OpenStreetMap
There are more than two millions of shop= key entries in OSM database. Definitely, everybody knows what “shop” or “store” actually is. But in the same time, certain vales of this key do not have any practical sense. I can even take the liberty to call these values “unusable”, because they tell nothing about what you actually can purchase in these stores.
I’m talking about shop=convenience (271k entries, top value), shop=supermarket(237k entries, second most used value), shop=kiosk (54k entries), shop=mall (34k entries).
These are “wildcard” values, indicating anything and nothing in the same time. If someone wants to find, where to purchase particular type of goods, it’s impossible to do that in case of these tags, especially - latter two mentioned above.
Malls are not shops at all - these are commercial buildings, full of everything, from supermarkets to cinema theaters and customer service or sales offices. So, every store, amenity or facility there should be tagged individually.
Kiosks are very different from country to country - just several years ago, for example, some kiosks in Russia were selling everything including strong alcohol (currently, it’s prohibited, but there still is some legal loophole, especially for countryside). But anyway, we can never tell, if particular “kiosk” in OSM database stands for something close to news agent, or it sells vegetables and tobacco products. However, building=kiosk does make sense, since concept of small standalone retail building is more or less similar in entire world.
Same thing applies to convenience stores, supermarkets and, at certain grade, to pharmacies (at least, in American meaning, since Walgreens or Rite Aid, being “pharmacies” look exactly like Safeway or Albertsons, being “grocery stores”, with just a bit larger over-the-counter drugs department and without fresh produce department).
Usable scheme for stores should be based on non-exclusive keys, reflecting certain groups of goods (including ones, specific to certain countries, since “alcohol”, for example, can stand for strong alcohol only or for any alcohol drinks, including beer, depending on local or national regulations). Currently, I’m not proposing any ready-to-use scheme. The main purpose of this diary entry is to draw certain attention to this large problem. I think I have to remind, that OSM is not a map, where particular wildcard term can be just converted into a single map icon and label, leaving all interpretation and guessing to end user. OSM data should be suitable for search and filtering.
Ones, who think, that “mall” or “supermarket” are obvious terms, have very limited experience, therefore, their opinion is irrelevant.
Discussion
Comment from CloCkWeRX on 24 April 2016 at 06:01
You realise the key shop in the OSM sense is more describing a physical thing that can be looked at, verified on the ground, and classified to a general type, not what a thing does in extreme detail, though don’t you?
A structure or area identifiable as a supermarket or mall is perfectly valid, and has its own name. It’s a lot better than just building=yes.
Other schema tackles the “business” problem - https://schema.org/Store for example. Not much of that is really applicable to OSM - we’d never want to map online-only stores, reviews, etc; but it doesn’t change the usefulness of that data living somewhere else.
You’ll notice that even in the wiki examples that is describes product values as really broad “focus”; as opposed to listing an entire product inventory down to the last items and stock levels.
I do find it ironic you are apparently unhappy with sweeping generalisations in tagging schema; but are ready to dismiss the observed behaviour of a heck of a lot of mappers “because they have limited experience”.
Comment from BushmanK on 24 April 2016 at 06:47
@CloCkWeRX, where exactly did you get that idea of “physical thing”?
Current Wiki description of
shop=says:If you are not aware of this fact, I have to point on it: many tags in OSM are not representing any physical entities, but describing logical and even abstract properties of objects. Anyway, even if it would be like you saying, practical meaning of values I’ve mentioned above is close to zero, since only minority of people want to just visit a store - they usually want to purchase particular products.
In case of gas stations we do have very detailed and effective scheme, and nobody has to guess, if they can get particular type of fuel and use specific payment method there. How shops are different from it, so it’s okay to make people guess?
Current scheme is not a result of careful planning and development - it’s a product of borrowing inconsistent terms from natural language(s), therefore, mentioning “focus on something” is just an attempt to justify fundamental imperfection of current scheme, which forces mappers to rely on mythical “focus”. This scheme is unable to pass even simple test: imagine a shop, which is called “Alpha”, with two equally sized departments, selling flowers and gifts. You can’t rely on “focus” since there is no one, you can’t rely on name, since it’s neutral. With current scheme you have two ways: 1) randomly pick
shop=floristorshop=giftsand make it incomplete; 2) tag both departments individually with both tags, which will be wrong, since it’s a single business.Your phrase about “listing an inventory down to the last item” sounds like rhetoric exaggeration, because I haven’t said anything about individual items, but clearly mentioned groups of products.
