Hacker News new | past | comments | ask | show | jobs | submit login
OpenStreetMap proven to be a highly accurate map in top US cities (lyft.com)
880 points by clarecorthell 63 days ago | hide | past | favorite | 243 comments



Good that OSM is accurate in cities as well, because in hiking, skiing, or even some ferry routes, OSM simply wins because that's often the only mapping provider to even have _any_ data. Some of my first hand experiences:

- French GR-20 routes. Google maps are laughably empty, while OSM has you covered with almost the same information as the hiking maps you can buy on the trail.

- Anapurna - similar to GR-20 situation. This route gets changed often due to landslides, but there is always some person doing a great job updating information not too late after.

- Volcanoes - OSM maps often contain camping grounds, water sources, etc. This is something I actively contribute too. The level of detail is amazing. Some guides in fact lose potential clients because of this.

OSM is a wonderful feat by all its contributors, and isn't appreciated as much as it deserves.


In my country, there is an amazing government mandated service called Swiss Topo (https://map.geo.admin.ch), which can be accessed freely. Here are some features:

- detailed maps from 1:10,000 to 1:1,000,000

- hiking routes which are actively maintained. You can create your own hiking routes, and it will estimate the time that it will take to complete them.

- historical data, like historical maps since 1864 and aerial photos

- aeronautical maps, naval charts, geological maps, ...

- basically any kind of data that you can find on maps, such as land registery, water planning, spatial planning, etc.

I am not sure if other countries provide this level of service free of charge. I would be curious to see what other countries offer on this topic.


NSW, Aus offers SIX Maps: https://six.maps.nsw.gov.au (this offers very good imagery of the NSW areas, better than google)

That's the easy-to-use-maps.

There's also the aus-wide https://nationalmap.gov.au where you get basic satellite maps with the option to add various datasets and even daily maps from Landsat 2A

There's also ELVIS foundation spatial data https://elevation.fsdf.org.au

There's don't really cover hiking trails that much, etc. There's more general purpose mapping applications.

It's also possible to find various topographic maps in static form somewhere on the government sites.

Councils are the ones that provide the hiking trails, etc. so it depends on the locality. The Northern Beaches council for example provides an interactive trails map, but most have static maps.


Wow - very interesting.

Btw: the URL for Six maps is: https://maps.six.nsw.gov.au/

Not sure if it was changed but it took me a few tries to find the right permutation


Thanks for this comment!

How do you view daily maps from Landsat 2A? I couldn't find it in the options on nationalmap.gov.au


Those are all government funded and the data is private, correct?


Here the national land survey bureau (ČÚZK, rough translation) maintains maps in a similar scale to the one you posted. But they were opened only recently (they used to be for sale for pretty high prices) and the bureau haven't built a compelling web interface over them. The current UI [1] is basically just an ArcGIS frontend, with no automatic layer choosing based on zoom level and quite slow at that.

Planning maps are usually unfortunately only held by the responsible subjects (towns and larger administrative blocks), so there's no one map or one format. Most of these maps are at least published in some viewable capacity, though, from what I can tell.

[1]: https://ags.cuzk.cz/geoprohlizec/


swisstopo is really good, but as far as i know, sadly, only some data is open, and not all of it


There's a bit of discussion in `osm_ch` that from March on, the SwissTopo data will be free OpenGovernmentData: https://www.swisstopo.admin.ch/en/swisstopo/free-geodata.htm...

The debate is what 'free' means :)


Oh, this is very interesting. Thank you for the link.

I think some of the questions about what 'free' of charge mean are answered below in the frequently asked questions section.


I know that I dearly miss the French IGN. It’s close to impossible to find decent paper map for hiking in the US.


I've found the Nat Geo waterproof maps to be quite good. Between the "Trails Illustrated" fold-up maps and the "Topographic Trail Guides", the east coast is pretty well covered.

https://www.natgeomaps.com/trail-maps

I also use Gaia GPS as a planning tool. Fiddle with routes on the laptop, then load the relevant map squares on the phone for offline use. It has the NatGeo maps, plus several other sources, available for overlay. With this, my NatGeo paper maps largely become backups.


Have you tried USGS topo maps?

I guess the paper will be out of date, I'm not sure what the new electronic ones do for local and regional trails (I think they would print okay, they just aren't distributed directly on paper).


OSM isn't perfect--especially on trails that aren't official in any way. (Although there are some totally informal trails literally out my door that are far better than you could reasonably expect. And the missing sections are probably on private land which is one reason I haven't filled them in.) But Google seems to have pretty much zero interest in mapping even popular official trails in many cases.


Nothing is perfect, not even official paper maps.

I remember a hike we did on holiday in the Drôme department of France when I was a little kid, long before OSM or Google Maps or even Microsoft's Terraserver. Must have been somewhere in the mid to late 80's, I think. We did have a paper map though, and I think it was an official topographic map. We started in one valley and the plan was to hike up the ridge and then down to the neighboring valley, where we knew the Tour De France was going to pass that day. Watching the Tour was not really the main objective, otherwise we would have driven to a more suitable spot by car, but it was an incentive to keep us kids going. In hindsight we could never have made it on time.

The map clearly showed paths on both sides of the ridge, and sure enough, walking up the ridge went smoothly (if tiring) along that path. But when we arrived at the top, there was no path down the other side to be seen. There simply wasn't one, and the terrain was not exactly suitable for walking down without a path. The world simply didn't conform to the map. So we stayed there for a while and saw the Tour pass far below us, much too far to recognize anything except the helicopters but even those looked more like flies than helicopters from our high point of view.

That wasn't the only time we saw significant differences between French maps and France itself, but it was by far the most memorable one.


For people not knowing it: https://opentopomap.org

OpenStreetMap with the traditional topographic view. But I was not able to find where you wanted to hike to see the Tour de France.


I don't know exactly either, but there are some clues I thought of after I posted that story. I remember we were cheering for Flemish cyclist Fons De Wolf since he had won the stage the day before. I found out that he won stage 14 in the Tour De France of 1984 (which means my earlier estimation of mid to late 80's was not entirely correct), so the stage we observed from a distance must have been stage 15, from Domaine du Rouret in the Ardèche to Grenoble (https://en.wikipedia.org/wiki/1984_Tour_de_France#Route_and_... and https://en.wikipedia.org/wiki/1984_Tour_de_France,_Stage_12_...)

Sure enough, that passes through the Drôme departement where we were on holiday. I don't have the exact route for that stage, but it's described as a hilly stage so I assume it passed through the Vercors Massif on its way to Grenoble. That's most likely where we were that day, but where exactly I can't tell. I could ask my father but I don't think he'll remember either. It's quite a long time ago, after all.


There's an OpenTopoMap too, at least on Android. I think it's amazing how much information on the world around us we can now get with almost no effort and cost.


That's pretty good - almost as good as OS 1:25,000 maps for landforms and actually better for showing paths.


I'm often hesitant about mapping the unofficial trails, particularly the ones you aren't supposed to use. There are many around the SF bay area, some are simply non-maintained non-system trails, while others are illegal mtb paths off the allowed trail network. Some of them are so well traveled in areas with minimal erosion that they are almost as good as maintained trails, although often less well graded.


Yeah, these are basically ATV trails near my property which are probably mostly not supposed to be there. Certainly on private land although conservation land as well. I've been appreciative for them during current times and I've even been appreciative that the ATVers tend to come in and clear fallen trees and the like. But that doesn't mean I feel comfortable plotting tracks on OSM--especially those parts of the tracks on private land and especially those parts where ATVs have ripped up the terrain in said private land.


