Moving amenity=atm to z19+ #3372
Conversation
|
I guess the issues you mentioned could have been resolved also with a less drastic cure like reducing the priority of atms in rendering (render them and their labels later).
Probably it once was a conscious decision to render these already at z17 because they are quite important to almost everyone, regardless of personal interest, especially useful for tourists.
For me it is ok to remove them or to ban them to z19, if there is some idea behind this move, just that it doesn’t seem there was a thorough consideration of the topic, and I haven’t seen real arguments why it will be removed from z17 and 18.
|
|
Yes, I just recently wrote an article about size principle in universal map: |
|
sent from a phone
On 31. Aug 2018, at 00:08, kocio-pl ***@***.***> wrote:
Yes, I just recently wrote an article about size principle in universal map:
https://www.openstreetmap.org/user/kocio/diary/44852
as I wrote, the size argument doesn’t relate in a meaningful way to atms, or in other words, I’m not convinced at all this map could work by reducing the whole selection process essentially to depend on the size, and even less on the “typical size” of things.
|
|
My argument is that in the universal map size of the object is the only universal property we may rely on. If this was touristic map style, we would have a simple guiding rule, but it's not. When some objects are usually standalone (in the outdoor for example), they could be rendered earlier, but this is not the case for ATMs. This is also similar case to defensive tower/castle: defensive towers and ATMs might be standalone, but we have no way to know that from tagging, so castle or bank should be rendered earlier as a generalization. |
|
Note the change proposed here was more or less already proposed and rejected back in 2015 in #1884 - along with a lengthy discussion of the typical physical size equals cartographic importance paradigm. |
|
3 years later this problem has grown in a significant way (from ~80 k uses to ~140 k uses) and it keeps growing constantly:
|
|
The increase in numbers is hardly increase in density, it is primarily increase in breadth to areas previously not mapped. As already mentioned back in 2015 the problem you are trying to address is self fabricated by the inflationary addition of new icons. This applies even more today. After all it is definitely not that ATMs have somehow become less relevant over the last years. |
|
There are more of all the objects in the database, not just ATMs, so their context is more dense too. We already made shops less significant and it made some room (because shops are very popular). Yet in the city there are also just the few old (it's not the "new icons" problem), but very popular amenities that make the biggest problem, especially restaurants, fast food and cafes:
|
|
Just to make things clear: of course density is not bigger in the best mapped areas, which were complete 3 years ago (they were the showcase of this problem back then). But density is bigger in all the places that are better mapped now (no matter if they're complete now) and there are more places where this problem exists. |
|
Can the ATM labels be removed too? They just add unnessecary clutter to the map. |
Well, actually, this is a slight side-step, but ATMs are fast becoming dinosaurs in some countries like the Netherlands where I live. We are fast becoming a paperless and cashless economy. I rarely visit an ATM these days, probably only once in two months. Everyone pays with their debitcard here, which is also possible almost anywhere, even in temporary beach clubs at the coast that are only there during the summer season. And it is not just in highly developed countries. I can't recall where I read it, or about exactly what country it was, but I remember reading an article last year about one of the emerging economies in Africa where the leadership was pushing IT, and that people were quite massively turning to digital paying methods there as well. Ah, found it, of all places, it was Rwanda: |
|
sent from a phone
On 31. Aug 2018, at 08:48, mboeringa ***@***.***> wrote:
We are fast becoming a paperless and cashless economy.
+1, there is an agenda to push for the abolition of cash in order to be able to better monitor the flow of money, but there is also a countermovement of privacy focused people.
And there are alternatives kinds of electronic money pushing into the market, like apple pay or google pay.
|
In US you can find ATMs inside bars and restaurants. The only two ATMS in this neighborhood are the two restaurants: https://www.openstreetmap.org/#map=19/37.50370/-122.48550 (I should check if they're mapped).