I’m not dismissing any observations, but saying, that observations, limited by particular country, are not enough to make qualified conclusions about any general case (while OSM is an international project, so it should have more or less country-indepentent tagging schemes).
By the way, if someone is going to say anything about “scheme with product groups seems too complicated”, my answer is simple: if it’s too complicated for you, don’t try to bring others to your level.
Comment from zarl on 24 April 2016 at 08:14
While your thoughts may have a point the way you express them you shows everyone else that you don’t care for other thoughts or even a discussion. Your problem.
In reply to the last sentence of your posting: Those who think that other mappers’ opinions are irrelevant seem to have a very limited field of view.
Let me just note that one of the large number of “applications” of the OSM database are i.e. GPS devices. Try searching for one of your overly specific tags (“Hey! I need more strong alcohol!”) out in the field with e.g. a Garmin handheld. So while there is no way of stopping you from adding specific tags to an existing POI (do it! promote it!) please stop telling others that they have mapped everything in a wrong way for the last couple of years.
Comment from nebulon42 on 24 April 2016 at 08:52
Somebody who writes that someones opinion is irrelevant is not worth listening to.
Comment from Sanderd17 on 24 April 2016 at 10:00
Thanks for saying my opinion is irrelevant. But I say that the current schema works very well.
First of all, you’re mentioning the big number of different shops. And then you’re nitpicking on some of the most used values. I know people who complain about the amount of different tags they need to parse, and I know people who complain about the lack of values to describe a feature. I’ve never met someone who complains about both at the same time.
The shop=* works good. I can perfectly see if something is a supermarket, a bakery, a clothes shop or whatever. If it’s some exotic shop, I just invent some tag for the shop, and it gets parsed by most tools as just being “a shop”. That’s perfectly reasonable.
Next to that, there’s no problem with having local definitions for values. I live in Belgium, and I know perfectly well what a supermarket or a convenience store is (we don’t use kiosk or mall very often). And so do all users in Belgium. So it doesn’t matter that our definition is somewhat different to other definitions. People here will want a supermarket to the local definition, and find one.
But as the almighty intelligence ofc already knows, tags don’t change in OSM because someone things some tag is better. Tags change because they are being used by a growing group of users. I wish you good luck with changing the shop tags.
Comment from MMN-o on 24 April 2016 at 13:43
Personally I tend to use, for example, shop=mall for the mall building. And then, when I get time, further details can be added (such as individual shops with higher detail etc.)
Comment from BushmanK on 24 April 2016 at 15:06
Oh, I see, people here just want to compete in being offended. That is typical, but it is, again, irrelevant to the topic of tagging. If certain scheme has serious flaws, why on Earth should someone be offended by fact, that he used it for years without understanding that fact? Constructive way of dealing with it is to discuss how to fix that problem, not defensive reaction. There were completely ridiculous, but super-widely used tags such as
wood=coniferandwood=deciduous. Does anyone here feel offended or betrayed because these tags were proven nonsense, deprecated and replaced with proper tags (leaf_typeandleaf_cycle)? Probably, everybody (including myself) used those obsolete tags for certain period, but constructive reaction on deprecation of them is to start using new ones, not defending old ones, especially, if your opinion is irrelevant due to lack of knowledge of biology. “What was good for ancestors is good for us” is always bad point.To discuss anything, one should be at least qualified for it by having enough knowledge on certain topic, otherwise, discussion always turns into repeating similar weak arguments, based on limited knowledge. It is easier, if such people will just abstain from comments. But now I see, that at least four people here just want to fight for their right to speak. That’s wrong place for it, probably. Nobody infringes this right, but it doesn’t make unqualified opinion relevant, unfortunately.
