> Over the course of a year, Google quietly turned its map inside-out – transforming it from a road map into a place map.
I've long been amazed how we somehow transitioned during the early 20th century from a mental model of roads and paths running through locations to places (house lots, etc) being the spaces between the roads. It's a natural thing to happen, but one of those invisible flips that happens on a timescale longer than a human lifetime.
But this anticipates the opposite: if you can stop worrying about how to get somewhere (because you don't have to drive or plan much -- self-driving or Lyft-style services can take care of the route planning) you can focus on the destination.
We see this phenomenon in subway maps which are famously schematic and not geographical.
(BTW the transformation is visible in literature, which is how I noticed it. The sense of geography in, say, Jane Austin is completely alien to today).
> BTW the transformation is visible in literature, which is how I noticed it. The sense of geography in, say, Jane Austin is completely alien to today
Now I'm curious, don't remember such a difference, but then it's been a long time since I read Jane Austen. Don't suppose you could put your finger on exactly how she handles geography that a writer like Stephen King or China Mieville doesn't?
At a guess, previous generations would talk in terms of distance-to-destination, whereas we talk in terms of time-to-destination. They would think of a highway and a neighborhood as having the same "speed limit" (foot or horse-and-cart). We, with cars, think of highways as having a much lower time cost (or higher time efficiency) than neighborhood roads.
> At a guess, previous generations would talk in terms of distance-to-destination, whereas we talk in terms of time-to-destination
I guess this must be a 20th-century phenomenon limited to industrialized nations. My grandparents' generation, who lived in the Carpathian Mountains and relied on cows and wooden-made carts for their method of transportation (when not going on foot), definitely only used time-to-destination. I've never ever heard my grandma' say anything about kilometers, meters, or the like.
In developing countries this shift has already happened ('distance' to 'time') due to things like traffic jams, inconsistent infrastructure (some routes are longer but faster due to lesser traffic signals or better maintained etc.)
Maybe it would not be wrong to assert that this shift is more in force in developing countries than other places. This could be due to unpredictableness. The routes that are optimal now - takes less time - can change in matter of minutes/hours (due to procession/rallies, rains, accidents, school & office timings, power failure - leading to stoppage of traffic signals, etc.)
Google maps always give multiple routes to a destination and they differ in terms of time to destination (distance to destination is seen as secondary information on map interface).
> In developing countries this shift has already happened ('distance' to 'time') due to things like traffic jams, inconsistent infrastructure (some routes are longer but faster due to lesser traffic signals or better maintained etc.)
Then I guess Germany has not yet developed. But when talking about places people always specify the distance. Berlin is 584 km away from Munich, not 5 hours and 15 minutes. Talking about time to destination is something I have exclusively seen Americans do. Maybe that's because in the US distance and time are basically the same, since everybody takes the car and drives at the speed limit. But in Germany there is no general speed limit, and you might also decide to take the train or airplane to Berlin (and may people do) at which point talking about time would be confusing at best.
Before trains were introduced in Germany, distance was often given in time it takes to walk (Wegstunde, lit. "way hour"). Early trains even advertised how many hours you can cover in mere minutes. People then probably realised how silly that sounds when more and more train lines where built and dropped that in favour of distance.
I disagree. Places that I have visited serve as excellent landmarks. I wouldn't want them to be the only thing on the map but having them dusted around makes it very easy for me to know where new places are.
Google makes money from place entries (booking ads, place ads) but not from geographic place names, this alone explains the move. Think of the places selected as ads at the top of your search results.
I was thinking or maybe dreaming about this the other day. There's a subtle shift through tech that blurs the original goal. You wanted to enjoy some place a little easier, so roads make sense. Then it grows and grows and now the beautiful places are surrounded by roads and annoyed crowds and there's nothing to enjoy.
It's also interesting to see how maps evolve (or fail to evolve) for places that are not San Francisco.
At the moment, Apple Maps seems to have a more thought-through design for public transit than Google Maps. Which is to say, transit view in Apple Maps is either visually clean and uncluttered, or completely nonexistent, depending on whether they got around to adding your city. Clearly a lot of by-hand design work goes into it, which isn't a very scalable approach.
On the other hand, transit data in Google sometimes appears to have been munged with no human intervention and never received even a cursory check by a graphic designer. For example, turning on Transit view in downtown Toronto will show a mess of ungodly rainbow spaghetti which is meant to represent the streetcar system. There are lines on non-revenue tracks where no streetcars actually run, lines on streets that don't have streetcar tracks, random artefact lines that appear and then vanish two blocks later, and lines drawn diagonally through the middle of High Park where there is no street at all. Somehow, the data behind this spaghetti is diligently updated year-after-year (e.g. the new Cherry streetcar was added in 2016) without anyone involved in the process noticing that the results are hideously garbled.