In .ar, getting money from the 'wrong' ATM network (Link vs Banelco) will get you an extra fee for inter network communications. Yes, that exists. But I think we can live without it. |
|
I just realized that ATMs might be a problem everywhere, not only in the big cities. You can find them also in the rural areas, but from my observation they are rarely in some distance from shops, like few meters (which would be OK in rural area). Usually they are located near the entrance of shop or are embedded in the wall. So they might eclipse shops no matter where they are located, they work effectively like visual "parasites" of shops. |
|
sent from a phone
On 31. Aug 2018, at 12:54, kocio-pl ***@***.***> wrote:
So they might eclipse shops no matter where they are located, they work effectively like visual "parasites" of shops.
it is a deliberate choice of the (former?) style sheet developers to give precedence to atms rather than shops and does not have to do with zoom levels (although you could use different priorities between the same things on different zoom levels).
|
|
Do you think that they knew how the map with this style will look like 6+ years from the moment of this choice? I don't. There was much less objects then and we had white list of shops - only few of them had icons and most of them were not even shown as a dot. And the icon for ATM was there. In 2009 I was hoping that the map will be complete - so I wanted to add all the house numbers in my neighborhood, because there were streets and they even all had names! This was all I hoped for then. I don't know when exactly Mapnik XML style has been designed (OSM Carto was just technical reimplementation initially), but it might make sense then. But we're in a very different world now, welcome in OSM 2018... |
|
Example of the area where the ATMs almost alone clutter the map (it's far from complete): |
|
example where restaurants and cafes clutter the map at the same zoom level: https://www.openstreetmap.org/#map=17/48.85325/2.33847
or
https://www.openstreetmap.org/#map=17/41.89943/12.47216
you can find situations like these for some features, sometimes it just shows the particularity of the area and is not necessarily a problem.
sent from a phone
… On 5. Sep 2018, at 02:23, kocio-pl ***@***.***> wrote:
Example of the area where the ATMs almost alone clutter the map (it's far from complete):
|
|
Re: "it is a deliberate choice of the (former?) style sheet developers to give precedence to atms rather than shops and does not have to do with zoom levels" I agree with the previous style choice to show ATMs at z17, and oppose this change. ATMs are very important in most non-European countries, where cash transactions are still the norm, and they are not as common. I've often needed to travel a kilometer to find the closest one, even in a town. Also note that the Humanitarian style has ATMs rendered prominently, which I mention as a proxy for their importance in lesser-developed countries. Instead, I would recommend deprioritizing an ATM near a bank, so that the bank shows instead, and deprioritizing the name lable for ATMs. That should solve both of your sample problems without removing the ATMs from the map entirely at z17 and z18. |
|
2018-09-10 15:56 GMT+02:00 jeisenbe <notifications@github.com>:
Instead, I would recommend deprioritizing an ATM near a bank, so that the
bank shows instead, and deprioritizing the name lable for ATMs. That should
solve both of your sample problems without removing the ATMs from the map
entirely at z17 and z18.
+1, indeed I had recommended this as well in the first reply, but it has
been ignored ever since.
|
|
Sorry for not answering this suggestion, I wanted to answer general issues first and then I needed to think a bit. Humanitarian style can afford to make arbitrary choices about importance, because it's much simpler than OSM Carto. They don't try to follow the database changes, because their scope is very limited, unlike ours, and they are expected to follow undeveloped areas more closely than we, so this is not surprising to me. Also searching is not what general style is made for. This is where mobile apps or web search field come into play. I believe size rule is good way to go, because it will make map more coherent and easier to declutter at the same time. But we could test your idea too. Is anybody ready to try the code with less priority of ATMs? Just please remember that we don't have a way to express "near a bank" property, it has to be "near anything". I would not like ATM to eclipse shop or amenity like post office or gym too. |
|
Re: " I would not like ATM to eclipse shop or amenity like post office or gym too." I'm not sure that ATMs are less important cartographically than a generic shop, even if they are smaller (though some ATM halls have a dozen different ATMs, and are the size of a kiosk) As you mentioned, "searching is not what general style is made for"; it would be easier to find a particular shop by searching for it's name, or by searching for it's category. Other services that provide reviews and more details about shops are also more useful. But while seeing a shop icon on the map is not very informative (especially without a name), an ATM icon alone is enough to know "I can get money out of my bank there". In your first picture you showed a bank icon obscured by an ATM icon, on a corner with 5 banks:
In this situation, the ATM icon is more useful than the bank icon, even though the bank is larger. What good is it to know that there are 5 banks, if we don't see the names. I can't get any services at a bank where I have no account, so I'll have to zoom in or just search for the right bank. But most ATMs will work with cards from other banks and other countries, so just seeing the location on a general map is useful. |
|
You see? You make a lot of "hand waving" about ATMs and you're still "not sure" what is cartographically more or less important. Of course - you can't be sure, because you have no rule even for shop/bank//ATM trio. I use simple size rule for them: I have seen ATMs being part of the bank or shop, yet I have never seen a bank or shop being a part of ATM, so typical size is clear - ATMs are smaller and therefore should appear later than both of them. And this rule works as effective also for castle/defensive tower, museum/toilets and many other cases we will test in the future. |
|
2018-09-11 12:10 GMT+02:00 kocio-pl <notifications@github.com>:
You make a lot of "hand waving" about ATMs and you're still "not sure"
what is cartographically more or less important. Of course - you can't be
sure, because you have no rule even for shop/bank//ATM trio.
