OSM needn't compete with Google Maps (as a consumer application) directly; it doesn't, really.
Lots of apps based on OSM data do need to compete, and that is very difficult to do on a global-scale. That's why most successful OSM-based consumer applications serve a few niches incredibly well. This is not counting the dozens of companies offering OSM-based base maps for use in other, mostly non-map-related uses (such as a store locator or data visualization tool).
OSM has the advantage of time, no need to make money, and a vibrant community around it. Google has to make money directly or indirectly via Google Maps to justify continued investment; OSM simply has to exist, and the longer it exists, the better the data will become. As the ecosystem around OSM improves (and continues building positive feedback loops), I strongly suspect we will see a successful consumer-targeted app using OSM data, and it will be much better than the many (good) apps right now, simply because another 5-10 years of accumulated data improvements outstrip the competition's willingness to continue investing.
Caveat: I cofounded an OSM base map provider. I'm biased. :-)
Some local public transportation "companies" in Germany (which are called "Anstalt des öffentlichen Rechts" - literal translation "entity of public law" - and are more or less non-profit businesses funded by the city) are using OSM data for their apps and websites and they often edit the bus/tram/train routes and stops if they change them. This is where OSM shines: public entities keeping their data up to date for the benefit of all.
"Anstalt des öffentlichen Rechts" - literal translation "entity of public law" - and are more or less non-profit businesses funded by the city
As a point of interest, the American equivalent of this is called an "Authority." Examples include the Chicago Transit Authority, and the Port Authority of New York and New Jersey.
It's not just about market share. In both cases there is an intermediary between the (public) service provider and the users, which is the public. In one case the intermediary is a private entity that answers to no one; in the other case the data belongs to all of us. If you go beyond a mercantile view, it's easy which one is the best for the public in the long term.
I was happy with how OSM was run before. Only recently became unhappy with it, when they decided to make that move. Note also, that it was a silent move for the casual OSM visitor. There was no big sign saying: "Hey, we switched to fastly! This means your requests will go through their servers. Just wanted to give you a heads up for you privacy conscious people." No justification, no asking for donations for running their own instead. I've supported multiple other projects, which I feel are important for the public and would have considered it for OSM to stay at OSM and not outsource stuff to fastly.
Now I cannot, with a clean conscience, recommend it to people any longer. What will my answer be to anyone, who asks me: "Well, if you are not using Google maps, what map service are _you_ using?" Now I cannot answer OSM, because then it will be: "But they are also allowing big companies to track their users! You are a hypocrite!"
Basically I need something better now, in order to stay consistent and not become a hypocrite. Perhaps I will use my months ago updated maps.me on my phone instead. However, with maps.me issues of late, I will need to switch that eventually as well.
Remember: Ignorants and uninformed people are not to be optimized for, if we want any quality standards to survive on the WWW. To keep standards, one cannot always go for the lowest common denominator.
Although I understand that using any external service provider is a potential concern, are there specific issues with Fastly's terms or practices that you consider worrisome? I'd be interested in anything we may not have fully considered when we decided to make that move.
Primarily I am worried about fastly being used by loads of other websites and them getting data about me from usage of multiple websites, especially location sensitive websites, like map websites like OSM. If I allowed their scripts on all those websites, holy moly, the profile they could build about me ... Best way to protect your privacy is to not let the data be collected in the first place.
I am no lawyer and do not wish to read many pages of their legalese. However, if they somehow managed to convince me, that they are not building profiles, that they are not transferring or selling data to anyone else, and that none of the software they use does so behind their backs, then I might trust them. Now that is very difficult to do of course, almost impossible. I would rather trust a single entity, that is not present in any requests to any other website I visit. That seems to be much lower risk than putting data all in the hands of a single party, which is present on many many websites.
That's for the website tiles. If a transit authority is using OSM data, they would download & process it directly themselves, and hence this doesn't apply at all.
I feel like when saying this, you have to mention that it is(or at least was when I was tasked to do so 2 years ago) notoriously crazy to host an OSM tile/nominatim server for a mid-size country, and even more so for the entire world.