It also took them about a decade to realize that the SkyTrain in Vancouver is a rapid transit system.
I really like the approach described in these links. Targeted algorithms solve specific problems (e.g. ILP to decide ordering of adjacent lines) instead of someone trying to put together a Grand Unified Machine Learning Algorithm that ends up having to be fudged manually after it gets 99 cases right and makes an egregious mistake in the 100th one.
The more general problem Transit seems to be trying to solve (that no one else is) is having a map that can be kept up-to-date with transit alerts and diversions with minimal hassle. For example, it looks like their launch-time picture of Toronto's streetcars (in the first link) was generated when there was trackwork at the College and Bathurst intersection and the corresponding streetcar lines were being diverted around it.
The second link explains rather well the reasons behind some of the problems evident in Google's Toronto map. For example, the unsnapping issue for adjacent lines occurs a lot because streetcar lines are heavily interlined.
they really are a great company. when there was an outage in san francisco's schedule reporting system, they had users tag themselves as being on specific buses, and tried to synthesise their own real-time schedule from that.
It drew SkyTrain as an ordinary railway (part of the map layer), which is probably treated the same for transit directions, but nevertheless an odd choice visually, since it gets lost in the tangle of freight railways which are drawn identically and which SkyTrain frequently runs alongside.
One thing I think Google has going for is it "Guides" program. Where user get "points" for correcting mistakes, adding new places, and publishing pictures. I have been in it for a few months and it feels good to contribute.
I was uploading images to Google Maps before that anyway, and if I recall signed up for it, it's just a point meter.
Google then sends you a summary of how your contributions were doing. Recently they send me some E-Mail to the effect of "Awesome! Your image on <business name> has been viewed 100 thousand times!" or something like that.
It entirely turned me off the whole thing and I haven't contributed since. I thought "why am I contributing to this ad-powered product that they're monetizing for millions, for free?".
It's kind of hilarious really. If you look at their landing page the benefits you can gain include things like "...and you can also moderate our forums, for us, for free!": https://www.google.com/local/guides/benefits/
I believe the idea is the motivate you to contribute further by proving that your edit made a difference. How does your edit being viewed 100 thousand times makes you less interested in contributing? Would you rather contribute to a product nobody uses?
> I thought "why am I contributing to this ad-powered product that they're monetizing for millions, for free?"
You already knew Google Maps was monetized by ads, before making an edit.
I don't think it's fair to see this as free labor for Google. Your edits to Google Maps could be related to your own business. Or you might be helping your favorite local business to get more visibility and customers.
Perhaps it can also be seen as a contribution to your neighborhood.
Huh, there actually used to be a real benefit - free Drive storage upgrades once you hit a certain tier. Guess things got popular enough they ditched it.
(disclaimer: work on maps, nothing to do with guides tho)
Also they started pushing 'rate where you were' notifications way too much. Initially notifying when in maps app moved to ever increasing out of app notifications. That's when I stopped and turned if all maps notifications.
I might not be motivated by that now (I'm 35) but when I was in my 20s I found it incredibly exciting just to see something I made get used by other people. I think it's a natural desire.
Since then I've gotten a little obsessive about specific goals, so I'm less compelled by it, but I don't think that's an inevitable aspect of aging, it's just how my personality turned out.
yeah, the only real benefit for you is that their map doesn't suck. But the bad part is that they could still shut down or charge for the service at any time.
Not the GP but I share the sentiment. I enjoy doing work that benefits others (in this case other map users), even if a third party stands to gain as well. I don't think of it as "working for": I understand the economics, but not all activities need be conceptualized as transactional.
i sort of agree... but on the other hand, at least they offer the "result" (Maps) for free.
LinkedIn for example lets user pay to search for (some of) it's "users own" data! :(
Wow, thanks for posting that, it is really nifty. And makes it much easier to confirm what I already figured for a while: OSM being open really helps in adding a ton of interesting details which matter to me, whereas the bigger map players are overwhelmingly undetailed in areas which aren't considered important enough (at leat I assume that is the reason). Like the small village where I live: OSM has even the smallest footpaths, even going so far as including an official one which doesn't really exist because it is currently part of privatly owned land, it is apty labelled 'old footpath xx' so one can see the new route is now 'footpath xx'.
This is the sort of thing I do in my local area; update OSM with anything I notice that's missing and seems like it belongs.