I use simple size rule for them: I have seen ATMs being part of the bank
or shop, yet I have never seen a bank o shop being a part of ATM, so
typical size is clear - ATMs are smaller and therefore should appear later
than both of them. And this rule works as effective also for
castle/defensive tower, museum/toilets and many other cases we will test in
the future.
to me, basing rendering priority only on "size" is arbitrary. Yes, it is a
concept, but it is not suited for solving the complex decision we have to
make, in a satisfying way.
I could come up with another arbitrary criterion, the "time rule". If
something is available for more time it is more important to show
(hypothesis). A christmas market is there only for few weeks in the years,
so we prefer to show the situation on the rest of the time. Banks are only
open a few hours on weekdays vs an atm which is typically available 24/7.
See, atms should dominate banks.
Another arbitrary criterion: "how big is the group potentially interested
in this?" Atms are useful for anybody that has a bank card of one of the
circuits it offers. That is orders of magnitude more people than those who
are interested in a bank (must be client of the bank and maybe even of this
particular branch for most operations).
Another arbitrary criterion: significance. How significant / important is
something compared to other things, how many people know it. A newsagent
will rarely be known outside its surroundings. A major museum can be world
famous and will be interesting to many tourists and will usually serve as a
reference point (many people know it). You might not want to go to the
museum every day, but it is useful to see it labeled for orientation.
There are probably more criteria, and to me what we have to do is balance
these and decide on a case by case base how to deal with it, how to weight
the criteria.
|
|
Nice try. 😄 Of course, any rule is arbitrary and it might be good to look at some alternatives - I just use what my experience with this style shows is working in pretty universal way. However how easy is to test all these rules? Is it visible somehow? For example - significance: it's so hard to find good test, that any tagging scheme using it is dropped in the Tagging discussion, even if multiple people wanted to establish some. The only one quite popular tagging of similar type that I'm aware of is road smoothness. Not too objective, nor universal, sometimes just unknown. How big group is interested in - how would you measure this? And what if this group already have special style available? In which part of the world is this group big and where it will be small? Not too universal for me and also unknown in many cases. Example: on cycling map layer there are pubs shown in the prominent way - in Poland you can't drink alcohol when riding bike (not even radler - sorry...), so even if we know exactly the group and style is very specific, it can be different in different countries. Time rule - we use it already for example for water, but most of the time it's a tagging rule: most volatile objects are simply not allowed. We could of course parse time tags, but this is object related, not for the whole class. And try to see this difference - also not easy to apply. All the permanent objects should be rendered from the same zoom level? How many different classes would we get this way (2 perhaps - permanent and seasonal)? Size rule is not my invention nor anything new, I just try to stick to it more because we have too many hard to compare class of objects. We already show cities later than countries and neighborhood later than suburbs (...so it was piece of cake for me to add quarters to this scheme - just slip it between them!). We even show supermarkets earlier than convenience shops - luckily there are different tags for them. This is also not the only rule - for example roads make a hierarchy so it's quite easy to depict it (however note that it's still somewhat related to their size). We have also more freedom with the outdoor features for example - there is not too much to show, so we can show them earlier, exaggerate them a bit. But this is when things are easy (yet even for outdoor objects relative size might be useful - defensive tower/castle case is typical there). When it's not that easy - like in dense tagged areas - we need some simple and universal rule. |
|
sent from a phone
On 11. Sep 2018, at 13:11, kocio-pl ***@***.***> wrote:
Size rule is not my invention not anything new, I just try to stick to it more because we have too many hard to compare class of objects.
to be clear, size is an important factor for deciding what to show when, but it is not the only one. Yes, one is bigger than the other, but within the range of object sizes (and common zoomlevels) banks and atms are similar (things smaller than a building).