Requires some dedication and lots and lots of disk space and bandwidth - probably most feasible in a box at home with only you yourself having access and not having to pay monthly for high-traffic storage.
It is a bit crazy, but at least it is possible. So if someone has special requirements (100% pure software, special rendering, special routing, special data analysis) they can do this.
What is both illegal and basically impossible and even more complicated technically to do with Google Maps data.
I think an easy to setup self-hosted vector/tile server could help increase adoption. I'am currently looking into replacing Mapbox with a self hosted solution. My last attempt ran out of disk space building the docker images (with european maps) from https://github.com/Overv/openstreetmap-tile-server. Haven't found the time to setup a larger VPS but this looks like a great project to get started.
Then it provides various services (routing, PoI search, geocoding, etc.) over this data set to all applications running locally on your machine. It targets mainly various mobile Linux distros but works perfectly fine on desktop as well and is available in flatpak form.
Like this all navigation and mapping can share the same data set and no potentially sensitive location related metadata is leaked to a remote server other than what data packs have been initially downloaded.
Also the community run infrastructure just needs to be able to store and store and distribute the ~150 GB of data packs covering Earth but does not need to have any extensive compute and memory requirements to handle lots expensive of individual API queries.
I use https://openmaptiles.org/ at work. For normal, Web Mercator maps we just use the free download of the whole world, available at [1]. Running the OpenMapTiles tools to generate a current tileset for a particular region is straightforward, but if you need all of Europe you will need some decent compute capacity. Unless everything can be done on a single computer, you'll also need to implement some missing pieces, e.g. a message system to distribute the rendering across your computers.
It’s simply more efficient to update OSM, which makes OSM more up to date.
Google maps pulled data from OSM so the government had ~zero incentives to publish to GMaps and every other mapping service directly. It’s simply more work without benefit.
PS: If I remember correct Google was on a 2 week delay for OSM data at the time, but that wasn’t considered meaningful. Also, what’s with all the hate here?
> OSM simply has to exist, and the longer it exists, the better the data will become.
Probably, but not necessarily. The world changes with time and if OSM contributions languish, it falls out of date with the world and tangibly becomes a worse product. Certainly not inevitable, but a possibility worth noting.
Agreed. In fact, I think the OSM total fraction of errors will exponentially decay towards some non-zero fixed fraction of errors. That's because, on average in the long term, the world probably changes at an approximately fixed rate (relative to total mappable features in the world) and OSM is updated/corrected at a fixed rate (relative to fraction of mappable features that are wrong or missing). Even given the benefit of a lot of time, the OSM can never catch up with a mapping service that updates faster because that other service will have a different equilibrium.
As of current technology every geographical database lag behind real world and took time to catch up. 20 year ago the average lag was often accounted in years between two survey.
It would be nice to determine some random controls points and compare lag on different mapping service. The average lag would maybe maybe drop under a year for some (and it will depend on which feature, road update faster than building if you account for gps data).
However I really don't see why OSM would necessarily lag more than Google Map or any commercial database. Because the more OSM usage is widespread (directly or through apps) the more human "field sensors" you get to update the map. Same way that I don't think many commercial encyclopedia can claim that they are more up-to-date than Wikipedia to record new facts.
I'm with you on the "free projects have forever, proprietary projects have a deadline" line of thinking, I use it often when talking about social networks and microblogging. But one thing to consider is the motivation for predatory or destructive action. The OSM community does not have the motive or the resources for destructive activities towards Google, whereas Google has a motive and resources to sabotage OSM. This sort of thing you see historically with Microsoft and EEE, with Facebook and XMPP, and with lots of other things, including some of Google's behavior.
I wrote about this in another comment in this thread (and talked about a solution to the problem), but I believe the existing OSM navigation apps that exist work well, and that the remaining friction point preventing them from competing with google in the consumer navigation market is address lookup. OsmAnd for example is very user friendly and I use it almost exclusively, but looking up addresses is the main difficulty.
I really wany to love OsmAnd, but I'm not sure I agree on the user friendliness yet. First thing user feels is how slow the rendering is while panning. I can't measure it but I'd guess about 10fps on my phone, and the area revealed by panning is all grey for a good while before it gets drawn.