Using Garmin search as a proof of something is wrong, because one of fundamental rules of OSM is not to map for particular device, but describe real-life objects in certain adequate manner. It means, if certain device doesn’t know how to search for something, we shouldn’t avoid tagging it. In addition, any detailed scheme is reversible by design. It means, that it is possible to indicate, that certain product groups is the most important for particular retail business. So, it should be possible to convert detailed tagging into simplistic categories when preparing data for navigation devices, for example. But opposite process is impossible: you can’t extract any information about product range from
shop=supermarket.Also, practice of assuming that your opponent wanted to say something while he didn’t, shows really poor rhetoric skill. It forces others to discuss imaginary things, which is obviously bad idea.
I do realize that any detailed scheme is only good for minority of mappers who can see, why it’s important. Healthcare 2.0 (which is still just a proposal) is, probably, the best example of it. But I’m appealing to people’s logic, not their feelings about precious old tags. And I’m aware of retrogrades, so I’m not proposing getting rid of those nonsense values - if someone wants them, he can use them, why not? But don’t tell others that there is no better way.
@MMN-o, you are only one having a point.
shop=mallmakes any sense by itself only in form of temporary solution to map commercial building identified by satellite image, for example, without knowledge of what’s inside. This is constructive approach to using it.Comment from BushmanK on 24 April 2016 at 15:30
If someone dislikes my manner of telling about this existing problem - for sure, they have right to feel that way.
But it doesn’t make this problem non-existent or any logic, linked to it, automatically wrong. While thinking, that it does is an ad hominem fallacy.
Comment from MarkusHD on 24 April 2016 at 16:08
You talk about searchability. But natural language, which you call inconsistent for tagging, is exactly the language people use for searching.
Comment from dcp on 24 April 2016 at 16:32
When I joined OSM in 2008, most of my home town was a white space. Being retired and needing something to do I started mapping but it would have been impossible to collect all the related OSM data using all the right tags at that beginning.
There was no Bing or anything similar in those days so it was a Garmin and Dictaphone. It did not worry me in any way that my data was incomplete or faulty though! After 8 years other locals have signed up and improved on my original entries, geometrical, geographically and tagging.
Isn’t that what it is all about BushmanK? Please, just carry on surveying and improve the data. Remember that the mapping standards are often different from country to country and some mappers go to extraordinary lengths to map useless things; eg a roller coaster track, where a single node would do the job.
Sanderd17 comment is extremely relevant: In Belgian they understand what is needed and meant.
PS A shop selling flowers, gifts and alcohol could be defined as shop=florist;gift;alcohol couldn’t it? A good renderer could cope with this syntax!
Comment from BushmanK on 24 April 2016 at 16:48
@MarkusHD,
First, I said no single word about prohibiting usage of those values for malls and supermarkets, so if your comment is about ability to look up particular term, I’m not interfering with this ability.
But you should also understand, that without tagging product range, there is no way to search for something without assumptions. And, as I already said above, detailed scheme is reversible, while current “focus-based” scheme is not. Take a look at my test example of shop selling flowers and gifts. Quality of tagging scheme as well as of any other piece of information architecture should be tested against the extreme cases, not average ones.
Indexing this “flower+gifts” shop for POI catalog of some navigation device, it’s easy to put it into two existing predefined categories (in case if this particular software supports it). But using “focus-based” scheme we will have either incomplete data or double data.
And please, keep in mind, that having no sources of data about product ranges is one of those reasons, why there is no by-product search methods. I already gave an example of widely used scheme for fuel, which is completely similar to what I’m talking about. Also, we have many other tags in OSM, like weight and size limitations for roads, not supported by any well-known search engine or navigation device. But we still tagging these things, aren’t we?
My point is, until there is no usable data, nobody will support non-existent data in any search engine.
Comment from BushmanK on 24 April 2016 at 16:59
@dcp,
Using semicolon-delimited lists of values is something that is suggested, but rarely used and rarely supported because of cumbersomeness of this method. Multiple nodes for a single business is a makeshift thing, and it creates redundancy. So, both approaches are a kind of existing, but in the same time, nobody wants to use it and to support it.
Your point about gradual improvement is true, but I don’t see how it’s related to this topic. Honestly, if you personally don’t care about having better data architecture, why bother? You’re doing some great job mapping your town, and it’s wonderful. But what’s the point in telling other people, that it’s wrong to try improving data architecture, if they want to do that?