In thr Uk, I believ, paths 'legalise' themselves - once a path is established, if used continously it will become a public right of way.

Then you cant close it or obstruct it even if you are the land owner


Help contribute back to make it amazing. That's why OSM is amazing :)


It gave my old Garmin GPS a new life, where Garmin stopped providing updated data. A packaged map file with OSM data made it useful again.

For those in a similar situation, I use this

https://www.openmapchest.org/maps/canada/


For those in a similar situation which want this for Europe, too, use http://garmin.openstreetmap.nl, which has worldwide coverage.


The other great mapping resource is CalTopo (https://caltopo.com). Outside did an article on how CalTopo is used by people who work and play outside: https://www.outsideonline.com/2229756/your-navigation-outdat...


That story (from 2017) is fantastic. CalTopo - and OSM, at best of times - a project of passion. Thanks for sharing!


I often notice when OSM is wrong/missing data while hiking and I'll record a trace and update when I get back home. I enjoy contributing but I wish there was a bit more of a community aspect to it (or maybe I simply haven't found that). In some areas the US government agency maps are hilariously out of date or simply completely wrong so OSM is great to able to simply fix it after visiting.


You can find something at https://wiki.openstreetmap.org/wiki/Contact_channels or https://community.osm.be/ (click on your area).

Personally I follow

- global Discord channel (and local one) https://wiki.openstreetmap.org/wiki/Discord - global and local Telegram - https://wiki.openstreetmap.org/wiki/List_of_OSM_centric_Tele... - forum of a local community - https://wiki.openstreetmap.org/wiki/Tagging_mailing_list (if you enjoy participating in design of tagging schemes) - https://wiki.openstreetmap.org/wiki/Talk_mailing_list


For hiking, what I also like to do is add small and "useless" features to the map, that can help navigation.

Like, if there is a small open spot or some minor structure in the middle of the woods, I will add it. If someone is walking and not sure if they are still on the right track, they can look around, see the open spot, see the open spot on OSM, and know that they are indeed at that place.


I've really enjoyed the community at slack.openstreetmap.us great for finding projects or people to work with. There are local channels!


Is there an easy way to contribute from my phone? If I could just turn the GPS on and let it track my path I’d be able to contribute all of my random hikes that might not be on OSM.


You can directly upload GPS traces:

https://www.openstreetmap.org/traces

That won't get them into the OpenStreetMap database, but it makes them available (if you set them to public) for anyone looking to add the trails.

There's also apps like StreetComplete. It has 'challenges', where it prompts you for answers to questions and updates OpenStreetMap accordingly (so what is the surface type of a road, things like that).

Vespucci is a full featured editor for Android.


OSM discourages adding raw GPS tracks to the map, rather they should only be uploaded to a sort of waiting room hosted on OSM's infrastructure. A GPS track from a phone or unit is something that needs to be carefully examined, compared to other GPS tracks from the region or aerial imagery, and then tweaked and refined. You need to be able to use specialized mapping software properly in order to contribute tracks to OSM. The good news is that the barrier to entry is fairly low for people like you and me on this "news for nerds" site.


There is bunch of suggestions on how to contribute to OSM with your mobile here: https://learnosm.org/en/mobile-mapping/

On the left side you have a list of apps/devices with guides and summaries


Street complete (fdroid/playstore or GH) is a very easy way to use your phone to quickly to edits or validate data. https://github.com/streetcomplete/StreetComplete


You can also add details using https://wiki.openstreetmap.org/wiki/StreetComplete, it's super simple and almost like a game :)


I quite like that interactive website that helps visitors choose a mapping tool: https://whatosm.pavie.info/

It asks the difficulty level, time you'd like to spend, and whether you want to map at home or on the terrain.


You should be aware that it rather biased (doesn't show tools that are in competition to those that the author was involved in), and out of date.


That could be, but it's a start, and helped me discover a few tools.

Do you have a list of such tools? Do you have a source for attributing this to malice rather than ignorance?

Now, my personal opinion on mobile editing: Vespucci is a bit complex, but basically allows you to do anything on the go. I ususally save GPX tracks using OSMAnd, but it can be heavy on the battery.


After registering, it took me only like an hour to have my track up and mapping it with their brilliant online editing tool. There are some gotchas, but the interface is really friendly and their wiki is vast with info, references and tutorials alike.


Interesting thought.

I have GPS traces for a number of Amtrak routes across America. I wonder if those would be useful.


I imagine something as large as a railway is already covered (especially with the railfan community). In fact in most locations they actually include the number of tracks, passing loops and other features that a simple GPS track will not have.


Why not? It is always informative to have more information. Especially if you know ahead of time, what side of the train to be on, to see more interesting things. Also having information on one map program is great. Why switch apps? Especially since OSM is so privacy friendly.


While I agree that OSM is great and highly accurate, I would be careful using it in remote locations where your life may depend on a correct trail. It is very easy to edit OSM (changes are propagated into map tiles after a few minutes), and malicious intent is not immediately spotted. The latter is especially true in remote locations.


I usually look at Strava's heatmap when I'm unfamiliar with an area. Chances are if there's a path on the map, it shows up on the heatmap.

https://www.strava.com/heatmap


That is a fantastic idea. Thank you for sharing. I just checked a few trails that were missing in OSM and they all showed up as bright yellow on heatmap.


Strava even has their own fork of the OSM online editor iD[0], that uses the heatmap data to align (or "slide" [1]) paths to where people have been. I haven't used it in a while, it was pretty slow when I did and I don't know about licensing etc, but it looks like it still works. So if you want to add the trails you could try it out.

[0] https://strava.github.io/iD

[1] https://labs.strava.com/slide/


Slide unfortunately hasn't worked for years. Not long after Paul Mach, who created slide, left strava, there was that whole national security incident with the heat map.

The heat map was changed to hide military bases, etc, but it broke slide and no one cared enough to fix it.


This is incredibly unfortunate because Slide is ideal for mapping trails in the woods. Lacking something like it I end up tracing them by hand, which is a pain. (I do a lot of trail mapping in Michigan.)


I mean...sure, never blindly trust anything. But is there any proof of "malicious intent" in OSM mapping, having lead to actual issues?


I mean...sure, never blindly trust anything

This. 'death by GPS' is a thing unfortunately, and those are usually maps you actually pay for which makes it worse, in some way.


In rural areas of the US, there are some places where OSM data is hilariously bad. At one point they imported a whole bunch of TIGER data that wasn't cleaned up properly, and it's still showing roads that don't exist a decade later.

It's a good way to tell if a mapping service is using their own data or just ripping off OSM though.


The amount of detail OSM has on Artis, the Amsterdam zoo, is insane.

Also, until quite recently (and maybe even still today) Google Map was terrible at bike routes in and around Amsterdam, whereas OSM was excellent at it. I remember how Google ignored major bike routes and tried to send you over highways or footpaths instead. Nowadays they just send you along bike paths next to major car thoroughfares, instead of sending you along the major bike thoroughfare that's discouraged for cars, running parallel to it.


You can look at French hiking maps online at https://www.geoportail.gouv.fr/carte - I find it works well for planning a route and then I use OSM when I'm actually doing the hike.


For volcanoes, what do you mean guides lose potential clients because of this? Like their maps aren't as good as other guides and so people don't hire them? I'm a bit confused.


With a sufficiently accurate map, hiring a guide is unnecessary.

Somewhat similar to the situation with IT, where sufficiently documented software or hardware cuts into consultancy fees.