It's fine for a tool, but has no chance to compete for casual end users like this.
But fast OSM renderers exist. Take a look at
https://play.google.com/store/apps/details?id=gr.talent.crui...
for an example. It's perfectly smooth. The application is closed source but the underlying renderer (VTM) is LGPL3. The renderer, https://github.com/mapsforge/vtm, is basically an OpenGL alternative for the mapsforge offline maps format and maintained by the same guy (the original mapsforge renderer is arguably a bit slow, struggling to hide its slowness behind cached tile bitmaps).
> the longer it exists, the better the data will become.
This is unfortunately not a given. The world changes constantly, and things like having up to date coverage of restaurants require an immense amount of essentially ephemeral work to keep up to date.
That's why there should be a decoupled architecture.
I am not an OSM expert. It may have this already. I am all for OSM, and sincerely wish it great success.
What I do have, however, is experience in founding a worldwide, federated infrastructure that supports a fairly specialized (read: "small") demographic, but does that exceedingly well. It has become a worldwide standard, and is being taken to the next level by a team that doesn't really include me, anymore.
It has a lot of structural support for federation. The idea is to push specialization away from the core as far and as fast as possible. The core is a relatively simple, tightly-focused, ultra-high-quality database engine, with a powerful semantic API. The services are addons that were written by some really energetic and skilled people that bring far more to the table than Yours Truly. In fact, these addons are what made the system suddenly become a "must have" for everyone.
I'm assuming that OSM should have a "core mission" of geographic data, which is really just geolocation stuff. Basically, tagged locations. The tags can be used to relate the locations to external datasets.
I would then assume that things like businesses, government services, and other geotagged datasets are attached as adapters to the "core," so they can be managed by others, who are specialized.
Google does this (with arguable success) by letting businesses and individual users contribute data. I assume that it uses its own tools to aggregate and clean that data, and probably applies the data in an adapter fashion. This is what they are good at, and at vast scale and performance. Say what you will about them, but they do "big and fast" really well.
I have found, through my own experience, that if I want local data to be accurate, it is important to let locals directly supply that data.
In my experience, that started with an infrastructure designed for federation, and is realized by writing data harvesters that provide a rich, simple UI. Also, it is important to keep the core "pure." Specializations should be supplied at the adapter level; not by changing the core. For the most part, the central database schema of my system hasn't changed in ten years, but it does a lot more now, than I ever dreamed it would, back then.
But that's just me. I don't think I'm anywhere near the level that many people live at. My experience is fairly small and humble.
I've once spoken to an OSM "evangelist" and he really struggled to get this point.
In a way I got his point the OSM model can virtually store anything so it's the data provider burden to develop a glue code to interact with OSM.
However because of potential vandalism or sheer goofiness objects and ids can potentially break at anytime within OSM so writing glue code is actually not trivial.
Maybe a half way solution would be to promote generic glue code framework. I've heard about this project by a french local authority that help sync back data from OSM to your own database. Dunno if other similar projects exists.
Open source software is often comparable or even superior to the proprietary sort in feature set, the kind of thing you can list in bullet points. But it tends to fall down when it comes to user experience.
I remember attempting to use OSM on my phone before, and I remember being similarly confused about "which app should I use?" and the ones I tried just being fairly clunky. But I'll give maps.me a try, I can't remember if that was one of them.
This can be a tumultuous time to try maps.me. It recently got a new owner that decided to use it as a distribution channel for their digital wallet. I'm not sure where it stands right now, but at one point they had thrown away the previous app nearly entirely.
For more context check out [0], which had a major discussion when it was posted. In short, the source code of Maps.me is open source and people are working on building from that, yet the Maps.me you find in the play store isn't the same as the one so many people recommend.
This kind of garbage is why open source software can't compete with the likes of Google. I started out reading this comment chain thinking "great, I'm going to try maps.me". 10 seconds later, I've learned I can't just try it, I have to find the right fork, of which I see several, each with people mentioning serious caveats. And who knows when the currently best fork will get taken over by the next wallet software huckster?
Come on, this has nothing to do with it being open source, rather the contrary.
The same thing could happen (and has) with proprietary software, and you are then out of luck.