Comment from BushmanK on 24 April 2016 at 17:02
@dcp,
And regarding of “in Belgium they understand…” - OSM is an international project, therefore, data should be consistently formed regardless of national flavour. I have another fresh example on this topic right here.
Comment from MarkusHD on 24 April 2016 at 20:28
“Gifts” is a bad choice in this discussion. What is a gift? You can buy any article as a gift for someone - from a cup to a car. You can even buy an article in a
shop=giftsfor yourself, then it is no gift anymore.Now rejecting the “car” example and assuming we have a similar view about what a gift is in this context: A more or less small object, which an average person can carry on its own and from which the majority of people think it would be a good idea for use as a present, sometimes if they just have no better idea. And I’m sure this view also only applies to sort of modern areas in the world, so it is still kind of a local assumption for an international project (even if it’s not bound to national flavors). Now the particular articles, and let it just be article groups or categories, are such a big variety that
shop=giftsis in no way better than or different fromshop=supermarket.Comment from BushmanK on 24 April 2016 at 20:46
@MarkusHD,
“Gifts” is just a random existing value I’ve used for illustration purpose, and there are 20k entries of
shop=giftin OSM, just 1.75 times less common than, for example,shop=mall. You can pick whatever value you like, that fits this case (shop=chocolate, for example), but result doesn’t change.Mentioning gifts as second category of products, sold by this hypothetical two-department shop in addition to flowers, I had no intention to discuss this particular value. My intention was to demonstrate, that currently there is no good, consistent, non-cumbersome and non-redundant way to tag any shop, not “focused” on single category of products. My view on using semicolon-delimited lists is already expressed above in reply to @dcp.
You’ve got carried away with
shop=giftand missed the purpose of this example completely.Comment from MarkusHD on 24 April 2016 at 21:13
Don’t take it too serious. I did understand the purpose of your example, I just explained why “gift” was a bad choice. Since you very much emphasizes consistent and clear tagging, I think it is perfectly fine for a short nitpicking. I didn’t want to start a needless discussion about that specific key.
No,
fuelandpaymentare not “completely similar” to tagging product groups in a supermarket. If you navigate to a fuel station, but you cannot use the offered fuel for your motor vehicle, this station is entirely useless for you and might even be problematic for your further tour. If you arrive at some shop without cash, but it doesn’t accept your debit card, then this shop is entirely useless as well. But if you visit a supermarket, which doesn’t offer bread and juice, you may still choose apples and beer to satisfy thirst and hunger.Don’t get me wrong. Despite my comments I might even agree with parts of your statements.
Comment from BushmanK on 25 April 2016 at 00:07
@MarkusHD,
Okay, I see, you’ve just changed the scale of problem. However, currently I’m not really interested in discussing particular cases since my diary entry covers the general one.
Case of gas stations is different only in aspect of consequences, which could be much more significant in case of lack of information or too general information, therefore, people are more careful about refueling than about purchasing food or clothes. So, it’s just a question of attitude, which is subjective. But in terms of information architecture, these are definitely similar cases, since product groups are as easy to identify as fuel types.
In one of my previous diary entries on topic of “myth of newbie” I already have it explained, that since we don’t have reliable model of mapper’s behavior, all assumptions such as “nobody will use it” and similar ones are nothing but personal assumptions.
Fundamental principle of OSM is “any tags you like”, and I don’t see any problem with introducing better set of tags. All that resistance I’m seeing here in certain comments looks like an expression of stubbornness and desire to protect precious familiar nonsense no matter what, because people trying to argue about things I have never said, presumably existing in their own mind only.
Comment from escada on 25 April 2016 at 08:37
Please add do-it-yourself, trade, electronics, and when I think of it, all the other shop tags to your list of tags that are “useless”.
Unless you list all products (brands, specific items and price) you will never find the shop that you want to go to. O, and don’t forget to list the number of items in stock, otherwise I might make a needless trip over there. And I can only hope it is not the yearly vacation for the shop or whether the shop is closed for unexpected reasons.