The bulk of my updates are local trails through the bush from when I go exploring, based on my GPS tracks. But I've also been adding pedestrian paths between streets, new roads from residential developments, and so on.
I'm guessing it just takes a few of this kind of person to keep OSM up-to-date. It's a shame it's not more widely used directly by the general public with some way of flagging errors.
Additional data point about the missing "coastline dropshadows", on top of the general "bleaching" trend: that kind of effect is also not free to implement in WebGL or in mobile apps. Plus the folks that were involved in designing it have since left.
How so? Drawing a drop shadow is not free, but on a GPU it should be rather cheap. It's just a blur of an all-black version of the object; not only are blurs fully parallel, they're separable (into successive operations blurring along the X and Y axes), reducing the number of input pixels that have to be considered for each output pixel. For a shadow in particular you don't need a very high-quality blur, you can act on single-channel greyscale images (less bandwidth), and the usual depth testing should skip running the pixel shader on pixels covered by the shape itself. And of course GPUs are only getting faster, not slower.
But the wording of your post suggests your info comes from the source, so maybe I'm missing something.
It's not expensive, but it all adds up. I put the words in quotes because it's not actually what is commonly known as a drop shadow (https://en.wikipedia.org/wiki/Drop_shadow). It's more like a silhouette shader. This picture shows the effect in full glory:
I can't remember if the highlighted water region was in a fixed location or if it followed the sunlight. (You definitely see building shadows change direction according to the time of the day in 2D WebGL mode, e.g. near the Eiffel tower. This is what I mean when I say that it all adds up, especially with WebGL 1.0 and an underpowered GPU like the original Chromebook Pixel's Intel HD4000.)
Do people actually use the Chromebook Pixel Mark 1 in the Googleplex and is 3D performance on maps a problem for them?
A while ago I took off the 'designer' font choices from a website (spider grey text on grey background, kerning mangled) with some local CSS. Everyone prefered the simple legibility to what was on my screen so we set the website to be like mine, legible.
I wonder if some people on the Google Maps team had a similar internal hack to optimise their workflow - turning off graphics cruft - get back to legibility - and if someone got this fed back into the actual design, 'as proven by the team'. I prefer this evolution of design, from people who have to get the data right in the first case and how they 'optimise' the presentation of that information for themselves.
> This all seems to suggest that Google’s location data is more precise than Apple’s. (Or that Apple’s geocoder is buggy.) And perhaps here we’re seeing the fruits of Google’s decade-long Street View project.
How can Apple catch up? Is there an obvious acquisition?
Waze was the obvious acquisition, because the Map editor is well done and accessible on the web, the community is very active and Apple even used their data, but Google was smarter and acquired them.
Apple Maps is limited to Macs and iDevices, while Google Maps is accessible essentially everywhere (web and Search page and Android/iOS). Because of this, it's easier for the masses to update information on Google Maps than Apple Maps.
Exactly. It works as a default included app and a tick on the features list. If better maps are worth it to you many options are available in the store, it doesn't hurt apple either way.
Last I checked, one of Garmin's key mapping data providers is Here Maps (formerly NavTeq). "Here" is owned by a consortium of companies (primarily VW, BMW & Daimler) and licenses its data to a number of companies. Apple can become another licensee.
For its massive cash hoards, I gotta say that Apple is neither strategic nor quick to the table with acquisitions.
My biggest problem with google maps is that it doesn't respect the font size settings of my device. A four point font is not ideal for my eyes in the best of conditions, much less when travelling and navigating.
I know it is useless to you if you are stuck on an old OS version but IMO Nougat did this right :
- changing the font size is a mess .. I work on UI on a large Android app and it is not trivial to make a design adapt to very large font sizes without sacrificing performances. Most apps don't bother though .. so you end up with mangled text fields.
-Nougat still support font size changes but has also added what I think is a better implementation : variable density.
The OS can 'lie' to the framework and the running apps and declare a pixel density different than its natural one.
That way the whole interface gets scaled (since interfaces are designed in density independent pixels).
Works like a charm and won't have any UI glitch because of it.
Justin O'Beirne's articles are extremely detailed as always. It's definitely a must-see for mappers whatever the company or products you're working with.
But he is consistent since 2010 in his point of view and analysis process. Getting even more precise, so or course it might have trolled a little bit but his arguments make a truly interesting whole.