This is good to hear. Any idea if it's accurate with coastal data? I'm asking for a scientific project that seeks to map the locations of various seaweed strains.


There’s OpenSeaMap: https://map.openseamap.org/


These experiences are interesting, but I think Annapurna might not be the right benchmark to evaluate against. One of the least climbed and most deadly mountains on Earth seems rather far from OSM's core use case.

It is cool that OSM has the data, but I hope anyone attempting the summit is getting information more directly from other climbers.


The reference is presumably to hiking the Annapurna Circuit, not for climbing Annapurna itself.

https://en.wikipedia.org/wiki/Annapurna_Circuit


I've been looking to replace Google Maps with OSM on my smartphone. OSM itself is great, but I've been having a harder time finding a good app to use it with. Is anyone familiar with a good option? I mostly use a maps app for location searches and real-time GPS navigation, but offline maps and location sharing are nice-to-have features.

I've been using OsmAnd and it's okay. The interface is pretty good. It supports offline maps, but not location sharing. Map tiles are kind of slow to load compared to Google Maps, which is odd since the tiles have to be downloaded to local storage. The location search often includes redundant results and whatever database they use to store place data is sparse on details like pictures, hours, or reviews. I'll keep using it to stay de-platformed from Google as much as possible, but I'd be willing to pay a bit for something more responsive and polished.



I'm pretty sure OsmAnd renders all the roads, buildings etc on demand, which is why it's slow. It has to be able to rerender since it supports different styles, but it doesn't seem like it caches the results or prerenders adjacent tiles.

EDIT: performance issues seem to not be due to rendering the vector data[1].

[1] https://github.com/osmandapp/OsmAnd/issues/8370


> It supports offline maps, but not location sharing.

Are you on iOS or Android?

On Android there is the OsmAnd Online GPS Tracker [0].

[0] https://play.google.com/store/apps/details?id=net.osmand.tel...


I think Gaia GPS has a free tier? The base map is OSM. I've seen changes that I've made in OSM pulled in within a couple of days. I've gotten a lot of use out of my subscription.

I can't speak as much to the location searching capability.


Love Gaia GPS as well. Location search is hit and miss, but once you're on a trail, it's a flawless app.


Magic Earth offers a more polished user experience for driving. Compared to OsmAnd, you gain address geolocation from a third-party source (not sure which), traffic alerts, and a dashcam feature, but you lose map editing functionality and many of the non-driving features in OsmAnd. Magic Earth is closed source, although it is free of charge and has no ads. It doesn't hurt to have both Magic Earth and OsmAnd installed.

https://www.magicearth.com


I downloaded magic earth and it works surprisingly well. I am concerned about the fact that it is free, though. I wonder what's their business plan.


I'm using an app called Maps.Me, which seems to be OSM-based. It's kind of commercialized, which I don't like, but it's usable.

I'd be interested in something more FOSSy, myself.


Maps.me was on HN a couple of weeks back.

MAPS.ME is gone. We must bring it back: https://telegra.ph/What-happened-to-the-old-MAPSME-12-20

https://news.ycombinator.com/item?id=25515004



Surprisingly, the original is still being updated too: https://github.com/mapsme/omim

Is OMaps available for phones yet? All I can see is source code.


There used to be a free fork of Maps.me called simply Maps on F-Droid here: https://f-droid.org/packages/com.github.axet.maps

Very nicely designed. It seems to be gone now however, maybe someone else knows why.


There's a new one called OMaps, currently in an early stages of development.

https://github.com/omapsapp

https://matrix.to/#/#omapsapp:matrix.org?via=matrix.org&via=...


I was looking for a more lightweight FOSS map app (as opposed to OsmAnd+, which is excellent but not the most performant), as far as I can tell the Maps maintainer had a falling out with / didn't respond to the F-Droid maintainers.

That said I had to piece this together from some commit messages by the F-droid maintainers so I'd prefer a better source myself.


I recently tried the newer iOS app and it requested my location in the background somewhere around once per minute when I wasn't using the app. Didn't do anything to investigate— I just deleted the app— but it was sketchy.


I've been using Maps, from fdroid, also. It is very good, quick, and you can choose tracking on and off, etc. It states under info, that it is based on maps.me, version that's working is 9.1.8-8.


I love MapOut (iOS only). It's an outdoor-focused mapping app with easy offline downloads and really beautiful cartography.


MapOut is such a beautiful app, I'm using it very often. If you are doing any outdoor sports or even just wanting to get to know your surrounding, I heavily recommend MapOut on iOS. Worth every euro. The developers are also very responsive in case you have any questions.


I previously had good experience with Maps.me. OsmAnd is great with its myriad settings and such, but somehow is a huge battery drain on my phone. I use it just fine in the background though, because it still provides turn by turn navigation that way. I think OsmAnd doesn't use generated tiles, but rather generates the map on the fly, hence the slowness.


I like Locus maps a lot. The UI is a little busy but once you get used to it you appreciate it because it really is a work of love.


And with Locus maps you can use OpenAndroMaps [1] as a map source.

With their map styles [2] OpenAndroMaps are IMO the best hiking maps you can get. They even added very precise elevation data from LIDAR scans recently.

[1] https://www.openandromaps.org/en

[2] https://www.openandromaps.org/en/legend/elevate-mountain-hik...


Locus also have a store where they sell offline maps very similar to OpenAndroMaps for very small amounts. Not sure if they include elevation data however.


We used Locus to hike through Norway.


Other than the ones that have been mentioned already, I also use Maverick, which has a free and a paid version.

https://play.google.com/store/apps/details?id=com.codesector...

I like its ability to show the different map layers, and it has great information screens that show things like GPS coordinates and such, great for geohashing or geocaching. Also has a cool display of how many satellites are visible and whether they're GPS or GLONASS or whatever.

The track recording feature is also nice, saving as a gpx file that can be easily uploaded to the openstreetmap online iD editor with drag-and-drop.


As Google has lost it's sheen of benevolence over the last decade public knowledge projects like this become ever more appealing. Wikipedia proves they can be wildly successful all without capital investment and the perverse incentives that follow.


I used to think Google was the "good guy" by providing so many resources for free, but I've realized how they've exploited their market dominance by essentially manipulating the masses with their ad service. We need to move away from conglomerates and start using software that doesn't sell your soul to the devil.


I thought I was helping the world by fixing things on google maps. That will be a perpetual reminder for how naive I can be.


You were helping google, but you were also helping everyone who used their maps from that day into the future. Your comment gives off the air of someone upset they helped the wrong party in a zero-sum game. IMO, this is clearly a positive sum one, and you shouldn’t despair at improving little corners of the world just because actors you don’t like will also benefit.


I DO see it as zero sum. OSM and Google maps compete for a limited supply of users. It's the "wrong" team because when Google gets users, they make money. When OSM gets users, the map gets better and people are freer from abuse and manipulation.


Apple and Facebook both use OSM data, so contributing to OSM helps them make money.


don't forget Amazon (for shipping).


You need to start paying for software because if you don't then you're the product.

Payment might include leasing out your infrastructure to the project until cryptos become viable for something other than ponzi schemes.


There is a fallacy in that conclusion though.

When you pay, you may still be a product. There is no (economic) law that dictates that when you pay, your data may not be sold.

Paying customers might dissapear if they find out you make additional profit through datamining or -sales. So there is more incentive not to sell or mine data, but it is no guarantee.

The only guarantee is when technology ensures the service provider does not have the data at all. E.g. through e2e encryption.