With open source? Use the last known good version, fix wat you need to, even if you are the only one person on earth interested in that piece of software. Maintain it. Or just find someone else who is doing it (that's a "fork").
I've stopped using a lot of proprietary software and apps due to this, though examples are fading from my memory. On Android, I think I can name at least ES file explorer and Touchpal that went down the Adware hole.
Nowadays I only use FLOSS. Not even GAPPS. OSMAnd (which I donate to) is pretty nice and featurefull, it suits me as a maps power user, but it can be a bit daunting. There used to be a maps.me fork on F-Droid, but it was recently removed: https://gitlab.com/fdroid/fdroiddata/-/merge_requests/7951/d...
Looks like the maps.me community is reorganizing, and hopefully it will re-emerge stronger and more resilient that it was before. Also, lessons learned with this might be useful to other projects.
> And who knows when the currently best fork will get taken over by the next wallet software huckster?
I agree, and I bought plenty of both paid and open source apps in the past. But I find that having the source code offers stronger guarantees than a receipt (some paid-for proprietary software also let me down, if only for becoming abandonware), that's why I only use proprietary software if it's in a comoditized niche with plenty of interoperable alternatives.
For instance, I bought every "simple mobile tool" app (from Tibor Kaputa) on the play store (even though I don't use Google Play, at least they are in my family library), as well as OSMAnd plus its plugins. I also bought some open source games on Steam, such as mindustry. Krita is there as well, but I prefer donating directly to KDE.
I am totally in favor of paying devs, though I generally prefer to make anonymous donations, trough liberapay (unfortunately not tax-deductible) for instance, as a way to cut the middle-man, (I tend to spend more than 600€ a year on free software donations, which is not negligible to me: almost half a monthly salary).
I don't really understand your point, to be honest: ES is "abolutely proprietary" as far as I know, so it won't be available on F-droid. I was trying to come up with examples of proprietary software that changed under me. Source access is a must for me, this is about freedom, not price. I also have trust issues with proprietary software.
Proprietary software doesn't have the same or worse garbage? At least this way when the "vendor" went bad, the users were only annoyed instead of fucked.
I'll take "this kind of garbage" Every Single Day to the end of time.
Maps.me isn't really "open source software". It's a commercial entity and one of the most popular mapping applications in the world. (Definitely #1 for navigating on foot.)
The point is that yeah, competing with Google in some field that isn't integral to Google's advertising business is quite possible and realistic. There are real-world examples.
There are benefits to dictatorships in software, which can help avoid fragmentation and give customers "the one right choice". However, such dictatorships tend to falter and fail as the desire to serve the masses with a simple, yet effective option morphs into disregard of user preferences and rent seeking as those users are locked in.
Google is a powerhouse today, but that power is built on a lot of the open source software (e.g., Android) that you claim cannot compete. I am willing to bet a significant sum of money that within the next 10 years Google (which by now forgot its "do not be evil" origins) will be successfully challenged on both the search and the mapping fronts by an upstart leveraging some open source software. My 2c.
Ah okay, I just downloaded the one on the play store and it did seem pretty bad? I mean initially it felt alright glancing around, but as soon as I searched for "chinese restaurant" it just immediately fell over and did nothing useful.
That might be down to data i.e. OSM being crappy in your area or the app being bad... Hard to tell, it worked for me back when the app was with its original developer. So maybe browse around on osm.org in the same area to see if you have data in your area.
I live in a major European city (Munich), so the data being that bad would strike me as a sort of indictment. Based on how people talk, I thought big European cities would be one of the strongest types of areas.
I tried out your idea -- I went to the official site, openstreetmap.org, zoomed in onto my neighborhood, and then searched for "Chinese restaurant" in the search. It returned a result for a place apparently literally named "Chinese Restaurant" in Portland, OR. That doesn't exactly fill me with confidence.
I love OSM but what you say is correct, search is hard, or rather no one cares about the search part of OSM. That is why you need to browse around on the map to judge data quality, and even then you might nor see all that is there.
The data is usually there if you live in europe, it's just hard to search. There are two related things that make it hard, bad tagging of the data and the extreme diversity of data. E.g. this winter I've used osmand to find fire pits, it was a great way to use the data but there seems to lots of ways to tag firepits so a search might not find all of them.
