How Mapbox Is Winning Over Developers to Challenge Google's Mapping Dominance | Hacker News Reader

How Mapbox Is Winning Over Developers to Challenge Google's Mapping Dominance

forbes.com

336 pointscoloneltcb111 comments

There are more alternatives to Google & Mapbox that mostly use OpenStreetMap as a backing data source:

    https://openmaptiles.org (only map tiles)
    https://stadiamaps.com (map tiles, static maps & routing)
    https://www.thunderforest.com (only maps)
    https://geocoder.opencagedata.com (only geocoding)
    https://geocode.earth (only geocoding)

A laundry listing: https://switch2osm.org/providers/

Tile providers are very thankful for the excellent rendering libraries built by Mapbox, Mapzen, and the Leaflet project. Most maps either use Leaflet.js or MapBox GL JS (open source) for their rendering agent, now that Mapzen is defunct (although their rendering library, tangrams, is now open source).

(I am a co-founder of Stadia Maps.)

Hmmm I evaluated a lot of those (mainly OSM-based) geocoders recently. Unfortunately the quality of results is orders of magnitude worse than Google. The only ones that even came close were Mapbox, Mapquest and Here.

Hi,

co-founder of OpenCage here.

It depends heavily on what your use case is: forward or reverse, which part of the world, what level of granularity, etc. Google offers a great geocoding service, no doubt. But you also have to pay their price, and obey their terms and conditions. Their great service is overkill for many very common use cases, at least those that we see from our customers.

Anyone who wants to try us out, our demo page is here: https://geocoder.opencagedata.com/demo

I’m curious, how close? Would you consider mapbox instead of google maps? In what cases is it not close enough?

Do you know of any tile service offering colored, shaded elevation reliefs? I've been trying to find something for our meteorologists to use and the best I've found is https://maps-for-free.com

> Do you know of any tile service offering colored, shaded elevation reliefs?

Does that imply variation in coloration based on elevation? (Just making sure I understand your question.) We have limited support for basic hillshading and are actively looking into contour lines. Coloration for various elevations is something we do not currently offer (nor am I aware of any other providers), but it is something we would be willing to investigate.

Depending on what precise application you need, this may be possible one the web using our normal vector maps and contour lines.

Yes, color for elevation and shading for slope.

Another use case that I've been searching for is a tile set not of images, but of just elevation data. So that I could implement something similar to google earth's lat/lng/elevation info wherever your mouse cursor is.

You don't need a web service for that. The data is free, search for "SRTM" and download the continents you need.

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

Sure but I have no expertise on how to take those files and turn them into a tiles that correspond with the same basemap image tiles. Also ideally the service would incorporate more than just SRTM, including for example the USGS high res DEM where available.

Same boat, but does this help at all? https://wiki.openstreetmap.org/wiki/SRTM

This is just what the comments you replied to were discussing

You’re right, but seems to include implementations the user may be able to leverage? Specifically "In Use" and "Data in OSM format (XML)"

> how to take those files and turn them into a tiles that correspond with the same basemap image tiles.

The format is very simple to make sense of. Each file is an array of 16 bit integer values in meters, and it maps 1x1 degrees areas. Each file contains 1201x1201 or 3601x3601 samples depending on resolution. The first/last row/columns of adjacent files overlap that’s why non-round numbers.

Both SRTM and tiles use Mercator projection, i.e. mapping them together is very simple, too. In the embedded software I’ve developed, I’ve layered SRTM-derived data on top of OSM tiles, worked quite well and wasn’t too hard to implement. Even though due to very slow target hardware (500MHz CPU, 256MB RAM, slow flash storage), I had to invent more optimized format for SRTM data (dense array of 32x32 tiles taking exactly 2kb/each, aligned by disk sector), and implement an offline converter.

> including for example the USGS high res DEM where available

I haven’t personally compared them, but before picking SRTM for that embedded GIS project I’ve read about them, and concluded SRTM is more reliable. Also the resolution is more or less the same, SRTM offers 25m horizontal resolution, that’s pretty high, IMO.

Have you evaluated SRTM (or some patched up derivative) for your latter use case?