Exactly. It is more lucrative to provide a product for free and make money mining data rather than providing a paid product, and this is even making it harder for competitors to enter the market and not die instantly. The next logical step is not paid products, but rather open source ones with distribued data storage.


Are you aware of software developed as a hobby or as part of the actual product (without making the user a product) or as a limited version demonstrating full product?


There is a reason why we don't visit surgeons who do it as a hobby.


I don't see anyone asking for professional knitters on Etsy


Wikipedia shows that average people are pretty good at assembling data. The problem is drive by contributors are not very good at developing software. Wikipedia is super basic software wise while GIS and everything around it is extremely complex.

The data in OSM is pretty good but there are no apps for it that are actually good other than leaflet JS for embedding on sites.

Have to remember that a maps app is way more than just data. Its efficient routing, its good text to speach, its pulling in data from multiple sources like the gps location of the bus I'm waiting for or the roads closed for construction currently.

You would also need a whole bunch of heuristics for the OSM dataset that the current apps don't have. OsmAnd would not be able to tell the difference between the turning lane on an intersection and an entire road that just has no name because the way the data is entered is there is just an unamed path joining the roads so the text to speach would say "bear slightly left" instead of "turn right on to foo street".


Graphhopper is an excellent open source routing engine.

I'm not sure if you are really saying that Open Source is unfit for GIS. Because there is a lot of proof it is a perfectly suited model for GIS.


Absolutely true, Graphhopper is amazing.

But I completely agree with the point of the original comment. Open Source software often comes with questionable usability and is often not well fitted to be used by the average consumer. While lots of open source applications offer the same, if not more, features than their commercial counterparts, it often feels like you need a special degree to use a simple android application or desktop app.

Without a good and usable, cross platform application to use with OSM, OSM will never be something used by the average consumer.


Then the conclusion would be "Open Source is unfitted as model to build end-user software for the average consumer".

Which I still disagree with, but that is a different discussion. It has nothing to do with Open Source being unfit as model for sofware around OSM or GIS.

Edit: to be clear: Bing Maps, Apple Maps, Facebook, WhatsApp are just some of the (average) consumer facing products that use OSM for their data. Some solely use OSM as data, others enrich and/or use it to mix with other sources. This to show examples of how OSM data is used in and by apps that you probably use. But where the data is Open Source, the app using it, is not.


> Have to remember that a maps app is way more than just data.

Agreed.

> Its efficient routing,

This is a hard one especially for long routes when calculated on a phone rather than on a server.

> its good text to speach,

Ummmm. Text-to-speech is almost always provided by the OS rather than the app.

> its pulling in data from multiple sources like the gps location of the bus I'm waiting for

There are surprisingly good apps for bus navigation using OSM in some parts of the world. Unfortunately, where I live, bus companies seem to "license" that data only to Google.

> or the roads closed for construction currently.

This is literally part of the OSM dataset.

> You would also need a whole bunch of heuristics for the OSM dataset that the current apps don't have. OsmAnd would not be able to tell the difference between the turning lane on an intersection and an entire road that just has no name because the way the data is entered is there is just an unamed path joining the roads so the text to speach would say "bear slightly left" instead of "turn right on to foo street".

Rather than "heuristics for the OSM dataset", we generally need a more complete dataset. A "turning lane on an intersection" and an "entire road that just has no name" should be mapped differently. The turning lane will probably be mapped as a "link" type road with a "destination:street" of the road that it is going to (if it is mapped as a separate road at all). The generic no-name street will be mapped as a road with "noname" (yes, there is a special tag for this). Further, in many cases, a "turning lane" will be mapped with the "turn:lanes" tag on a road leaving no room for confusion with generic unnamed roads.

I will admit that in current mapping in many places you will often get "bear slightly left" rather than "turn right on to foo street". However, this is almost always caused by a (current) lack of completeness in the data rather than a need to create some sort of set of heuristics to figure out what is "actually" going on.

I think the hardest part for a good maps app that you did not mention is search. Searching in OsmAnd (offline) is abysmal. In fact it is worse than that. It is so bad that I will switch to a web browser, search on DDG, open a link in GMaps, find the coordinates and copy them to OsmAnd for routing because I know Osmand search does not have a snowball's chance in h** of remotely finding what I want in any sane manner.


> "entire road that just has no name" should be mapped differently

https://wiki.openstreetmap.org/wiki/Key:noname (used for example by StreetComplete that I really recommend if someone has Android, wants to contribute and is not interesting in learning to use a complicated editor).


Lyft has access to some pretty good data on where the roads actually are and if there are any closures. Looks like they are already contributing back, which is awesome. Something I've never heard being done, but could be very helpful for cities, is tracking the bumpiness of roads as a metric for road quality. The city could have near real time access to information on where the potholes are. I'd think they'd be willing to pay for it.


We're working on a project in that direction [0].

The basic idea of the app is that cyclist record their bike rides through the city and accelerometer data is used to report dangerous situations (e.g. a car crossing a cycle lane without regards for the cyclists). It turns out that this accelerometer data can also be used to check the cycle path quality. Some cities respond well to the idea while others just ignore the problem

[0]: https://www.mcc.tu-berlin.de/menue/forschung/projekte/simra/...


There's a similar-ish precedent: in my city, OsmAnd can display whether streets have proper lighting. Though I guess that doesn't change as often as asphalt—but then again, I can only envy people if potholes are fixed too fast for them.


Volvo cars will notify other Volvos about slippery roads and other road hazards, and there’s work to share this data with other car manufacturers as well. https://www.cnet.com/roadshow/news/volvo-european-data-task-...

Is there something like an Open Routing project where this dynamic data would make more sense than in OSM?


I did a prototype for that idea back in the J2ME days when the first phones came out that had accelerometers. It works surprisingly well, provided you have lots of data (multiple cars driving through the same route to average out any random fluctuations due to driving style etc).

There were two issues: 1) It was way too early for this idea as the devices that had the HW for it were rather rare 2) I had no killer app that would entice people to install and run the app. There were no app-taxi companies at the time and just telling people that "please download and open that app when you are driving to help the local municipality fix the roads" wasn't really going to fly

I'd still love to work on that idea though.. one day :)


The Lyft app could measure bumpiness in great detail by using the phone's accelerometer.


There have been projects like that.

The problem is that potholes do not exists because cities do not know where they are. They exist because of lack of resources or priorities.


The wiki states that this data type is being proposed: https://wiki.openstreetmap.org/wiki/Proposed_features/Surfac...


Says the proposal is abandoned in 2011. The thing is OSM does not deal with real time data, the general rule is if the data is likely to change within the next month it doesn't go in OSM. A lot of OSM apps and servers cache the OSM dataset so anything like a road closed for one day would be stale before it even reached users.

Thats one of the big issues with OSM though, they have an attitude of "Thats not my problem, it should go in another dataset" but then no one actually has an alternative place to put it so users just miss out.


There are people working on layering the data better. I was at a talk about this at two State of the Map conferences. I don't have links or names ready, on mobile, sorry.

The idea, in my words, is to have a standard on how to link between layers: how to merge them, so to say.

That way, a routing app for a car would only need the roads and their features layers. But the search engine for tourists needs the poi layer and not all the road details.

A 'live' layer for e.g. traffic density could then be a neat commercial product for OSM consumers. But without standards to link your layer to, say, the OSM streets, it is almost impossible to offer and process such a layer.


I used OSM as turn-by turn loaded into my Garmin [1] for my three year drive around Africa, and then to cross North America six times in 2019.