At least here in the USA, address locations are not public data in most of the country. City maps are generally copyrighted and cannot be used in OpenStreetMap. Google has paid for access to proprietary address data. Osmand isn't a source of data, it is not at fault here. OSM contributors need to add addresses manually by walking down the street and using an OSM mobile app like StreetComplete to fill in the gaps.
So I'm no power user, but I've been using OsmAnd instead of Google Maps as a simple map app for about a year. If I just need Street names, fine. If I need to search for any business address, it sucks. I still need Google Maps on my device.
You say "contributors need to add addresses..." but this is so unintuitive that it's simply not going to happen in over 99% of cases. I say this as someone who really believes in the cause and who therefore makes up the 1% that would be willing to spend the time improving the data. (I still buy digital music and manually add my own mood/instrument tags, if you want a 'for instance')
So why don't I? Case in point: I wanted directions to a restaurant that opened near me about 2 years ago. Couldn't find it. So I walked there, ordered my food and while waiting for it to be served, manually navigated to the map position and found the previous restaurant was still listed. As far as I could see, there was no way for me to edit the map entry within OsmAnd. As an interested party, willing to invest some initial setup and learning time, and ongoing edit time of a minute or two, I completely failed. There's nothing in the UI to even hint at how to edit the address. A 'regular user' isn't even going to bother trying.
At this rate, the world will change much faster than the OSM data.
FWIW, I looked into it and apparently the "online search" functionality gets way more hits when searching for house numbers. That search must use nominatim or pelias as the backend. You can find it by hitting the search button, going to the categories tab, then scrolling to the bottom of that tab.
It really depends on the jurisdiction. Lots of cities and counties have open data, and lots of states too. There's also a national database that looks pretty interesting:
The UI improved but it's still a maze of options and menus. I use it to plan my cycling routes but I gave up on the navigation features. Creating a route from markers is a nightmare of complexity. Then there is the problem that with a bicycle I can ride on pretty much every surface and it's impossible for the poor thing to guess an itinerary I would like. Not it's fault.
What I do is discovering roads and paths with OsmAnd, checking them with the satellite view of Google Maps and sometimes also with Street View, then
placing markers in OsmAnd and following them when I'm on the new areas I still don't know.
If OSM had satellite pictures I could do without Google Maps for that. I would still use it to check the opening hours of shops.
Well, you can add online tiles to OSMAnd as an overlay, it supports multiple providers (even custom ones, so you could use the high quality IGN Ortho for France, for instance). I actually quite like the route's output.
> Creating a route from markers is a nightmare of complexity.
Is it really so complex? Globe icon or menu -> switch to cycling profile -> itinerary icon -> adjust start position and destination, markers are suggested when tapping on either. You can add more intermediate markers by tapping the "+" between source and dest.
You can press any part of the map -> directions -> add as {subsequent, first intermediate, last intermediate} destination. You can also tap existing markers to do so.
In my opinion, it doesn't get much easier than this? I was so confused at google maps' interface when I was handed a phone to plan a route three days ago, it felt like it was burried down more menus, reorganizing destinations was a pain, while it's just edit -> drag and drop in OSMAnd.
For biking, also try brouter: http://brouter.de/ I know it can be used with OSMAnd, but I never spent enough time to figure it out. The web version is quite good already.
Yes, and it's only on F-Droid. Apparently there's OsmAnd and OsmAnd+, and the one with the tilde is basically the + version with no dependencies on google services. At least according to german wikipedia, which had a convenient little table
maybe they renamed it, I was just reciting wikipedia, it might be outdated. In any case, OsmAnd+ is the paid version that OsmAnd~ is supposed to be equivalent to, so I imagine they just decided to rename it to make things less confusing
e: I can see it on the f-droid site, but i can't find it in the f-droid app either:
Maps.me was recently sold and got a lot of negative press for their latest changes (becoming more GMaps like, hiding detail etc). Discussion from 11d ago:
It just has a cluttered and sometimes confusing interface (in my experience) which makes it difficult for me to intuitively use it for routing especially. So I typically use Apple Maps for routing—because it just works and is easy.