When you write „stick to it more“ I understand „give more importance to size than to anything else“.
Should an office building or a parking have priority before a post office or a church, because they have a larger footprint? Would we prefer a forest name over a city label?
|
You're right and I've just said it.
You took too big granularity when comparing them to buildings (that is the downside I mentioned with time based rule). We could also say that they all are the same compared to country - yes, they are but it's too big generalization. But it helps a lot when we try to clean z17 a bit, which is very dense in WMAs - at this level they have different size and this is simple to tell which are bigger and which are smaller.
I mean "let's not omit it from our set of rules" (it's not there now) and "let's use it as a universal rule to avoid subjective importance wars".
Parking may have very different size, but we already deal with it properly - "P" sign is shown only when it's big enough. The same with the forest. They are automatically prioritizing themselves. So are the city labels. It's better than typical class size, it's per object, which is closer to the truth. |
|
if I counted right, there was one comment (lakedistrictOSM) in favor of removing the atms from z17, some general remarks about atms and at least 3 against it (imagico, jeisenbe, dieterdreist) , plus the rejection by the former team in #1884. You are still insisting in deploying it? |
|
Yes. Up to this moment no good reasons have been given against this solution. I responded to all the critical comments, I guess. |
An alternative rule would be to prioritize objects based on how often they are used. This is of course hard to count exactly, but we could make an estimation. This would increase the prominence of ATMs, because ATMs are used quite often even though they are small. How would you feel about such rule? |
|
Am Mo., 1. Okt. 2018 um 10:40 Uhr schrieb Matthijs Melissen <
notifications@github.com>:
I believe size rule is good way to go
An alternative rule would be to prioritize objects based on how often they
are used. This is of course hard to count exactly, but we could make an
estimation. This would increase the prominence of ATMs, because ATMs are
used quite often even though they are small. How would you feel about such
rule?
similarly, it could be an indicator to look at how big is the potential
userbase (not necessarily in absolute terms, which is hard to do, but also
"subjectively" relative, i.e. we guesstimate it). ATMs can be used 24/7
(typically) and by everybody (who has a card of one of the
circuits/networks that the atm supports), while a bank has typically very
limited opening hours and is useful only for people who hold an acocunt at
this specific branch.
We could also look at how useful we estimate showing the thing on a map.
Usually, maps are more important for tourists than for locals, because
locals know much more often without a map where the things are that they
need, especially if it is the branch they have their bank account at. Of
course, things on a map also serve for general orientation, i.e. it could
help you to see something on a map even if you don't want to go there, and
this would be a point in favor of prioritizing banks, because they will be
more visibly prominent than atms when you stand in the street.
The way the "size rule" is presented currently seems that it also useful
for comparing apples with pears. Maybe it is not intentional but zhe way it
is written, to me it suggests we should give priority to a slurry tank over
an atm, because the former is bigger. This is just nonsense.
|
|
Another consideration is that the current prominence rules are not size
based. For example, “major buildings” are shown mire prominently- like
train stations, churches, mosques, synagogues, airports etc
This is reasonable because these places are community landmarks and are
also public places that people are likely to look for. These
characteristics do not relate to size.
We are also currently considering deemphasizing minor buildings like
garages, based on their use and importance, not due to their size. This is
reasonable.
My folks live in the county, where the sheds and barns are often bigger
than the house, but certainly the house is more important. When there is a
forest fire, they count the outbuildings and houses burned separately even
though the outbuildings may be bigger
…On Mon, Oct 1, 2018 at 5:40 PM Matthijs Melissen ***@***.***> wrote:
I believe size rule is good way to go
An alternative rule would be to prioritize objects based on how often they
are used. This is of course hard to count exactly, but we could make an
estimation. This would increase the prominence of ATMs, because ATMs are
used quite often even though they are small. How would you feel about such
rule?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#3372 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AoxshJRIssmcS5nR9SWNG7P4yOh7j6-Qks5ugdUBgaJpZM4WURWp>
.
|
|
First of all thanks for all the comments. Sorry for not responding to all the problems this time - there are too many things and I will try to reply step by step:
Sure. There is not and I have said already that I don't propose to remove all the others, so it might be a straw man argument. I just think this is the best rule for universal style.
For such a rich and still growing database I believe deciding case by case (=effectively no rules) will not work good. All the downsides of "vote everything" idea (including rising lack of coherency between multiple elements) apply.