I have been utterly staggered how accurate and complete it is in every conceivable scenario - dirt street villages in Sudan and Mali, capital cities in same, capital cities in South Africa and all over the US and Canada.

I also used it to drive down through Southern Europe, to cross parts of the Sahara and Namib deserts and even in the mighty Congo it had more detail and info than any other mapping source I've ever seen.

My hat is off to the OSM team.

[1] loading OSM into a Garmin is a nice easy way to get free routable maps on your dashboard for the entire world - https://www.youtube.com/watch?v=nZRQrG6bC3k


I am glad to hear that you found it accurate and complete. I know a lot of mapping of sub-Saharan Africa has been done by foreign armchair mappers working from aerial imagery for e.g. a HOT effort. I myself did a great deal of aerial mapping West African villages in preparation for a trip through the region that I ultimately had to cancel, and I was nervous that without being able to actually verify my edits firsthand, I was introducing some errors.


I used an OSM-based map on my Garmin on trips to India; did five trips to Bangalore and often played tourists on weekends; I had a particular interest in the ancient temples of Karnataka. I used a driver who knew the sites well, but not always perfectly, and the map was a huge help; coverage of roads in rural areas was quite good, not perfect but several times got us around closed roads the driver was familiar with and onto good alternate routes.

Not sure how much of the mapping was done from aerial imagery and how much was actual GPS tracking, but it was better than I expected. And this was about 5 years ago.


There's a lot of humanitarian groups contributing to OSM. Usually they get to work after natural disasters, so that emergency workers can get around easier.


On the other hand, I wonder if OSM's accuracy in Europe has taken a hit in the last year. The lockdowns have led to a wave of bankruptcies of restaurants and fitness centers in several countries, but because mappers are forced to stay home and aren't circulating in the city doing their business, they aren't noticing so much what has changed and removing those POIs from the map. Also, the extremely strict lockdowns in some countries have prevented people from going out for mapping expeditions in the surrounding regions to add missing infrastructure.


This is an issue with all map services, really. I see the same thing happening on Google and Apple Maps, lots of places that are permanently closed haven't been updated yet.


Apple, Amazon, Facebook and MS have been lately contributing a lot to OSM, so what you see on Apple Maps might very well be OSM data.

https://joemorrison.medium.com/openstreetmap-is-having-a-mom...


Apple Maps has been OpenStreetMap for years.


No, Apple is USING OpenStreetMap which is a different thing - they pulled the data but haven't contributed much back as they're building their own proprietary backend.


Apple does contribute to OSM quite a bit, see around 4:00 in https://2020.stateofthemap.org/sessions/SPRQVZ/ for details of corporate contributions: "Apple is currently doing the most work on the map".

Here is the base repository of their contributions on GitHub: https://github.com/osmlab/appledata

And if I see this correctly, the #adt hashtag (for Apple Data Team) contains already +100000 changesets for this year alone: https://osmcha.org/?filters=%7B%22comment%22%3A%5B%7B%22labe...


They also pay Kaart to do it, so most if not all of the Kaart contributions are paid for by Apple.

https://wiki.openstreetmap.org/wiki/Apple#Contributions_thro...


Apple's old map was OpenStreetMap plus other sources but its new ones are proprietary. But its new ones are only available in the US, UK, Canada and Ireland.


I believe the old map was mostly TomTom (TeleAtlas) with a little bit of other sources sprinkled in.


Yes, more TomTom that openstreetmap, that was clearly show by the dataquality. Much less walk and vike information. If you don't drive a car all my sampling say OSM is way better.


Interestingly with almost nothing to do but walks around the city, Covid is the first time I started actively contributing to OSM. Although StreetComplete [1] definitely helped there, albeit just as a gateway for actually updated POIs, which still required the web interface.

[1] https://github.com/westnordost/StreetComplete/


> gateway for actually updated POIs, which still required the web interface

Vespucci app can be used to do it directly on phone.

Vespucci vs StreetComplete are quite interesting, in many aspects they are complete opposites.


Also, OSM Go![1] allows you to edit/delete any object on the map :)

[1] Available as PWA and Android app: https://wiki.openstreetmap.org/wiki/Osm_Go!


I do not think so. Well restaurant may be a little behind.

I try to maintain all restaurants etc in Denmark by comparing OSM POIs to public health inspection reports. see: https://digitalfrihed.dk/restaurants/all.html

So today there are 243 restaurants, cafes and fast food places in OSM that do not have a valid inspection report (the pink ones on my map). I think that many are just closed temporarily because of Covid and therefore have not been inspected and the inspection report just timed out. Probably some that have an inspection report will never open again, but I will soon find out. Many of them are cafeterias for employees in larger companies and will likely open again.

Many restaurants are open for takeaway, also many that did not do takeaway before the lockdown.

I have lately been doing speedlimits, mostly from Mapillary. But I have also done some surveying on bicycle.


If you notice this you can add a note to the map. This does not require an account or logging in and it will be visible to mappers.


Google Maps is funded by corporations who want to mine your data and sell you stuff, whether you need it or not.

OpenStreetMaps is funded by... well, actually, it's complicated. There are lots of individual donations, but there is also money by Amazon, Facebook and Apple. And some other corporations.

Regardless - consider making a donation:

https://donate.openstreetmap.org/

I would also entreat you to ask OSM to get off corporate funding. And of course, if you can, contribute map data about where you live.


I'd be really interested in a comparison to other mapping providers. 95% doesn't sound that great in the abstract, but perhaps that's better than the competition?


What is the best-in-class iOS app using OpenStreetMap? For iPhone, MagicEarth looks extremely good (https://www.magicearth.com/), but I'm wondering if people have other recommendations...


Arguably, “Windy Maps” is up there:

https://apps.apple.com/us/app/windy-maps/id1201228182

Same devs as the stunning https://www.windy.com/ weather site which also has an app:

https://apps.apple.com/us/app/windy-com/id1161387262

If you're more about hiking, maybe “Gaia GPS Hiking, Offroad Maps”:

https://apps.apple.com/us/app/gaia-gps-hiking-offroad-maps/i...


I wonder what's the best-in-class iOS app using OpenStreetMap in terms of privacy.

Functioning offline without sharing your location (or data derived from it) with some big-tech company selling it for $.


OsmAnd (https://apps.apple.com/ch/app/osmand-maps-travel-navigate/id...) does only collect usage data and diagnostics according to the App Store.


Mapy.cz (https://apps.apple.com/ch/app/mapy-cz/id411411020?l=en) does collect some data (see App Privacy in the App Store), but is not linked to „your identity“.


Magic Earth (https://apps.apple.com/ch/app/magic-earth-navigation-maps/id...) does not collect any data according to the App Store.


Magic Earth seems the best option of the apps mentioned, as they collect no data at all, indeed.

Thank you!


Have you tried OsmAnd?

Their iOS offering has gotten fairly good recently. It does cost a small amount of money to get the full version. The free version is complete (it is only limited in how many regions you can download) and more than useful for trying it out.


I have been using maps.me for about 5 years. It is becoming more and more commercialized but it's fast and has a simple interface.


Maps.me was recently purchased by a weird payments company and replaced with a generic reference MapBox OSM app.

This might have been reverted later but it was big on HN a few weeks ago.


It was reverted, but a fork started nevertheless, OMaps, currently in an early stages of development.

https://github.com/omapsapp

Matrix chat: https://matrix.to/#/#omapsapp:matrix.org?via=matrix.org&via=...