I can't for the life of me understand why Apple pays TomTom for maps of places where OpenStreetMap is unequivocally more accurate. OpenStreetMap also has the correct layout of the footpaths at the north and south of the park, points of interest for every bench and a marked area in the location and correct shape of the playground, making the data (at least assuming Google renders their best data) considerably higher quality. I think this says a lot, especially considering the default leg-up that Google has here. They have images of the contours of every one of those paths, they have GPS traces, they have photogrammetry-quality photography and location, they have lots of people writing photogrammetry software; and yet for some reason they still don't appear to fuse the streetview, aerial, and satellite imagery, or seem to do feature detection for points of interest.
The big problem which seems to make it not worth expanding your dataset for graphical maps, is that it is quite difficult to display a lot of data, and still be easier to read than an aerial photograph.
> I can't for the life of me understand why Apple pays TomTom for maps of places where OpenStreetMap is unequivocally more accurate.
I suppose they can't easily, since the OSM license change.
My understanding is they'd have to open a lot of the geo data that they integrate together into their maps (POI, public transit, etc.), which might not be feasible as I suspect most of what they integrate is commercial and not owned by them.
(someone could check, but while Apple Maps had been using OSM e.g. for Pakistan, did they ever switch to post-relicensing data anywhere, or do they just keep using 2012 data?)
I mean, if OSM is more accurate anyway, then they might as well have a staff to generate their own data. There's no point in being greedy with their POI annotations if the end result is not sufficiently detailed to be useful anyway.
It's often easier to acquire it at least partially, it's not necessarily being greedy as much as not owning the 3rd party data in the first place (but having a license to it).
OSM is more of an all or nothing, everything has to be shared back, so it's hard to be pragmatic and mix different sources some open and some proprietary.
(and not every company wants to run a full fledged geo operation, and those can do it at scale will often end up keeping their data proprietary)
You can however, put separate and distinct data layers on top of your map, such as icons showing specialists points of interest, routes, track logs, shaded areas, contours and the like, then Share-Alike does not apply to these elements as long as they do not interact with the map underneath.
In the UK Apple maps is superior in one important way: it knows about road closures. Google maps here still doesn't seem to take into account planned closures, which are published by the council, so often it'll direct you right into a roadblock.
(Reading the article, Apple sources data from TomTom, which probably explains why they have good up-to-date satnav information.)
The thing I still don't get is, why don't any of these map services indicate traffic lights? When navigating to an unfamiliar place, counting the lights, or looking for the next traffic light, is much easier than using distance or street names.
In places like India, where landmark-based navigation is the norm or there aren't many road names (especially outside of large cities), Google Maps uses businesses as landmarks. A random example:
OpenStreetMap has this. I don't see a use for it, but if someone sees a use for it, they're free to add it to the dataset. Seems you and others find it useful, and it's in there :)
You wouldn't be able to rely on "turn right and go through 2 stop lights" type directions derived from OpenStreetMap, traffic lights aren't mapped all that consistently.
Depends on where you are. I don't remember seeing inaccurate crossings in the Netherlands, and scrolling over some roads just now (some rural, some in towns, some in cities), it matches 100% with my memory (true positives and true negatives).
Germany also has good data, but I drive there much less so I can't vouch for it.
Belgium on the other hand... I haven't yet found missing roads, which is surprising given what the map looks like.
It isn't super complicated data to capture, I think a single person could fill in a small to medium sized town in an easy weekend.
There's an issue where intersections aren't really mapped in OSM (at least in terms of priority, turn restrictions are increasingly captured), the pieces are all well modeled but there isn't much information about how they fit together.
Google Maps in Japan seems to be a few steps ahead of Google Maps in America, on a large number of details. I wonder who they have to compete with there.
Yeah, Google Maps in Japan knows a lot more, e.g. it shows actual road widths with delimited sidewalks or it tells you about train platforms and elevators when switching lines. The data is from a third party (ZENRIN, check the bottom of the screen) that provides an insane level of detail. Yahoo uses them, too, but Apple doesn't.
Ah. So it's not a question of Google having to assemble better data to compete with (insert other provider), it's a question of having local agencies that are much more proactive about assembling and maintaining accurate data than Google would be.
You could, but the number is usually much higher, counting can get iffy with complicated intersection, you need to worry about both sides, etc. Lights are usually only at major streets, so unless you're at the final steps of your journey, you will probably only be turning at lights.
It may seem strange but Yahoo is indeed the best maps app in Japan. Yahoo (Japan) also makes the best transit app here.
People give Google Maps a lot of credit but I find it really hard to use as a means for keeping track of places I like, or for discovering new places. For example, if I want to label a custom pin, I have to grant Google some sweeping permissions to use my web and mobile activity data. I couldn't grant that permission if I wanted to, as it's turned off somewhere in the bowels of my G Suite domain management settings.