HERE Maps (called 'Here WeGo' on the App Store) is usually better than Google maps for me in a car, at least in Europe. It's owned by the German car industry and used to be Nokia maps.
Prior to being bought by Nokia, HERE was Navteq. The very early version of Google maps actually included a notice that data was licensed from Navteq -- I learned of Navteq from Google Maps. I believe Bing maps also licensed data from Navteq.
just a quick check for a route in my (european) city showed it tries to route me through a road closed 5 months ago, and doesnt even know my 4 years old house exists.
this isnt the case with gmaps
id really like to use less google, but unfortunately gmaps doesnt appear to really have a viable alternative for my (pretty common) use case of traffic-aware city navigator
It's likely it will never compete. Either its data will never get better than Google's, or it will. If it does, Google might switch the mapping source for Google Maps over to OSM, and make loads of money off it that way. A bit like MS Edge on top of Chromium.
You seem to operate under assumption that Google maps have great data or UI. They don't have either.
Google maps may or may not be good for road travel, I don't know. I don't drive. They certainly seem to suck for public transport planning because of silently added walking times that you can't turn off, and overcomplicated 3D-chess like UI, that's hard to navigate and automatically jumps around the map for no really discernible reason.
On a recent holiday, I had to constantly switch between gmaps and one other app with better actual mapping data/display (just because one has better public transport data, and the other is better at actually being a map - having things like a scale!, or contour lines).
A lot of walking/cycling tracks were just not there in gmaps at all, nevermind the washed out display style of the map tiles and lack of clearly visually defined features you typically orient yourself by when walking, like houses, or other important things you need when planning a walk, like contour lines. Or let alone a scale for quick orientation on how far something is at the current zoom level.
All they really have going for them at this point for me is public transport data (not the UI though).
In some ways, I'd argue it is better today. I find using an OSM-based app is far more likely to show trails than Google Maps is. (Probably because Google 1.) Doesn't really care and 2.) Official data for trails is pretty fragmentary.)
Exactly this - Google maps seem to really have all the drivable roads in all areas I've been in but hardly have anything at all that is pedestrian only as far as I can tell.
As an example in 2019, we climbed on Mount Shiroyama above Kagoshima and looking at Google Maps the only way down to the city was the one we came from or some car only tunnels.On the other hand OSM listed a path with lots of stairs the locals apparently use to get up and down the mountain quickly.
If we used only Google Maps we would have to backtrack quite a bit over parts of the city we already went to but instead we arrived back in the city quickly just in time to get a tasty evening ramen. :)
> Either its data will never get better than Google's, or it will. If it does, Google might switch the mapping source for Google Maps over to OSM, and make loads of money off it that way. A bit like MS Edge on top of Chromium.
Sure, that's reasonable, too. I think OSM (the project & database) wins in that situation, even if its derivations don't compete.
Again, I don't really see OSM directly competing with Google Maps (the app).
Openstreetmap is not an end-user oriented product - but it is perceived through such products. Understanding what might make it more useful to developers benefits from competitive analysis of the consumer products.
Yeah I was here to say this. OSM doesn't even compete with Google Maps, or rather, Google Maps doesn't even compete with OSM if you want to do any kind of large scale analysis on streets without silly licensing restrictions.
Lots of apps based on OSM data do need to compete, and that is very difficult to do on a global-scale. That's why most successful OSM-based consumer applications serve a few niches incredibly well. This is not counting the dozens of companies offering OSM-based base maps for use in other, mostly non-map-related uses (such as a store locator or data visualization tool).
OSM has the advantage of time, no need to make money, and a vibrant community around it. Google has to make money directly or indirectly via Google Maps to justify continued investment; OSM simply has to exist, and the longer it exists, the better the data will become. As the ecosystem around OSM improves (and continues building positive feedback loops), I strongly suspect we will see a successful consumer-targeted app using OSM data, and it will be much better than the many (good) apps right now, simply because another 5-10 years of accumulated data improvements outstrip the competition's willingness to continue investing.
Caveat: I cofounded an OSM base map provider. I'm biased. :-)