Now if only there was a sensible way to self-host OSM data. From what I've found you either have to build a dedicated server to continuously building the tiles (which takes a ton of storage and processing power) or pay for it at an insane monthly price. I wish OSM would directly make their tiles available to mirror. (Maybe they do, but I haven't found it yet)


I just looked it up and apparently, all tiles would take up 54 TB[0], and even just the important tiles are around 1TB. I now understand why just downloading them isn't an option.

[0] https://wiki.openstreetmap.org/wiki/Tile_disk_usage


It specifically says: OSM does NOT pre-render every tile. Pre-rendering all tiles would use around 54 TB of storage

that 54TB figure is only if you pre-render into raster every zoom level of every tile for the whole world. a more typical self hosted openstreetmap renders the tiles on demand as you scroll or move the map.

Disk space usage is MUCH less, such that my test/development setup fits well on a virtual machine with 300GB of disk space on a 12TB hard drive that cost $145. By this measurement I believe it's using just under 4 dollars of disk space, 8 dollars if you count its duplicate where it gets backed up to.


Maybe the tileset could be distributed via torrent, like how they distribute the main dataset.


that would be a bad idea since the tiles are ever changing. the master copy of the data is in a vector format. the idea is you want to run the (free) software engine that turns the vector data into various zoom levels of raster tiles, on demand. the vector data is really quite small, an entire large-population US state with many complex cities is under a few hundred megabytes.


Well their master data, planet.osm, is ever changing also, and they have a good time distributing that via torrent.


I built a tool for downloading tiles last week - but with a warning in the README to make sure you are following the OpenStreetMap tile downloading policies.

https://simonwillison.net/2021/Feb/4/datasette-tiles/


That actually looks pretty close to what I've been looking for. I'll have to look into it more when I get time.


You can download tiles directly from openstreetmap.org, although bulk requests are obviously discouraged.

Running your own tile server isn't too bad. The tiles are cached so that not much processing power is needed after a while of usage. Yes it requires storage but so would any self-hosting solution.


It really depends on your requirements, but if you only need one city then you're hardware requirements are in the raspberry pi territory: https://switch2osm.org/serving-tiles/

The site linked already provides docker images, I'm tempted to build a pi one.


https://hub.docker.com/r/maptiler/tileserver-gl

hosing the entire planet is not sensible, but you can safely host a region/city/small country.

that said im not in love with maptiler and trying out a bunch of other tile servers with limited sucess.


someone was kind enough to upload their recent (oct 2020) build of OSM vector tiles to the internet archive: https://archive.org/details/osm-vector-mbtiles


You nailed it wow. Didn’t even think to check there.


https://openmaptiles.org/ might be an option, although it doesn't include everything Open Street Map itself includes.


Can you explain what is not included? Do you mean that they don't render some things that are present in OSM? You refine and set up the tiles from openmaptiles to your liking quite a bit...


There are features in OSM like power lines, pipelines, ski runs, fences, cliffs, hedges, traffic lights, marker buoys, ferry routes, bus routes etc. These aren't in the OMT vector tiles, to keep them small. (Some of them might be included since I last looked.)

The features that are there are included from a particular zolm level. If you want to make a global railway map - so you will show major railways at zoom 2, say - you would need to hack around, since the first zoom the railways appear on is 4 (or whatever).

What you can easily customise is showing those railways at zoom 4+ (or not) and the style of the lines.


This is very correct, but for 100k requests a month it's free to use their API (https://www.maptiler.com/cloud/) [1].

If you want some of the features you mention rendered in your application you probably have to "Generate your own vector tiles from selected OpenStreetMap tags" as they also mention on their website.

[1] I'm not even a customer of them, just browsing their website.


A friend and i where once on a mission, making bike-tours in the swiss mountains every weekend, and map every forgotten water-fountain in villages we crossed, great fun!


I have been considering looking at OSM as "yet another data source" for my maps, but overall I run into the highly abstract issue of comparison: how do I know that two segments (one from say OSM, one from perhaps the county) reference the same stretch of street for comparison?

Even the slightest nudge of a single coordinate point on the line would mean that they were not geometrically identical, so I would not know to compare their attributes. Names are almost laughably bad for this purpose, for having multiple forms ("N SECOND ST" vs "NORTH 2ND STREET") and for being terrifying vague (so many, many Main Streets). Worse yet, a given segment might be split at a point not an intersection (speed change occurs here, municipality change, directionality change, road type change, etc) so I cannot even rely on those. And it is entirely possible for different maps to include a driveway, splitting a street at that point, while another does not, leaving it continuous.

Large-scale automated comparison of this kind of thing is challenging.


One of the under-appreciated advantages OSM has here is that they have a fairly well enforced convention of using fully expanded abbreviations in addresses and street names.

It's _much_ easier for automated comparison to replace `street` with `st` than it is to guess which one of many possible expansions to replace `st` with (is it saint? street? sankt? strasse?).

This isn't foolproof but much of the "authoritative" data out there has no such convention, making analysis much more dangerous.


Oh, cramming the names into a common format is something I have under my belt, but there are additional wrinkles and, worse yet, local bizarreness (aka "historical convention"). The names are the least of it, really, even if they are ... odd.


The official USPS abbreviation list has some pretty crazy ones:

https://pe.usps.com/text/pub28/28apc_002.htm


https://sharedstreets.io/ may help!

> SharedStreets Referencing System is a global, non-proprietary system for describing streets. The Referencing System is the foundation of the SharedStreets toolkit and it is used to connect a wide range of street-linked data. The Referencing System connects street data from private companies and cities. Referenced data can then be transferred between different maps seamlessly.

I think it can align maps that have similar streets that may not align perfectly coordinate wise.


> Names are almost laughably bad for this purpose, for having multiple forms ("N SECOND ST" vs "NORTH 2ND STREET") and for being terrifying vague (so many, many Main Streets).

libpostal (open source, open data location parser) might help with this. Between using it to parse the name, and supplementing this with the city name (using reverse geocode with imprecise coordinates), you might be able to get something useful.


Lyft study shows crucial OpenStreetMap road attributes are fresh and high quality in 30 North American cities, as compared to groundtruth. Blog post and paper detail the process, methodology, and results.


"Road directionality" being only 98.9% accurate seems to be a huge problem for a navigation app!


Road directionality is variable over time. I'm not sure at what rate it changes but I bet it's a lot more than 1.1% per year.


Hmm, I’ve seen maybe a handful of two-way streets change to one-way over my entire life, and have never seen a one-way street change direction. If ~1% of streets changed direction every year, there’d be mass confusion and frustration.


One major thoroughfare in London changed (if I remember right) from one-way in one direction, to one-way in the other direction, to two-way, in the space of a couple of years.

People got used to it very quickly, to the extent that they would not remember the change ever happened.

If you have a broad enough roads to have two-way streets everywhere, there's little motivation for it to change. [Edited to note: at the level of individual cities, OSM's directionality was often estimated to be 100% correct]

But, with some narrower one-way streets, there becomes a clear motivation to tweak things to improve traffic flows (and also tweak the surrounding two-way streets).


Mapping OSM in NYC, I've run into about a dozen one-way streets that have changed direction at some point, sometimes for just a block or two.


There’s apparently a trend of converting one-way streets to two-way; cities have realized that one-way streets are a little too efficient, they increase bandwidth but often cities want you to stop and shop.

Indianapolis has been moving that direction, at least, but I don’t know how widespread it is.