Well, I don't propose anything new. Some of them are (we never show cities before countries, for example), some of them aren't. Whenever we have not too many elements, we can exaggerate (like showing capitals early or nearly all outdoor-specific objects) and we can modify colors like you said (or other attributes, like showing dots for shops initially). We will still have a lot of room for choosing and voting.
"More important" - OK, but again: for whom and for what purpose? There is no universal meaning of "importance", free from the context. You just gave the reason for only one purpose. But it's very specific and sounds to me like HOT task, which has it's own very specific map style, and not for a general map we try to produce. We could modify shade or pattern for less important buildings - see #2532 (PR for which was rejected). But not showing big things before smaller ones is unpleasantly surprising idea for me - for the overview of this area hiding big, visible things does not work.
My English is too weak here, and so is my farmland knowledge, I don't know what is slurry tank - is it https://en.wikipedia.org/wiki/Slurry_pit ? If it's bigger, it should be rendered earlier. Still I can't imagine how does it look like... However I have very similar (also stinking...) analogy - we already show landfills from z10, which is much earlier than ATMs. Is there any problem for you? If yes, what do you want to do with it? |
|
Landfills / waste dumps are useful because you can take your trash there,
in most places I've lived at least, even if most people hire someone else
to do it.
They often serve a whole rural district, town or city and are well-known
landmarks due to their function, not just due to their size.
So I agree with showing landfills/dumps earlier than ATMs, and I don't
think we have to resort to size to explain this decision:
A landfill/dump is usually a regionally significant facility, while ATMs
are significant at the village or neighborhood level, in most places.
…On Tue, Oct 2, 2018 at 9:02 AM kocio-pl ***@***.***> wrote:
First of all thanks for all the comments. Sorry for not responding to all
the problems this time - there are too many things and I will try to reply
step by step:
IMHO there is no single rule that lets us decide everything,
Sure. There is not and I have said already that I don't propose to remove
all the others, so it might be a straw man argument. I just think this is
the best rule for universal style.
every issue has to be judged on an individual basis
For such a rich and still growing database I believe deciding case by case
(=effectively no rules) will not work good. All the downsides of "vote
everything" idea (including rising lack of coherency between multiple
elements) apply.
Another consideration is that the current prominence rules are not size
based.
Well, I don't propose anything new. Some of them are (we never show cities
before countries, for example), some of them aren't. Whenever we have not
too many elements, we can exaggerate (like showing capitals early or nearly
all outdoor-specific objects) and we can modify colors like you said (or
other attributes, like showing dots for shops initially). We will still
have a lot of room for choosing and voting.
My folks live in the county, where the sheds and barns are often bigger
than the house, but certainly the house is more important. When there is a
forest fire, they count the outbuildings and houses burned separately even
though the outbuildings may be bigger
"More important" - OK, but but for whom and for what purpose? There is no
universal meaning of "importance", free from the context. You just gave the
reason for only one purpose. But it's very specific and sounds to me like
HOT task, which has it's own very specific map style, and not for a general
map we try to produce. We could modify shade or pattern for less important
buildings - see #2532
<#2532> (PR for
which was rejected). But not showing big things before smaller ones is
unpleasantly surprising idea for me - for the overview of this area hiding
big, visible things does not work.
The way the "size rule" is presented currently seems that it also useful
for comparing apples with pears. Maybe it is not intentional but zhe way it
is written, to me it suggests we should give priority to a slurry tank over
an atm, because the former is bigger.
My English is too weak here, and so is my farmland knowledge, I don't know
what is slurry tank - is it https://en.wikipedia.org/wiki/Slurry_pit ?
If it's bigger, it should be rendered earlier. Still I can't imagine how
does it look like... However I have very similar (also stinking...) analogy
- we already show landfills from z10, which is much earlier than ATMs. Is
there any problem for you? If yes, what do you want to do with it?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#3372 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AoxshE4zpT86l8D6UfITDs7Nu--Qh47oks5ugqz5gaJpZM4WURWp>
.
|
They wouldn't be landmarks if they were just few meters big. It would be (probably) the case of slurry tank than. AFAIK you can't even get into big landfill, at least in Poland, so for me personally it's less important than ATM. But for the local community it's the other way around. So importance is hard to get and compare and we're serving map style for all uses. What if big landfill is operated by single entity, which for example imports trash from other countries and doesn't serve community at all? I know such places exist. Would you not show them early enough just because they are not useful? |
|
In your example, there is a landuse=landfill with access=private, if I
understand correctly.