No, seriously, I understand what you try to achieve, and I’ll agree that you cannot distinguish a supermarket from a hypermarket (even in Belgium, sorry Sander). We had this discussion just recently on the tagging mailing list. I think that our current scheme works in general, but of course if you want to buy a very specific hifi installation, our scheme does not work.
But this is also true for brands of sodas in supermarkets, some discounters do not sell the well-known brands, only their own. Do we have to solve this in OSM ? (honest question) or can we have a OpenShopCatalogue project where you list all this stuff ?
Comment from escada on 25 April 2016 at 09:11
What do you expect from the OpenStreetMap data ? Should it be able to answer any question about finding locations to buy item X of brand Y below price Z ? Should we be able to find any school where they give course A ? Should we be able to find any neighbourhood with less than X % of people belonging to religion/race/… ? Should we be able to find all hotels that accept dogs, have a swimming pool, speak Dutch, have a buffet that is rated highly by previous visitors and have an offer for 150 EUR for this weekend ?
Or do we find that information in another database and given the name / address (or another method of linking the data) of the item, then use OpenStreetMap to help in the navigation ?
Depending on the answers to those questions, we need a more detailed schema as you have in mind, or we can live with the more general one we have now.
Is our simple key-value paired schema the best place to store all that extra information ? Are our tools suited for the input/verification of this type of data ? I mention verification, because when you want to list hundreds (or even tens) of products per store, I expect we need other tools to verify what previous mappers did.
I’m interested to hear your thoughts on what you expect from OSM data.
Comment from BushmanK on 25 April 2016 at 17:27
@escada,
Some of your arguments are pushing everything to the limit of an absurd.
Here, I’m not trying to push any maximalist ideas of “complete revolution in shop tagging” - OSM is perfectly flexible for having many independent schemes of tagging, satisfying different people. But somehow, majority of commenters just want to exercise in telling me that only one (old and often meaningless) scheme has a right to be used. I also have never said, that we have to break everything down to a single product by attaching shop’s full inventory to its contour. However, there is significant advantage of non-exclusive product category tags (even with quite rough and general categories) over blanket tags such as “supermarket”. At least, it potentially allows any search mechanism to narrow search results down. I have to remind, that in OSM we have huge amount of hard to use, complicated and very detailed schemes, which are in use by mappers, but nobody really knows if they are supported by some end-user software or hardware. So, why so much intolerance towards just another one?
And there is nothing unrealistic in attempts to introduce this kind of scheme. I have several years of experience in product database development both for retain and b2b, with product ranges from sport gear to avionics. So, I know something about reasonable product categories, right?
Regarding of your question - from OSM data I expect simple things: consistency, uniformity, schemes without self-contradictions and mutual contradictions, clear definitions of tags and values, proper usage of abstraction and generalization. By “uniformity” I mean equal definitions regardless of country, language, culture. Nothing of things I expect is new, everything is more or less reproducing core principles of OSM.
One thing I’d want to expect from OSM contributors is some sort of will to evolve for better results. Which is rare thing, if you’re judging by comments here.
Comment from escada on 26 April 2016 at 05:08
@BushmanK
I know my examples might be a bit far stretched, but why do you call them absurd ? Do you think people do not decide on the neighborhood in which they are going to live depending on the other people living there ?
So the questions you want to answer (which questions) with the data are ok, but the examples I made up are absurd ? Can you tell me which part of the questions are absurd ? The questions or the fact that I ask you whether all of the data needed to answer them has to be in OSM ?
In order to define a data model, you need a functional specification. You define already a part of it in your last but one paragraph. But I miss some examples.
So from your reply I still don’t know which questions you want to pose and that can be solved with the new scheme that you have in mind and that cannot be answered with the current schema.
Correct me if I’m wrong, but do you want to be able to answer a question like:
give me all places where I can buy a bike ? (Some supermarkets might sell bikes, some sport shops might sell bikes, and of course shop=bicycle)
Then what about the question: Give my all places where I can buy a race bike of brand X, and where I can get a proper service to adapt it further to my needs ?
Absurd ?
Do you want something like shop=supermarket product:wine=no product:bicycle=yes
and somewhere in the wiki a page where the default products for a shop=supermarket are defined? But are those categories enough, or do we need more detail ?