Im yet to read TFA, but I could foresee there being plenty of one-way private drives and parking lots - if not included in the study, I tend to agree with you


When I worked in HFC/Fiber Design (and automation via code), we tried looking into OSM as a replacement for our workflows that used google earth and hack on top of hack (Google earth kept removing parts of the API that were useful for automation/semi-automation, at least I got to learn lots about P/Invoke I guess?)

Back then, it unfortunately wasn't usable for us, but I kept the research bits aside just in case when I left the job in 2013.

When I spoke with someone still at that company a few months ago, I learned that over the years OSM had evolved/matured enough that they were using it far more than google earth.

I miss GIS/Mapping. It's great to see it is still a collaborative community. :)


Unfortunately, working in the Commercial Real Estate world, OSM is no where close when doing a forward geocode... around the block is not close enough.

Only ones accurate enough are Google (won't use) and Mapbox.


I thought mapbox used openstreetmap data?


Not exclusively. They probably dump OpenAddresses data (or something similar) into their geocoder.

https://openaddresses.io/


Hi, founder of Geocode Earth (https://geocode.earth), another geocoding service that _does_ use OpenAddresses and only other open data sets.

Mapbox purchases data from, I believe, TomTom, and other proprietary sources. That's why they have cheap(er) pricing if you want to get results without the legal ability to store them, and much more expensive pricing if you do want to store it. I'm sure negotiating that arrangement with the proprietary address data vendors was...fun.


A friend of mine runs opencage, which is a geocoder based on open streetmaps and a few other data sets. They got a lot more business when Google raised their prices a while back. Anyway, go give them some love if you need a geocoder because they are an awesome small company with a great service. Their freemium layer is also pretty generous.

Nominatim is the defactor open streetmaps geocoder but its weakness is that it relies on only that; so it is only as good as OSM is.

Geocoding is a surprisingly hard problem to solve with and without open data. Picking apart text strings into unambiguous addresses is just super hard with all the conventions for this around the world. One area open streetmaps is still inconsistent is house numbers and postal codes. Getting solid open data for that is just hard in some places.


That seems like it could just be an up charge.

(or a discount for the ephemeral results if you want to look at it that way)


Yup, we pay them quite a bit to be able to store it.


The problem I have with OpenStreetMap is the license, it seems you can use an OSM derived base map and then layer anything else on top without problems, but you cannot process the data or mix in proprietary data without releasing the algorithm or the final dataset. So any competitive advantage derived from processing the data has to be immediately made open.

So you see a lot of use of OSM as a base map, but very little use as a mapping database for commercial applications. I see why they did this, to prevent companies taking their data to bootstrap a closed competitor, but the consequence is that the data is locked in and is getting much less use than you’d hope as a database.


Can you elaborate where license would be a problem?

Keep in mind this:

https://wiki.osmfoundation.org/wiki/Licence/Community_Guidel...

Which eases the issue a little bit. For instance, MS uses OSM buildings for Bing Maps. Or Apple uses OSM for road data in selected countries (Ukraine?)


I’ll give a contrived example, say you’re a property startup, and you want to produce a walkability rating for each house you sell. You might want to do a calculation on OSM for restaurants within 5 minutes walk, maybe a chef rating for whether there’s a greengrocer, fishmonger, delicatessen and butcher within walking distance, or you want to take into account a walk back from the train station and the opening times to see whether the shops are still open after someone comes back from their commute. This is the kind of thing which would be interesting for big cities with a lot of commuting by public transport.

My understanding is if you did this with OSM data you’d either have to make the algorithm that did the calculation public, or the output dataset (locations as keys and the final ratings for each as values).

OSM has exceptions for horizontal slicing (layer separate data on top of an OSM basemap), for vertical slicing (union OSM data with separate data) and for trivial manipulations, but it seems it doesn’t if you do this kind of non-trivial processing of OSM data.

So that means that it’s almost impossible for a company to find an interesting way of processing the data which can give them an advantage over a direct competitor, and there’s no real incentive to do so.

This is my reading of the terms from a few months back, hopefully I’m wrong about it. But I suspect this is why it’s rare to see the data being used like this.


Well, companies do offer isochrone APIs. That, or graphical maps, are Produced Works under ODbL and only need attribution. Actually enabling that was the point of licence switch in the fall of 2012 which while mostly successful wasn't without pain points (e.g. in Australia key contributors rejected new Contributor Terms and their edits had to be redacted).


Thanks, that’s interesting, it doesn’t match with my reading of the guidelines, again, glad if I’m reading them wrong and you can correct me.

https://wiki.osmfoundation.org/wiki/Licence/Community_Guidel...

It says clearly there that the underlying database for a produced work has to be published. So can a competitor not just request the derived database directly from a company selling the API?

Also, “ If the published result of your project is intended for the extraction of the original data, then it is a database and not a Produced Work” - does that not exactly describe an API? It talks about mugs, data visualisations, map images, physical maps and so on as produced works. I thought this meant you could protect (i.e. not release) essentially a styling of map data, but not the map data itself, which must be released.


One thing that someone might consider is working with Data Portability laws to allow for export of user contributed data to corporate map platforms.

I'd like to add my Google Maps contributions to OSM

I'm a Level 7 local guide there and was an active user of MapMaker back in the day..


Which sort of contributions would you like to copy from Google Maps to OSM?

If you personally surveyed these places you can go to OpenStreetMap.org, sign in and then you can add your changes - OSM may already have the info so a straight import wouldn't be possible.

Make sure not to copy any of Google's details to OSM that you didn't personally survey as that would violate copyright.

Please reach out to me if you'd like a hand with getting started mapping on OSM :)


Wrong answer.. This effort does not scale. You want to make it such that facts added by anyone can be easily imported with a simple opt-in.

GDPR data portability could be expanded to include defined formats for export so that import can be done in an automated fashion.


> Wrong answer..

That comment doesn't provide any value.

> You want to make it such that facts added by anyone can be easily imported with a simple opt-in.

OpenStreetMap is much larger than just me or one company.

> GDPR data portability could be expanded to include defined formats for export so that import can be done in an automated fashion.

Imports are not an easy problem - how do you deal with conflicts? How to verify which data is more accurate?

Imports are supported, and there is a detailed process for them to make sure OpenStreetMap stays well maintained: https://wiki.openstreetmap.org/wiki/Import/Guidelines


Google map reviews is the only thing keeping me from OSM based maps. I love to explore, take photos and review places. OsmAnd has some review functionality but it's very barebones unfortunately.


I think this adds to the case for Rousseau's Theory of human nature... https://en.wikipedia.org/wiki/Jean-Jacques_Rousseau


Except my street. No matter how many times I email Lyft support, it always sends them the wrong place. Most Lyft drivers don't read the directions on pickup either explaining the problem.

/sigh #FirstWorldProblems


No bothersome problem is too small! If you'd like to share, we would be eager to look into this. Reach us through the app or at osm-questions [a] lyft.com.


The real problem of GP is with the drivers not reading the note. Each system should have a manual override, Lyft does seem to have one, but the human element failed.


Have you tried fixing the map yourself? It's easy, just like Wikipedia.


That is a nice feature of OSM, and I've done this a few times.

But I wonder if there is any propagation delay from OSM to Lyft. When I made some changes to OSM data, I expected the Strava app to pick up the changes for route creation, but they don't receive updates constantly. Instead, it took several months until the changes took effect on Strava.


As an OSM mapper I would not say that debugging and fixing address issue is easy...


It's maybe a little snarky, but the most common issue is relatively straightforward to fix, just add the missing address.