Usually we render access=private with a lighter color icon (eg private
parking) or dashed instead of solid line (private roads)
But I think we usually render access=private at the same zoom level as
public, so far? I don't know the stylesheet code very well.
Perhaps a private landfill, or private parking lot, or private street,
could be rendered one zoom level later;
Eg zoom11 for private landfills, zoom10 for public landfills, if this is
actually tagged in the database.
This would actually be quite a good idea for airports, and perhaps for
parking lots too?
…-Joseph
On Tue, Oct 2, 2018 at 10:48 AM kocio-pl ***@***.***> wrote:
They often serve a whole rural district, town or city and are well-known
landmarks due to their function, not just due to their size.
They wouldn't be landmarks if they were just few meters big. It would be
(probably) the case of slurry tank than.
AFAIK you can't even get into big landfill, at least in Poland, so for me
personally it's less important than ATM. But for the local community it's
the other way around. So importance is hard to get and compare and we're
serving map style for all uses.
What if big landfill is operated by single entity, which for example
imports trash from other countries and doesn't serve community at all? I
know such places exist. Would you not show them early enough just because
they are not useful?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#3372 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AoxshOpbHLxn1otrY6ozQh53J7GKGmgPks5ugsXRgaJpZM4WURWp>
.
|
|
I don't like to go into details of rendering private objects now (it's a problem for few separate issue tickets IMO), especially since I still haven't responded to some other questions. My point is that operator, access, relative usefulness or ugliness doesn't matter - if something is big enough, it will be always visible much earlier than ATMs. |
|
Am Di., 2. Okt. 2018 um 02:01 Uhr schrieb kocio-pl <notifications@github.com
:
The way the "size rule" is presented currently seems that it also useful
for comparing apples with pears. Maybe it is not intentional but zhe way it
is written, to me it suggests we should give priority to a slurry tank over
an atm, because the former is bigger.
My English is too weak here, and so is my farmland knowledge, I don't know
what is slurry tank - is it https://en.wikipedia.org/wiki/Slurry_pit ?
exactly, as the article states, "also known as slurry tank". These can come
in different shapes and structures, but the function is the same (storing
animal pee). Here's an example how they may look like:
https://de.wikipedia.org/wiki/G%C3%BCllegrube#/media/File:Don%27t_jump;_Haddon_House_Farm,_Bakewell_-_geograph.org.uk_-_1264500.jpg
or
https://www.picclickimg.com/d/l400/pict/202165400029_/Ehemaliger-Jauchetank-G%C3%BClletank-jetzt-Wassertank-2200Ltr-feuerverzinkt.jpg
these are objects, like atms, and unlike landfills or other landuse.
|
Thanks for another example of rule that could be useful, I'm glad that we're testing alternatives. This one looks quite promising, since it gets away from the "importance wars" and deals with numbers, but is also not universal - it's not only harder to measure than size and can change over time, it's also very limited in scope. We might estimate the usage of ATMs, banks or shops at given moment, but there's no real equivalent for most memorials, towers or artworks for example. Niagara Falls might have some "userbase", but most other waterfalls just don't have any.
I think this is another nice helper rule, so it's good to have it in our toolset. It's more universal than just voting, but size still looks to me as the most universal rule for the map visualization.
You just gave the example of local authorities needs, but there can be much more uses you don't even think of. It's not our job to dictate which usage is more important. I understand why it's so hard to not look from the specific points of view, but one ends up with switching them and lacking tools to compare them. As OSM is still growing in all the directions, we have more different consumers and I think it's time to make "universal" style really generic. We used to be "street map" - but it was just the beginning. We even had the UK roads style for many years (until 2015), because it was London-based project, but IMO as we get global it's getting more important to look for the most universal rules for default map style. |
|
There were some arguments about the past - that we should probably trust in the original design reasons (even if we don't know them), that density is not that big problem, because this is only about tagging new places, that it's all the fault of adding new icons etc. Here is simple comparison of 2012 and 2018 data in the center of London with current style. Original ideas, which come from some time before 2012 (OSM Carto did not make a big departure from them), work reasonably good with smaller set of data, but look horrible with more of them (which was hard to predict, of course - it's also nobody's fault that OSM grows so fast). There is much bigger POI density even if London was probably mapped the best from the beginning, so this is not only about mapping new places. And there's only few new icons, much of the clutter here comes from adding more gastronomy and shops POIs. Small amenities are part of a problem now, even if we'll make orange dots on z17 (look at bigger streets): London center in 2012 (from the last CC-BY-SA licensed planet file)
London center currently
London center with proposed gastronomy dots
|
|
It looks like the current data includes many more offices, shops and
restaurants. The house/building address numbers also add to the crowding in
2018. The eating places are most noticeable since the shops and offices are
just dots.