It can't be coincidence that Google's increased focus on areas and places brings it closer to Apple Maps. It's easy to miss because of Apple's poorer data, but I think their app legitimately bested Google in some areas (heh), and this is Google absorbing those characteristics.
Id love to see them compare Here maps. I use Here for all my driving. Its much better interface, directions timing and has some unique features.Also I like the idea of supporting an independent player. While Google Maps is better when looking for businesses like a restaurant etc.
The majority of Here maps is owned by the trio of Dieselgate companies Volkswagen, BMW, and Daimler. I would hardly call them an independant player. I've also tried Here maps and about the only good thing about it was the ability to download maps and speed limit warnings. Everything else was mediocre and outdated.
Speed limit warning must have saved me a few tickets! I find its a better driving platform as its simpler. You can more easily read it at ta glance, whereas Google maps keeps getting worse on this front as they keep adding to it. Also their direction timing seems better. I think its a tad earlier and doesn't tend to get caught in loops where it tells you the same or irrelevant information again and again.
How can I configure OpenStreetMap to use the old Google Maps colours with an emphasis on roads?
I'm currently using some offline cached maps tiles downloaded with MOBAC, and waiting for Google to change their colours back again. Now I realise that it's a place map, and places generate advertising revenue, I think that Google is unlikely to fix that.
Another problem that happened recently in Taiwan was when Google removed pinyin (latin letters) from the street names, leaving only Chinese characters. Foreigners living here couldn't find their way around. I threw together a quick alternative to GMaps, and told people about it - until Google put the pinyin back about a week later.
There is Mapbox, they are a hosted provider which has a pretty straightforward tool for styling the map. I think they also have a starting style which is pretty similar to older Google Maps. I think they also have some level of customization for labels, and if you have a big need to have Pīnyīn I'm sure that something could be done for that.
It seems that with the increasing attention given to 'places' and less emphasis on the actual roads we are looking at Google getting Maps ready for the roll-out of self driving cars where the focus is on the destination rather than the journey.
I'm pretty map-aware and a bit of a pre-planner, but even my usual usage is to choose a destination and have Google give me directions. (Then maybe I'll tweak it or whatever.) Same with my 65-year-old dad who still buys a folding paper street map for each new city he visits. Does anyone still visually plan a route on Google Maps by just looking at the road markings?
No, because it is a pain to do, not because I would not want to. Rarely because it wants me to take a closed road. More often because I would like to travel through some part of a land and am not interested in the most efficient way of reaching my final destination.
I'm a little disappointed that the "Cartography Comparison" only looks at two commercial maps. Not even all commercial maps, let alone including OpenStreetMap. A big post detailing the minute differences between two of the big ones are not that interesting...
Well, here in Venezuela Apple Maps is practically useless. Traffic data just arrived about a year ago to Google Maps and it's been a fantastic tool to get on time to most places. Apple Maps doesn't even give directions.
My main complaint is the decrease is display priority of personal labels except for the stars.
Back in the day before labels I would star places I needed to "bookmark" regardless of importance in time and how ephemeral that mark was. Then when labels appeared I thought that was ideal to mark places which are always important (because I can personalize the label) as opposed to a generic star which most likely meant a temporary bookmark.
It seems that with this new Google maps, the stars always get display priority (it's shown even at smallest zoom level) whereas labels only appear at a algorithmically defined location (which seems arbitrary).
Because the Street View rig is expensive to manufacture; ordinary "driving the roads" can be done with normal vehicles. (Heck, you could even source data from usual drivers, if you owned e.g. Waze)
I think you mistook the section separator for the end of the article; that sentence isn't even halfway through it :) The rest of the article tries to answer exactly this question, and has a more interesting hypothesis than A/B testing.
> Over the course of a year, Google quietly turned its map inside-out – transforming it from a road map into a place map.
I've long been amazed how we somehow transitioned during the early 20th century from a mental model of roads and paths running through locations to places (house lots, etc) being the spaces between the roads. It's a natural thing to happen, but one of those invisible flips that happens on a timescale longer than a human lifetime.
But this anticipates the opposite: if you can stop worrying about how to get somewhere (because you don't have to drive or plan much -- self-driving or Lyft-style services can take care of the route planning) you can focus on the destination.
We see this phenomenon in subway maps which are famously schematic and not geographical.
(BTW the transformation is visible in literature, which is how I noticed it. The sense of geography in, say, Jane Austin is completely alien to today).