Yeah, its unfortunately missing a lot of alaska. But the main problem is I have no idea how to reproject it into web-mercator and generate usable tiles from it.

https://github.com/andrewharvey/srtm-stylesheets might help

(Full disclose I'm the author)

This looks awesome. Using those scripts I think I could actually do this myself.

We don’t have a current offering, but if you want, please email me at luke@stadiamaps.com. We’re actively investigating ways to integrate more datasets (such as elevation) and we would be happy to learn more about what you need.

If we can’t offer something, we’ll at least be able to point you in the right direction!

Give http://services.imagico.de/relief.php?lang=en a try.

Which zoom levels do you need? What coverage?

I recently wrote a guide (https://blog.mapbox.com/styling-mapbox-terrain-rgb-c75d1cd71...) about extruding the contour lines within the Mapbox Terrain vector tiles (https://www.mapbox.com/vector-tiles/mapbox-terrain/#contour) that could be helpful to you as soon as I update it to match our recently re-released Studio product.

Check out https://www.mapbox.com/help/access-elevation-data/ for more background and let us know @Mapbox if you have questions!

Note - I work at Mapbox

Try Thunderforest - relief is one of their specialities.

Is there anyone providing free high-res satellite tiles?

Simple answer is no. But it depends on what you consider high-res, ESA does provide live feed out of they one satelite snd you might use those for constructing your tiles, still it is nowhere close to off shelf solution.

Depends on what your definition of "free" is - beer or liberty? There are some free-as-in-beer satellite providers, but none that are free-for-anything - e.g. ESRI is freely available for OpenStreetMap edits, but probably not for commercial use: https://wiki.openstreetmap.org/wiki/Esri

The team behind OpenMapTiles started to work on global satellite tiles too.

See https://openmaptiles.com/satellite/. Next step is merging in very detailed aerial photos for US and EU - often with 1m or 25cm/pixel resolution.

It is going to be available for both self-hosting and as part of the hosting platform https://tilehosting.com/

Just wanted to add another geocoding provider to the list https://geocod.io - founded about 4 years ago based on frustrations with Google’s terms and pricing.

More crucially, they are winning over executives of medium-to-large internet businesses with superior sales and marketing.

Individual engineers at these places do not have the power to make decisions about which map software to use, which are often six- or seven-figure contracts. Mapbox offers a service that is comparable to Google Maps in terms of price, quality and flexibility. From a technical standpoint, there are pros and cons to both services; Mapbox's toolkit is more modern than Google, which means documentation is a pro for Mapbox but legacy browser support is a con (though it's just a minor annoyance).

Where they far outstrip Google is in sales and customer support. Google Maps won't bat an eye for you, even if you're paying them millions of dollars a year. The Mapbox team, in contrast, will devote developer time to answering your questions, respond to GitHub issues for you in a timely manner, and even create new plugins and repos if you have a legitimate need. That is their big advantage and that is what is driving adoption.

Source: I was employed at a medium-sized internet publisher, and tasked with evaluating Mapbox to replace Google Maps, was a primary developer working on the switch-over, and even wrote a blog post about the experience [0]. They even invited me to their office to work with them for a day (which, regrettably, I wasn't able to follow up on).

[0]: https://blog.rentpathcode.com/webpack-progressive-enhancemen...

What are your main use cases?

Showing apartment listings and houses on desktop and mobile web.

Papa johns was (is?) paying for mapquest purely for the privilege of calling up the noc every few days when their servers crashed. At least it felt like that when I worked in the noc.

what is noc?

Noc stands for network operations center

Out of curiosity, did you have a paid support plan for Google Maps, or what are you basing your impressions of their support on?

Also, part of the Google Maps Platform launch is that support (as in, being able to file tickets and get answers) will be extended to all users: https://mapsplatform.googleblog.com/2018/05/introducing-goog...

Disclaimer: I used to work on Maps Support at Google.

I'm just relaying the reason the senior director gave us for the switch. I evaluated Mapbox, but more from the angle of can we use this rather than should we. It was really as if the decision had been made already.

We did have a paid package with Google Maps. Good to hear that Maps is amping up their support.

Last I checked, MapBox's WebGL based vector tile renderer was a cut above the rest. For simple mapping, Google's stuff may cut it, but MapBox has the ability to draw complex vector polygons over a wide area that render/zoom/scroll fast and provide a nice experience. This is pretty important if you want to provide fine grained analysis of geographic areas.

