DWG.

Andy Townsend replied, informing me that the wiki is managed by wiki admins. (I don’t think the wiki communicates this well enough, but I digress.) He also linked back to Hiausirg’s talk page, where I finally saw their reply that the changes had been discussed on the key:industrial talk page. I don’t generally check talk pages due to poor UX and use the forum instead (and I don’t feel like the talk page is the most appropriate place to discuss such major changes). Hiausirg said my rollback of the page has not been discussed, unaware of this forum topic (another example of the communication difficulties caused by having too many ways to communicate), which Andy then linked to.

I apologize for the confusion. As for the wiki page, I’ll post my thoughts in another comment soon.

4 Likes

What I like of the then and now version: sweet and short. What I do not like of the interim/reverted version, section Comparison of related tags, much too wordy, rather a section Examples can convey the same in a much more reader friendly way. And please below section How to Map, and please section How to search the list below Values.

Living in a place with little industry, I though can understand, that industry is a thing in itself. What is a top level tag? building, man_made, landuse no problem. industry on a node/way/area without any of those? Samples please :slight_smile:

PS: Tourism regarded an industry here, even though tertiary sector. Looking at what it entails, I do not wonder. In other regions primary sector might also be industry. Certainly not so here.

We are talking about industrial= as in landuse=industrial here. There is another industry= I mentioned there, that might be used for identifying any industries.

I created a page to work at changes of my version.

I rewrote this section, based upon the shop key page. It should be more clear to read, and sound less “authoritative”.

  • The perscription of industrial being primary has been removed.
  • “large industrial polygon should be split up” and “it should be used in combination with all industrial=*” has been removed
  • “statement that product=* should be as detailed as possible” has been removed (this is not against the definition/usage of this key, but yeah, it also isn’t within the definition)

The purpose of this long list of possible values is to cover every possible type of industrial establishment, and not just the handful already established tag values. Otherwise, people have to make up own values, which will not be documented anywhere at all and never meaningfully used other than 1 or two times by the person who made the value up. There are already 1,200 values in taginfo, the majority being used less than 5 times - thats because only a few industrial values were documented in any way - too few.

It is important to highlight that none of those values should be considered as somehow “approved” by proposal or considered “authoritative” or similar. This list is not complete, there are probably values which are too broad and values which are too specific, there are values with bad descriptions, with bad names, bad examples and so on. But as mentioned, the list of values is intended as a guide in this complex world of industry tagging, and I think it’s definitly better than the old version of “these ten values exist, make up the rest for yourself”.

Unsure what you mean by that. industrial=factory and man_made=works were always overlappings, I didn’t change the definitions that the use of those tags would change.

works is a pretty old tag, significant usage started in 2010, wider usage of industrial started in 2018. Back then, factories could be tagged as man_made=works, and the product produced as product=*. However, this scheme is flawed, as it is not possible to determite the type/category of industry, which i’d say is more important than the exact products produced. Also, it is not possible to tag industrial establishments, which are not factories, such as scrap yards and warehouses.

An example: man_made=works+product=steel is a used combination. But is this now a steel mill, where metals are smelted, or is it just a metal processing factory, where steel parts or so are built together? This is a key difference which is not obvious from this type of tagging. Adding industrial=steel_mill or industrial=metal_processing solves the issue. And there are countless more examples of such problems.

That explains a lot, thank you!

It is indeed, but the problem is that this is too short for the complexity of this topic (also explained above in this post) Compare for example with the key description page for shop=*. The list of values is just as long than my industrial values list.

You’re right, I tried to improve that. (based on your suggestions and on the key description page for shops)

(Unsure what exactly you mean) Well, I think industrial should be viewed as a top level tag just like shop/office for commercial/service/tertiary establishments - currently there is no reliable way to tag primary and secondary industrial establishments, besides some exceptions like power plants.

shop=* / office=* are establishments commonly located within landuse=commercial/retail, and industrial=* are establishments commonly located within landuse=industrial, or more commonly, because of their size, they just share the same way. (Example, Example, Example) (Note that there are also cases where shop=* mapped as area share their outline with a landuse tag, because of their size (consisting of multiple buildings), such as here)

This is also a possiblility, but it is way more far-reaching than my changes, as it would mean that almost all industrial=* values used currently would need to be moved to industry=* - but for what benefit? What should the “industrial” key be used for then? Because the industrial key has the same name like the industrial landuse value, dosen’t make industrial=* automatically a subtag of landuse=industrial.