But I only see 3 ATMs on each version of the map. They don’t seem to be the
problem?
It would be great if someone could make a code that would render the map
differently in crowded urban areas vs sparse rural areas. Maybe it could
also fix the difference in zoom levels vs map projection at 60 degrees
north vs the equator.
…-Joseph
On Wed, Oct 3, 2018 at 12:04 AM kocio-pl ***@***.***> wrote:
There were some arguments about the past - that we should probably trust
in the original design reasons (even if we don't know them), that density
is not that big problem, because this is only about tagging new places,
that it's all the fault of adding new icons etc.
Here is simple comparison of 2012 and 2018 data in the center of London
with current style. Original ideas, which come from some time before 2012
(OSM Carto did not made a big departure from them), work reasonably good
with smaller set of data, but look horrible with more of them (which was
hard to predict, of course - it's also nobody's fault that OSM grows so
fast). There is much bigger POI density even if London was probably mapped
the best from the beginning, so this is not only about mapping new places.
And there's only few new icons, much of the clutter here comes from adding
more gastronomy and shops POIs. Small amenities are part of a problem now,
even if we'll make orange dots on z17:
London center in 2012 (from the last CC-BY-SA licensed planet file)
[image: oecwrnln]
<https://user-images.githubusercontent.com/5439713/46356371-e839d900-c662-11e8-95fc-3b8778a6a5b9.png>
London center currently
[image: nn903zm5]
<https://user-images.githubusercontent.com/5439713/46356368-e5d77f00-c662-11e8-89e4-c6a3b38cb935.png>
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#3372 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AoxshKCRuN7496AumrkZVRDI4fdmmYy0ks5ug4CHgaJpZM4WURWp>
.
|
|
All small amenities are part of the problem. I see 9 ATMs there and 4 of them are eclipsing roads. There are also phones and post boxes also eclipsing roads. This is not obvious until gastronomy is dotted, but roads have already some related features (like names, arrows, traffic lights, some bus stops or bike rentals) and phones or ATMs do not belong there. |
Yes, this looks like a very important problem, but still we have no clue how to fix it: #1957. |
|
Oh, now I see. I must have only Seen the top half of each image due to a
slow internet connection.
Is there some sort of code like:
“Don’t render x if there is are more than 2 other x within 50 pixels
distance”?
Joseph
…On Wed, Oct 3, 2018 at 9:15 PM kocio-pl ***@***.***> wrote:
All small amenities are part of the problem. I see 9 ATMs there and 4 of
them are eclipsing roads. There are also phones and post boxes also
eclipsing roads. This is not obvious until gastronomy is dotted, but roads
have already some related features (like names, arrows, traffic lights,
some bus stops or bike rentals) and phones or ATMs do not belong there.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#3372 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AoxshE8aq2aWkeSFT_d2b15KUM8Ca2Akks5uhKpzgaJpZM4WURWp>
.