We were shopping in a supermarket in Germany last year and wanted to buy some white wine. But they only had local wines and Chardonnay wines. None of which we wanted to buy. With just product:wine=yes I don’t gain a lot.
Which detail do we want ? What do we gain by listing all the categories of products, when you do not have details such as brand, tastes etc. ?
When you define a more detailed tagging, I would still like to have the possibility to quickly define something as a supermarket, but also when more detail is added, still look for supermarkets. I’d rather not start looking for something that sells vegetables, meat, fish, clean products, etc.
In case you proposed Education 2.0, that was what I was missing with the driving school thing. I still want to be able to search for driving schools and rather not have to look for education=course, course=driving-license-of-type-X
Comment from escada on 26 April 2016 at 05:09
@BushmanK
I know my examples might be a bit far stretched, but why do you call them absurd ? Do you think people do not decide on the neighborhood in which they are going to live depending on the other people living there ?
So the questions you want to answer (which questions) with the data are ok, but the examples I made up are absurd ? Can you tell me which part of the questions are absurd ? The questions or the fact that I ask you whether all of the data needed to answer them has to be in OSM ?
In order to define a data model, you need a functional specification. You define already a part of it in your last but one paragraph. But I miss some examples.
So from your reply I still don’t know which questions you want to pose and that can be solved with the new scheme that you have in mind and that cannot be answered with the current schema.
Correct me if I’m wrong, but do you want to be able to answer a question like:
give me all places where I can buy a bike ? (Some supermarkets might sell bikes, some sport shops might sell bikes, and of course shop=bicycle)
Then what about the question: Give my all places where I can buy a race bike of brand X, and where I can get a proper service to adapt it further to my needs ?
Absurd ?
Do you want something like shop=supermarket product:wine=no product:bicycle=yes
and somewhere in the wiki a page where the default products for a shop=supermarket are defined? But are those categories enough, or do we need more detail ?
We were shopping in a supermarket in Germany last year and wanted to buy some white wine. But they only had local wines and Chardonnay wines. None of which we wanted to buy. With just product:wine=yes I don’t gain a lot.
Which detail do we want ? What do we gain by listing all the categories of products, when you do not have details such as brand, tastes etc. ?
When you define a more detailed tagging, I would still like to have the possibility to quickly define something as a supermarket, but also when more detail is added, still look for supermarkets. I’d rather not start looking for something that sells vegetables, meat, fish, clean products, etc.
In case you proposed Education 2.0, that was what I was missing with the driving school thing. I still want to be able to search for driving schools and rather not have to look for education=course, course=driving-license-of-type-X
Comment from BushmanK on 26 April 2016 at 16:37
@escada,
Absurd, because you’ve made an extreme assumption and all your argument is about that extreme variant with inventory list, broken down to every single product. That’s not what I’m talking about and not even close to it. Several times I’ve mentioned “product groups”. Groups can be small, large, very different. But being non-exclusive, this scheme should allow to use any scale of abstraction and verbosity. And any level is better than “convenience store”. Obviously, more verbose categories are harder to maintain than less verbose ones. If someone will want to go super-verbose - why not, it’s up to mappers. I just want to discuss a better tool for store tagging (better scheme). Categories have to be uniform and documented, since more general any category is, more chances that certain culture has own way to use it (like I said - in Russia beer is not an alcohol drink according law, while somewhere else it is).
Again, I have to mention “reasonable verbosity”. It is practically impossible to make everyone satisfied by describing product range of particular store in every detail. There are several reasons for it: amount of work, constant changes in inventory, constant changes in supply availability. It even makes certain things non-verifiable, so - no place in OSM, sorry. That’s why I’m insisting on more or less general, but straightforward and clear categories. If you personally would like to extend it with brands or whatever and make more detailed - you are welcome, just don’t make me doing that. What I actually really dislike and oppose is direct linking product ranges and stuff like “supermarket”, “convenience” and so on. This would be too synthetic and unnatural without any gain.