Umm, industrial= was once proposed https://wiki.openstreetmap.org/wiki/Proposal:Industrial , and is found on ~80% of landuse=industrial ? Only ~2% of industrial= are ["industrial"][!landuse][!craft][!building][!"building:part"][!man_made][!office][!shop][!amenity][pipeline!=substation] (especially excluding building=industrial as well), It doesn’t automatically support your attempt to make it a feature either. 3545 of these are node , suggesting they are cases of considering landuse= can’t be used as a point. For the remaning 1466 wr , there are still other situations, eg lifecycle disused:landuse= , or power= without landuse=industrial .
While I find many of them should be refactored, industrial= is not exactly the same as industry= . Viz =warehouse (debated by you and others before) might be found in all industries. An industry= will have different facilities. At a glance, it is not the case that “almost all” industrial= used now will be moved directly. And what about vertical integration, different end products, and different stages of productions? That’s some question that needs be considered.
In the end, the topic here is not whether other possibilities that are more, or less, “far-reaching” than your changes. It is that you are changing the existing documentation with mixed content, and without consensus or reflecting the usage.

According to Taginfo, there are 1,299 instances of man_made=works and industrial=factory, which accounts for 0,6% of man_made=works and 8,7% of industrial=factory. That’s only minor overlap.

That’s stuff to put into a proposal and to discuss in the RFC period.

That’s way too prescriptive. Try to be more descriptive. That means doing your research on actual usage, not what the tags should mean.

A proposal draft was started (not formally proposed) …11 years ago, over eleven years! This is how long people tried to fix this industrial tagging problem and we still haven’t found a solution!
As shown by you and also with examples above, " industrial=* " is very often combined with landuse=industrial. Thats not because industrial=* is a subtag, it’s because industrial=* features are best mapped as areas - just like landuse, so they often share the same way. There can also be a large landuse=industrial polygon with many smaller industrial= polygons or nodes for individual facilities.
There is absolutely nothing wrong when industrial is combined with landuse=industrial or building=industrial.

I don’t quite understand that. building=warehouse and industrial=warehouse is not the same. building=* describes building type, nothing more. The building itself could be abandoned, it could be used as an art exhibit. industrial=warehouse describes that there is a warehousing company. There could be multiple warehouse buildings, but they belong to the same company/facility. industrial=distribution indicates that there is a distribution center by a company, for example Amazon Fullfillment Centers. This is different than a warehouse building which is part of a larger factory/facility.

This is all stuff which can be considered but… it’s been 11 years since this draft proposal you linked was started. And we still haven’t documented reliable ways to tag the most basic industrial facilities. As long as we don’t have some basics, we shouldn’t worry about this, otherwise we’ll be sitting here also for the next 10 years and won’t move a single step forward.

Well, this thread was started, because my changes to the page were accused of being too far-reaching, which I don’t agree with. I was not “mixing” existing documentation/definition, I was removing things which were mixed into the orginal documentation/definition.
“reflecting the usage” - the current usage dosen’t exist. Everyone uses this tag slightly different, because there is no clear documentation. And as long as there is no clear documentation, there won’t be any clear usage.

This dosen’t mean anything. Both tag’s definitions are the same. Of course not every OSM feature has both tags. landuse=reservoir is considered a duplicate of natural=water+water=reservoir. There are 260k landuse=reservoir, and only 28k of them have natural=water, and only 10k have water=reservoir.

I do not want to find and discuss the 100 perfect values, if anyone is expected to to that, this mess will never ever get fixed. I want to have a usable scheme, and when someone has improvements to that scheme, then just add those improvements. We should use this to start off, then see how things develop, keep improving and collecting experience with the values/their problems and eventually start formally proposing individual things when the whole thing matures somewhat.
That’s how OSM started - there was some veeery rough guidance, and things developed. No one sat there in 2004 and wrote the perfect proposal for the shop or amenity key.

I did research into the tag values listed on my page (creating this page alone took +1 month), and in most cases, they match the description.

Important to highlight: I’m not here to forge the perfect tagging scheme and talk all day about how things could be mapped, I’m here to actually map something - but to do that, I need some guidance to orientate myself on (not just make up random values how it has been done so far, and I assume other people need that too). And if no one managed to create guidance for something as basic as that over the last 11 years, then this means that I need to create something to work off. I don’t think it is my responsibility to take every slightest possiblility into account - because if anyone is expected to do that, we’ll never getting anywhere. If anyone thinks this is expectable … then you should simply do it. Cause’ no one did so far, 11 years.

2 Likes

People take issue with your content, meaning that it may not be as perfect as you believe it is.
And if you can’t be bothered to discuss it and to build a community consensus, it indicates that you’re already well aware that it’s not going to pass quality checks by other people, let alone a formal vote. You may prefer to keep your documentation somewhere in your private projects rather than in the community-driven Wiki.