|
|
@jeisenbe, there is a discussion about it in issue #1957. I also brought up a similar idea in the gastronomy color change issue based on population density. Which I was reading an article about and could be totally doable. Both solutions could introduce their own issues though, like slowed down rendering speed due to the added code complexity or it still being interpretive. Plus, it would have to be thoroughly tested. Id imagine by the time the code got written and it was tested properly there would be vector tiles implemented on the main site that would allow for a better layers selection system, which would solve most of these types of issues. Like there is an interactivity function in the project file that would allow for informational popups on the map (at least I think that's what it does), but it slows things down to almost not be worth it from my understanding. Which is why the main CartoCSS maps don't use it, I think. Not to say something like that or your idea isn't worth trying to implement, but there would be technological hurdles etc and it might not be worth the time if something better is coming along soonish. Its kind of a similar issue with showing the surface of roads. There's one map that I know of that shows road surfaces and it renders way slowly compared to this one. Plus, it only shows it in a very small area. That's all just my opinion though. |
|
Thanks all for the discussion and showing different points of view, it was good to inspect them to better understand the issue, which is more general than only ATMs. It looks to me now that any systematic approach to cleaning rendering needs to regard size differences to achieve really universal and generic OSM map style. Other rules (like private/public differentiation) can be still useful, but they are secondary tools. I also see the need to apply such cleaning, because database growth seriously changes the look of the map, even if the style hasn't changed for years in some aspects. I think ignoring such a big thing makes the resulting map gradually worse over the years. |
For the record, I do not agree with this statement. But from a community perspective, I think it's important that decisions are made by the people who actually do the work, so I'm happy you went through with this proposal. |
|
I am also kind of glad about this outcome because it marks a clear end to any remainders of the cartographic consensus that formerly existed in the development of this style. In particular the statements of @kocio-pl also demonstrate more clearly than before the rejection of the evidence and argument based discourse - right down to aspects of simple logic and the nature of the geographic reality of this world.
As i have emphatically pointed out before this is all a matter of what kind of development community you attract. There are a lot of potentially stable development community scenarios that - looked at from the inside - work harmonically together. The vast majority of them however would produce a map style that would ostentatiously be unsuitable as the public face of OpenStreetMap with its current goals and values. To what extent this is already the case for OSM-Carto everyone needs to decide for themselves. |
|
@imagico, I think your blowing it out of proportion. I lurked around here for several years before I contributed. There was way less consensus or discussion about things before @kocio-pl came along. His openness to trying things was one of the reasons I decided to finally contribute. There were many issues before he arrived that where shot down by a single person for nothing despite multiple people wanting to see the issues tested. There has also been many issues that he reopened because people requested it. He also went against his better judgement a few times and implemented things he didn't necessarily agree with, but that others wanted. I never saw that happen before. There's also still plenty of lively debate about issues. If anything there is more debate then there use to be, because contributors are less worried about getting coldly blown off or dismissed off hand simply because they don't have years of experience at this like you guys do. Which happened a lot before. Ultimately, its just one PR and map goals can change. Why shouldn't they? The database its based on, along with the needs of the mappers, constantly change. Maybe you can keep the goals of the map consistent forever, but if it doesn't evolve then no one will contribute and you'll have a dead project. At least the map will look pretty though right? To you at least. To me, an imperfect map made by a vibrant community is better than one that will be outdated in a few years and have no one left to update it because, they were all pushed away by rigidness. Right before @kocio-pl took control, it seemed like that was direction it was going in. The way I see it, he brought it back from the brink of obsolesces. So, rendering an ATM at z19 isn't the end of the world and if anything him merging it is a good thing, because its a sign of much needed change in attitude overall that doesn't involve fundamentalism. That's just my opinion though. |
This is sad for me to see such statement. I find it so unfair that it's hard to believe you wrote it. I took a lot of time and effort to look at multiple arguments, think about them and respond with my arguments and showing proofs. Everyone is free to have their own position and not like them. But they are here - you even took a small chance to respond and made your arguments against. And now you sound like there was just vague comments, probably. I took discussion very seriously. Not agreeing with arguments is one thing, rejecting their existence supports lack of discussion at all (why try, when I can simply wait and dismiss everything at the end?) and this is dangerous. What I think is wrong, is discussion without the outcome, as we had before. But demoting active discussion (in any way) is also wrong. I believe in active discussing, deciding and then implementing it. Each step requires activity.
Nice to have it, but it's not something that just magically happens. What is your proposition to reach consensus then?
I see. What is your position and suggestion then? |
Sounds nice for me, but this is all just theory. We could also have vector style transition and this idea is not our problem - just the lack of such code. So is about "potentially stable development scenarios" - it's also nice, but we don't have other scenarios than the current one. You can try to attract this kind of harmonic community if you feel like - it's up to you and I would love to see that in reality. |
Changes proposed in this pull request:
ATMs are small facilities which make clutter on the map on z17 and z18 when located near some other POIs, which is common case in cities. ATM is smaller than bank, where it can be usually placed, so it should be rendered later, and since their size is smaller than all the shops, I think they should be rendered from z19.
Just two examples out of many:
Bank is eclipsed by ATM at z17:
Before
After
ATM eclipsing label of a bigger amenity at z18:
Before
After