If you wanted this on GMaps, you were stuck rendering all your vectors to image tiles, hugely increasing the size. MapBox's vector support made this very easy! It may be that Google/free options have caught up in this space, but I haven't re-evaluated in a few years.

I think MapBox could also be a winner in the GIS space as the GIS options have not made the most graceful move to the web.

Mapzen, a now sadly defunct mapping startup, also had an awesome (if I say so myself - I used to work there, but not on that team) vector tile renderer for browser and mobile. Check it out at https://github.com/tangrams

Also, there was a great interactive style editor (https://github.com/tangrams/tangram-play), although it seems the demo site is broken now :-(

RIP Mapzen. I loved how much much you all were contributing to the open source web space.

Any chance mapzen would open source the code used to make metro extracts? The formats your team had to offer were so much nicer than working with raw OSM data

I'm glad you found them useful! The code which made all the metro extracts was embedded as a Chef recipe, although I'm sure you could just extract the bits which do stuff from the Chef wrappers: https://github.com/mapzen/chef-metroextractor

Cool! Thanks you!!

> MapBox's WebGL based vector tile renderer was a cut above the rest.

It really is. I've done a lot of work in the web with various rendering engines, and it's hard to beat the quality of MapBox GL.

Since Google Maps jacked up their prices a few days ago, I've been exploring alternatives, including Mapbox. Unfortunately Mapbox is also very steep for a solo developer developer who wants to build a side project - at least if it's considered 'commercial'. $500 / month for only 250 users, and no way to plan beyond that without inquiring with them about an "enterprise" plan.

Also, if anyone from Mapbox is reading this, you need to be more clear about what is considered "commercial". At the top of the pricing page you define a commercial app as:

  Paid web app or website (fee or subscription)
  Private web app or website with restricted access
  Asset tracking app or website to monitor people or things

But then lower down on the page, you seem to indicate a more permissive definition:

  If your app is for internal business use by employees,
  or if it is not accessible or functional for members
  of the general public then we consider that a
  Commercial app.

Even with these two definitions, it's not clear how this pricing works for freemium websites or websites paid for by advertising.

There are some other alternatives that are more truly "pay as you go" or have affordable tiers for commercial apps:

TomTom - https://developer.tomtom.com/store/maps-api

MapQuest - https://developer.mapquest.com/plans

Stadia Maps - https://stadiamaps.com/

There is also MapFit, which says it has "Clear pricing that scales with you" but doesn't actually explain any of their pricing:

https://mapfit.com/pricing

If anyone else has good alternatives that are affordable for solo devs, especially ones with good alternatives to the Google Maps Places API, I'd love to hear about them.

We don't have Places API support, but we definitely are aiming for products from solo devs and smaller businesses at Stadia Maps. We have a few suggestions for places to look when our services have a hole.

For instance, Foursquare has a pretty good places API we've used before.

Ah thanks for mentioning Foursquare - I meant to list it. It's another one that has a huge upfront cost for solo devs building a commercial app ($600 / month).

> It's another one that has a huge upfront cost for solo devs building a commercial app ($600 / month).

Yeah. :(

This problem seems to extend to the whole industry. Everyone charges big bucks to even very small players. We want to solve pieces of that we can, but it's a tough road to build quality databases.

Ali from Mapbox here - love to help clear up the confusion & answer any questions. If you want you can shoot me an email ali@mapbox.com

> no way to plan beyond that without inquiring with them about an "enterprise" plan.

I find it interesting how often people (myself included) are willing to consider spending hundreds of dollars on a product or service, but aren't willing to take 30 seconds to send an inquiry email or ring a sales line. Possibly it stems from cynicism (or from a perception that "if you have to ask, you can't afford it") but a few times now I've grumblingly made contact with a supplier only to find that the prices are very reasonable and they're great to deal with.

You're right that part of it is "if you have to ask, you can't afford it". But it's also frustration that they don't have a predictabe, structured pricing plan - they almost certainly tailor it to specific clients above the intro tier (otherwise, why not list it?).

Now I understand custom pricing has to be implemented at some level, but you usually see multiple pricing tiers before it kicks in.

I have no evidence to support it, but this article feels like bought and paid for PR.

Because it's in Forbes?

Native advertising.

forbes.com/sites might as well be wordpress.com, it's just some blog.

Pretty amazing how little such a big brand is actually worth.

Did you miss "This story appears in the May 31, 2018 issue of Forbes" below the byline of this article?

Note that the author is actually a Forbes Staff member. It's not _just_ a blog post.

But in general, definitely look closely at the forbes.com/sites articles.

All media is bought and paid for PR. The only question is who's paying for it and why. That's not necessarily a bad thing, either. It's just the way it is.

Huh? Plenty of media is created and sold as a commodity. Your average movie is not PR. Nor is your average news story from a reputable news source.

I wouldn't say that all media is PR specifically, but it's true that all media is bought and paid for by somebody. It might be a giant corporation paying a mechanical turk to write what they want. It might be somebody with the luxury of taking their own time to write. It might be a professional journalist working for a publication that sells ad space to pay its bills, and which doesn't want to piss off ad buyers.

That is the truth.

I know directly of at least one recent instance where a local news company was not allowed to interview a professional directly related to a story it was airing...because one of their big advertisers wouldn't allow anyone in that field to be interviewed unless it was an affiliate of theirs.

I'm being purposely vague there, but even the reporter was really upset about it.

Indeed. We will be switching away from google to an internal tile server for reasons I won't go into detail for now. Our account manager told us that the pricing of the new Google maps will change with their newly announced platform changes and our current usage would cost us now even less than before. When we asked for an offer from mapbox about a year ago, it was already much more expensive than google. I assume it's going to be even harder for them now to win against google now.

Paul Graham's essay on the topic:

http://www.paulgraham.com/submarine.html

On Google I/O day, no less. It's targeted to ride the search trends.

Indeed. Very carefully omits any mention of OpenStreetMap, too.

Mapbox has some really cool tech. They care a lot about rendering performance and do a lot of open-source work in the GPU vector graphics space; e.g. https://github.com/mapbox/mapbox-gl-js

I've always enjoyed chatting with their engineers about vector graphics stuff. :)

Not so surprising after Google's recent pricing changes. They reduced the free tier by an order of magnitude and 10xed prices.

Google maps is no longer good for hobby projects as the free tier is so low.

It's like they're trying to drive away everybody but but big enterprise shops.

That's probably a good business strategy for the future of Maps. The fewer clients you have, the more focused you can be in terms of product development and pricing. This is true in many businesses, actually.

It's risky to leave customers out that will feed your competition.

Every business decision carries risk. The question is whether or not the risk is acceptable.

If you don't want to serve any customers making under $10M, then other companies serving those customers are not your competition. At least not today. Is there a risk they might take their broader customer base and use it to compete with you upmarket some day? Yes. But that's why you do things like build a brand, relationships, technologies, and other competitive advantages within your niche.

Fancy steakhouses generally don't worry about McDonald's. (Not that this is a perfect analogy to Google vs MapBox. Just speaking generally here in response to your general point.)

Maybe they are. There's a huge difference in approach between the two. With enterprise customers, customer support will be very different for example.

> Not so surprising after Google's recent pricing changes.

Didn't that happen just a week or so ago? I don't think this can be attributed to the pricing change this early.

It depends on your hobby project. Google's free tier gives unlimited map views, commercial or not. (Edit: On native apps.)

Mapbox's free tier only gives 50k map views, and only for non-commercial apps. If you're a paid app, you also have to pay $499 per month on top of your usage.

Mapbox also gives unlimited map views for native mobile apps. Pricing is instead based on monthly active users (50k free).

Not if your app is paid though, and then the $499 only covers you to 250 seats (!).

I like what they're doing, and wanted to use them for a web app with restricted access, but certainly not going to pay $499pcm for a side project.

Happy to chat custom pricing, I work at Mapbox with many of the solo / independent developers. I'm @EqMapbox on twitter.

Sounds more like Google is turning it into an actual business. I assumed that MapBox was making some money, given that it relies on OSM and can cut corners, but it isn't.

Speaking from the perspective of someone who did some amount of web stuff with maps.

Mapbox feels really good. API, docs and rendering performance all of this is really good. Although it's not an obvious choice if you are starting fresh. I personally find the pricing impossible to digest. Probably sounds reasonable for well-founded startups from developed countries. If your doing a bootstrapped product and can't pay 500$ it's not for you.

Google Maps. the v3 it's here for a while. The API feels good, but not great, Stack Overflow is your documentation. Rendering performance is bad, rendering performance with (many) markers without some hacks is terrible, rendering performance on mobile is even worse. I'm trying to figure out what new the pricing means for me.

Leaflet and OpenStreetMap look like a better choice for all hobby projects and small product from now on.

> Widespread adoption has other benefits. Mapbox doesn't need to send out pricey cars or satellites in space to map the world. "Map data is not like food, air, water that's just around. It takes active work to make it," says Young Hahn, Mapbox's CTO. Whenever someone opens a Mapbox map, that person's phone or computer sends three pieces of anonymous data back to the company: longitude, latitude and a time stamp. These billions of data points constantly improve Mapbox's real-time plug-and-play map of the world. "Every time you touch the map, the map is learning," Gundersen says. "It's that flywheel."

What do they mean by that?

It's strange to say that food is "just around" in urbanized industrial society. Maybe for hunter-gatherers...

The Mapbox map is tracking its users. Based on that location data they can derive unmapped roads, turning restrictions, traffic jams (and time-of-day congestion), construction. And feeding other machine learning system to improve the data set. In a recent press release they talk about deployed sensors, no longer users. Google Maps knows when rush hour at your local gym is, Mapbox wants the same. It's a big selling point to their investors because staff editing/fixing the map doesn't scale well.

Any ideas about how they're solving the need to constantly update meta data like speed limits or POI names? This is the part of the OSM data set I found particularly weak when comparing vs. Google Maps.

For those I assume licensing data from commercial providers. For Germany geocoding for example Mapbox uses government data, not OpenStreetMap (found out reading a github issue when another user asked why an OSM change wasn't reflected).

https://www.mapillary.com/imagery is using image recognition for traffic signs and lists Mapbox as a user.

Mapbox has some excellent software. And they're hiring. Sadly, they're moving away from remote work, which probably eliminates a lot of good talent who are also heavy map users.

"Every time you touch the map, the map is learning," Gundersen says. "It's that flywheel."

Curious quote. A flywheel stores energy. What does that have to do with a system learning as it's used? Also curious phrasing. "That flywheel"? Which flywheel?

I suspect it is a reference to "The Flywheel Effect" espoused by business author Jim Collins: https://www.jimcollins.com/concepts/the-flywheel.html

I'm not an expert, and I'm not sure that this in fact a valid interpretation of that concept, but I know that it's a popular concept in certain business circles.

Amazon is also keen on the "flywheel" terminology:

http://www.samseely.com/blog/2016/5/2/the-amazon-flywheel-pa...

The era of mashups / cool web projects using Google Maps has ended. Their recent 28x price change (from $0.50 / 1000 views to $14 per 1000) has been cleverly disguised by the simultaneous rebranding to the "Google Maps Platform".

I wonder how many of the projects showcased over googlemapsmania.blogspot.com are just going to close the door.

Mapbox is a better product and a better development experience. It's also a better styling experience. It's a great case study on the pursuit of quality as a way to draw people in.

We started switching to MapBox some weeks ago, really liking it so far. There are some quirks but one can see that a lot of things are happening at a fast pace.

I generally like Mapbox but I've had some problems with the quality of their geocoding and still would get better results from Google's geocoding. It was a pretty niche thing, though: mapbox doesn't have as many European postcodes, but Europeans in several countries don't really treat postcodes with much respect to begin with. In particular, mapbox has no Eircodes whatsoever.

The quality of OpenAddresses' geocoding (which, I'd imagine, Mapbox uses) seems to be a bit spotty here in the United States as well.

Google's entire advantage here is the amount of resources they've poured into their data; nobody else has been able to match it so far.

EDIT: After doing a bit of poking and prodding, it does seem to be a lot better than it used to be.

We're using it for https://www.citymayor.co and it is getting expensive quite quickly (.5 for 1k page views), but the API is really good and the themes are really pretty as well. I've tried playing with open street map directly to create a theme and it was quite a mess.

Check out one of the services that base their data on OpenStreetMap.

My company (https://stadiamaps.com) does that. OpenMapTiles (https://openmaptiles.org) sells vector tiles you can use. There a many more over at https://switch2osm.org/providers/.

A bit tangential, but I have been unable to find a price-friendly static maps provider for embedding in PDF and print. Anyone have suggestions?

We would be happy to support this type of application! Please email me at luke@stadiamaps.com if you'd like more information pertaining to your use-case.

https://stadiamaps.com

Surprising to see competition in a space that is so dominated by Google. I hoped it’d be some open source endeavor, but seeing the link was Forbes I shouldn’t have gotten my hopes up... that said competition and options is good!

It's not surprising at all. Google have essentially built a moat that is very hard to cross: https://www.justinobeirne.com/google-maps-moat

Specific map rendering projects from mapzen:

Web Renderer: https://github.com/tangrams/tangram Native/Mobile Renderer: https://github.com/tangrams/tangram-es

Native renderer can be accessed for android through gradle, ios through cocoapods and even available on raspberry pi.

Scene file documentation: https://github.com/tangrams/tangram-docs https://mapzen.com/documentation/tangram/ (still seems to work :D)

Scene file authoring tool: https://github.com/tangrams/tangram-play Which continues to live at http://tangram.city/play/

One of the great features of tangram is to embed any glsl shader in the scene file, which gets injected during rendering and "transforms" the map rendering and the map experience to a all new level, giving the some great flexibility cartographer/designer/developer will like.

As an example check this demo "tron" scene: https://tangrams.github.io/tron-style/#15/37.7926/-122.4003 (source: https://github.com/tangrams/tron-style)

Tangram had some great consumer base including this amazing project called StreetComplete (https://github.com/westnordost/StreetComplete/). Check it out and use it to improve osm street data.

Also mapzen's geocode engine Pelias, continues to live as geocode.earch, which it out at https://geocode.earth/.

Disclamer, I used to work for mapzen on the tangram-es (native rendering) project and still supporting the project in the open source world (along with other tangram-es developers).

If anyone interested has any questions, you can contact me @tallytalwar on twitter.

Much of the data they use is open source, coming from OpenStreetMap.

A huge amount of Mapbox's tech is open source, including the entirety of their WebGL rendering engine. They rely heavily on OpenStreetMap, which is all open data.

Mapzen, a now sadly defunct mapping startup, released all of its software open-source, including WebGL mobile and browser SDKs, map tile rendering, search and routing. Take a look at:

https://github.com/tangrams/

https://github.com/tilezen/

https://github.com/valhalla/

https://github.com/pelias/

(Full disclosure: used to work at Mapzen)

What are the stumbling blocks to self-hosting Mapbox?

None. Start here: https://openmaptiles.com/server/

Am I crazy, I can't see a live demo page.

You're not crazy. Go to products -> maps, click on one of the themes, click on View live map. Like this one for example https://www.mapbox.com/maps/streets/

It's like they don't want to show it.

I love this thread. I have not one but two tickets asking we elevate the actual maps - I'm going to tap it again! Stay tuned for a nice redesign shortly that hopefully pulls them front and center.

Note - I work at Mapbox.

Damn it's actually kind of breathtaking how responsive and smooth the zooming and scrolling is on that compared to Google Maps.

Also since I'm a Safari user I figured I'd try Google Maps in Chrome as well to see if Google Maps' lag was due to Safari and nope, Chrome is actually way laggier than Safari.

How does it handle Japanese addresses?

I have a messy address list I would like to geocode, but not sure where to turn

How does mapbox compare to carto.com ?

Carto offers many of the same services, but their focus is more on the ability to analyze vs display data. Their big sales push is "Location Intelligence" vs simply providing various standard geoservices.

I consider Carto and Mapbox to be tangential services in many respects.

Carto is using Mapbox technology in the background. This blog posts explains some differentiation on top https://carto.com/blog/partnering-with-mapbox/

Carto uses geocoding and routing services from Mapbox - but their vector and raster map tiles are powered by OpenMapTiles. See: https://blog.klokantech.com/2018/05/openmaptiles-gives-you-f...

Reply on news.ycombinator.com