Using Healthcare as an example, you can clearly see the difference between single blanket tag (medical facility in general), tagging down to clinic specialization (which seems to be enough verbose and verifiable) and tagging down to every doctor and every room (which is an overhead for the most cases).
Education 2.0 is not my proposal, and currently I’m abstaining from working on it (however, I acknowledge, that it’s very important), but Healthcare 2.0, at least in its minimalist form, is very good, and I’m always using it as an example, since it is capable to describe specialization of any medical facility. As I said before, good scheme is flexible and extendable. I can’t see, why mapping product ranges can’t work this way.
Comment from escada on 27 April 2016 at 04:57
Maybe brands do not work for supermarkets, but what about bicycle shops, hifi (shops only selling Bose or Marantz or just a couple of end-end brand names), Apple Stores, etc ?
Let me take a look at categories for supermarkets in Belgium, we have the following types of supermarkets
Hypermarkets (a few Carrefour and Cora I think)
The others ? They probably have the same “important” categories, you can buy food, drinks, cleaning products in all of them. You just have less choice within each category. I know that in some supermarkets you can buy newspapers, but that is not the deciding factor to go to a supermarket in most cases.
In Belgium the name of the supermarket and the size are more important than trying to find categories that are sold in one and not in the other. The problem with categories is: not enough: all shops types mentioned above are the same. Too much categories, hard to search and understand why you would pick one above the other. People don’t like too much choice (there are studies about that).
Further, why do I have to specify a shop in my navigation app by clicking many categories, while searching for “supermarket” is much more convenient ? And yes, in a foreign country that might mean that I have to learn that the supermarket does not sell wine (e.g. Switserland, Sweden) and I have to go elsewhere.
Another anecdote: while rice pudding in cans are sold in almost every supermarket in Belgium (including the German Lidl and Aldi), you do not find this product in their German counterparts. Since this is about a product and not a category, your store 2.0 is not going to help.
While I would like to differentiate between hypermarkets and regular supermarkets myself, and I would like to express that the bakery in our town is part deli, I can understand why people do not like to map a supermarket as a shop that sells X number of categories. It doesn’t solve enough problems and is much less convenient than the simple shop=supermarket tag.
Just like you think product level is absurd (you do not see enough problems are solved vs. work in tagging), they think your store 2.0 is absurd ( not enough problems are solved vs. work in tagging). So for me, they react the same to you as you to me.
I have to constantly remind myself to be open to new tagging schemes, so I am open to improvements to the shop tag, but not at the cost of loosing the convenience to tag shop=supermarket. And that’s the same critique I have for the education 2.0. IMHO, You have to keep some clear defaults (whether it is supermarket or driving school or hospital), perhaps even on a national level (see highways), and have additional tags to mark the odds.
A scheme where you have to study a long list of tags to see whether something matches a well-known concept (which might be different from country to country), does not work for me.
Comment from BushmanK on 28 April 2016 at 15:57
@escada,
As you can probably notice, I’m not opposing adding brands, I just don’t want to propose anything for it, since I’m considering it less important (for myself). If you want - okay, no problem. But I’m positively against any attempts of matching wildcard entities (such as “supermarket”) with default sets of goods, since it is quite different from country to country and having separate reference lists of defaults is redundant.
Comment from LLlypuk82 on 17 May 2016 at 11:27
Видя комментарии, родилась такая аналогия:
У ребёнка есть кубики, из которых он может складывать незамысловатые конструкции, он это делает и ему нравится. Потом родители дарят ему на день рождения набор-конструктор со множеством разнообразных деталей, из которых можно собрать большой спектр фигур (тех же кубиков, например) и даже моделей, похожих на объекты окружающего мира. Казалось бы: надо подпрыгивать до потолка от счастья и приниматься за дело. Но, «вдруг», мы наблюдаем иную реакцию: наш «малыш» начинает нервозно хныкать и отпихивать замечательный конструктор, нежно прижимая к груди свои старые кубики (которые у него никто и в мыслях не держал отнимать) и опасливо посматривая на большую коробку со множеством «диковинных деталюшек». А причина проста: ребёнок ещё «не вырос из кубиков» или — иначе — не достаточно развит для того, чтобы оценить все преимущества нового конструктора и примитивность своих старых кубиков.