If it is OSM data issue then you can ask for help on one of https://wiki.openstreetmap.org/wiki/Contact_channels

Or me directly - though either would likely require revealing your address and location to diagnose what is going wrong.


Not just in US.. Here in Spain too. It's really great. Especially niche things like walking tracks are super accurate whereas on Google Maps I'm walking in one big grey square :D


I don't know which mapping options Lyft/Uber/etc use, but in my neighborhood, Google Maps has given street names to the allies behind people's houses. You can tell the drivers that have a clue, and will realize they should not turn down an alley. Drivers that are less cognizant of how the world works, follow the instructions dutifully, and waste time trying to figure out why everyone is confused.


OSM is usually quite good. My colleagues simulated nationwide travel using OSM roads:

https://www.researchgate.net/publication/337332041_Agent-bas...


I've just checked (some years have passes since the last time) and it seems improved a lot for Europe too.

I personally am not really interested in geospatial precision, I mostly use public transport so I need all the bus/tram/etc stops placed and named accurately.


Can anyone recommend an OSM based turn-by-turn, iOS app that is CarPlay compatible?



Mapping should be something done by the government. They know the most (construction plans etc) and they need to do it anyway. The data should then be available to anyone who needs it.


Dutch government does this partially. That's how OSM got some great address location and house outline data in the Netherlands :)

OSM worldwide is a mix of open datasets and original data. We'd not be as complete if one of those two were missing.


I took my honeymoon trip to New Zealand almost a decade ago using a Garmin gps and a map I made myself from the OSM database flashed on an SD card. It worked great.


I used it while backpacking across Europe back in 2019. Very useful when I was offline.

But a lot of building numbers were missed.

So I had to use Google Maps when I had Internet connection.


Off topic - is there a way to get an index of all businesses in OSM (restaurants, hotels, shops, etc)? I need the data for training a NLP model.


Yeah, http://overpass-turbo.eu/

use the wizard.


What small, growing companies are in the general field of mapping?

Are there companies that fund employees to work on Open Street Maps?


Not only small companies do, but big ones, too: https://2020.stateofthemap.org/sessions/SPRQVZ/


What is the barrier to adding more data? Do they need funds to purchase it from satellite providers?


Satellite probably won't give you highly detailed models. There'd be some ML going on to detect features. Plus, your image will lack ortho-rectification. Imaging is still primarily a ground-based/very low orbit operation with photogrammetry and lidar.


knowing what kind of structure a rectangle on your satellite image is, is also a big problem. A shed or car looks roughly the same. So I would say it's more about getting better Satellite and plain images + combining that with knowledge about the area and government.


You can donate at https://donate.openstreetmap.org/ - though money is not used for buying satellite imagery, but rather for server upkeep, maintaining core software etc.

You can also contribute directly - there is edit button on https://openstreetmap.org

StreetComplete may be useful if you have and Android phone and you are not interested in learning how to contribute (registering for OSM account is the worst step).

Vespucci (for Android), Go Map!! (for Apple devices), JOSM (Java standalone program) are full-scale editors that I also recommend.


OSM Foundation does not buy imagery. The community just relies on what is made available to them: Bing, Esri, Maxar, Mapbox all make their imagery available worldwide, on top of that many state or local governments have freed up their imagery.

As for street-level imagery there's Mapillary and KartaView, both relying on volunteers with smartphones.

The most important stuff, and the real asset of OSM, is what can't be added with aerial imagery: POIs aong with their attributes, road surface, speed limits, turn restrictions, paths in forests covered by trees and so on.


they need more contributors...


My experience with OSM is limited I only contributed a few times to the data in a local city in the Netherlands. I also used the xml data export only a few times for small experiments a long time ago. but from my experience is see the following:

* If many people enjoying or live somewhere they will update it very fast and accurate. Way beyond google and apple will ever have. (If a bench is placed somewhere else on a school campus then it will most likely be update within a few days, from my experience at least)

* Small structures are detectable on air and satellite images but it is really hard to know what kind of structure it is.

* Data from the dutch local governments do, from what I see on OSM, often contain good updates on new buildings. But if something old is changed. (Like a temporary school building is removed then that change is not reflected by a data update from the goverment, and it is often removed by a user after a while) So getting better data from the governments is probably a thing.

* A lot of layers are build on top of OSM. For example if a building is easy to use as a person who uses a wheelchair and if there toilet is wheelchair friendly. My ex-girlfriend was in a wheelchair so I used a app for that a lot. But she never used It. They app was very slow and had a strange UI. So changing the data of custom layer on top of OSM with better tools is hopefully going to be a thing.

* Users use it mostly because they like to contribute. Google has the power to send many people notification questions about places where they have been. OSM does not have this power.

* I would dare to say that OSM is already way more precise then google or apple maps. But users do not always care about that. The big driver for me to keep using google maps is that I often just want to go somewhere, and to go somewhere means that I want to travel by using the train or bus (after covid). In the case i'm in the city I want to know how long it takes to go somewhere by bike or by foot. In the case i'm going to a new city I often want to see how something looks and Google streetview or general images are better then a map. Google prioritizes all that information above having a more accurate map.

So no I do not think it's about getting more people involved. I think it's more about what kind of user does the platform want to have.


Is this because Google has increased the api pricing and people have switched to OSM?


If at all, then only as an indirect effect. I haven't really noticed an uptick in contributors since then. There has been more publicity and more people using it (and abusing the community-run openstreetmap.org tile servers for commercial purposes²).

² It's fine to use OSM for commercial purposes, but hosting the tiles costs bandwidth and computational power. Embedding the community servers in your taxi tracking product and refreshing the page and map background every second is just a dick move.


And I helped with that. :)


Me too :) feels great to see the changes online, and then as they propagate to different providers.


Wish Google MAPS and OSM share data with each other;


I wish that too! The OSM data license is similar to GPL in that it requires, when mixed, the resulting derived work to still be shared in an open license. Having Google's data to complement OSM's would be a huge benefit for researchers that can use geographical data for all sorts of purposes, users of course, everyone really. As someone who contributes to OpenStreetMap and avoids Google Maps (for privacy reasons, not map reasons), I have nothing but respect for how much data they gathered. I'm just sad they keep it locked up.


this is an excellent design interview question


OSM is lovely, except for the UI. It is way over complicated.


The thing is it doesn’t really have a UI; it’s just a database that people can do what they want with.

The interfaces that you see—even on the OSM website—are a myriad of separate projects.


Which UI do you mean? The project is much larger than the frontend at OpenStreetMap.org - I agree that could do with some work :)


Apparently, I'm not in the 30 cities that are accurate.

Used Magic Earth on my phone which is based on OSM. The very first address I input was off by about a mile. Didn't exactly inspire a lot of confidence.


To be a little pedantic, converting an address to a location is more of a function of a geocoder, not a function of the map. Software could have a very accurate map, while at the same time using a poor geocoder, so addresses are located incorrectly.


Not even pedantic, I think it's an important realization. Before knowing OSM, I though of Google Maps as a single service. Not realizing at all that mapping, addresses, POI entries, directions, navigation, etc are huge, separate problem domains. I think parent has the same perception.


Address search (geocoding) isn’t what this article is about. OSM by common agreement has a long way to go on that.


Right, but if you're Lyft you have to find people and locations from their addresses.


That is actually a good point, what happens is that many streets does not have any address points, and your search puts the marker on an arbitrary end of the street, so a mile off sounds about right. Address points are really cumbersome to crowd source, you really need either open data or buy a geocoder. Not easy to do if need to allow everyone to download the data and do what ever they want with it.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: