Hacker News new | past | comments | ask | show | jobs | submit login
Why Nukemap Isn't on Google Maps Anymore (nuclearsecrecy.com)
787 points by Fej on Dec 13, 2019 | hide | past | favorite | 289 comments

As a software developer for a lab within a big University, I appreciate the bit about Google offering edu discounts, but only if working with a representative _for the entire institution_. A lot of big companies offer discounts this way and it is maddening. There are so many different development projects within a university setting and they often operate with zero involvement of campus IT. It's like expecting a foreign company to involve their federal government in purchasing a software license.

The worst is when it's clear that bargaining has happened: limited use site licenses. There is a hypothetical microscope that costs over 250k and has a service contract. All output files are in proprietary format that can only be analyzed and exported using the proprietary software, and the license is such that only 10 people can use the software at once.

The bean counters in the university aren't going to cough up whatever Zeiss wants for an unlimited use license, and Zeiss isn't going to expand the license without trying to squeeze even more cash from the university. In either case, demand for the microscope would be exactly the same, only now researchers are setting alarms for 3am just to export a damn file.

That really sounds like a scenario where adversarial interoperability should be applied. Maybe build a service for queueing, processing and exporting to a mail/ftp or something like that.

> It's like expecting a foreign company to involve their federal government in purchasing a software license.

I love this analogy.

Also please let me pick a nit: It gets even worse when said foreign country isn't a federation!

Bizarrely, there's no good word for "pertaining to a country". The likely preferred option "national" is subject to a similar objection ("It gets even worse when said foreign country isn't a nation!"--not all countries are nation-states, and we have more than one salient example in the anglosphere!), and "state" has the unfortunate property of meaning something very different to an American.

And interestingly, government is not that clear either.

In Hungarian, for example, we use kormány to mean a mix of government, administration and cabinet, but mostly cabinet. Still it's usually translated to and from "government". But in English government is a very broad word that basically means all the tax funded institutions, not just the ministers. We rather call that "the state". For example in Hungarian nobody would say we have "government-funded" education or healthcare, rather we have "state-funded" ones. But in English "state" is very overloaded with many senses, so "government" took its place. It used to be strange to my ear in American movies when I knew less about the system, e.g. really the government is trying to find you? Like ministers and stuff?

"National" is a weird word in English, because it doesn't typically mean "pertaining to the entire nation" but more like "pertaining to the entire country". I think it's usually the proper term. It translates to "country-al" in the other languages I know.

That said, I don't know about the UK. If a Londoner talks about the "national level", is that England or the UK?

Oh damn I'm wrong and starting to see your point. Eg in German, "land" refers specifically to the federal state and not the federation. You need to use bundessomethingsomething to address all of Germany. Weird stuff!

EDIT suddenly I realize how weird a name the UN has. There's not really any proper nation states left in the word. Bhutan maybe?

I have no idea, but I like to think that it would've been the United Countries but then Scotland would've wanted a seat.

Fun time to nerd out!

The name of the United Nations is an artifact of WWII - that was the name chosen by the alliance on 1 January 1942, just after the US entered the war, to describe their now-common cause. Given the wide variety of regimes fighting on the same side [1] and the common ideological basis of the combatant countries, "nations" and "brotherhood of nations" style of nationalism were an umbrella everyone could get behind while still having an appropriate amount of emotional oomph.

[1] Mostly. The USSR wasn't fighting Japan.

> There's not really any proper nation states left in the word. Bhutan maybe?

I beg your pardon?

France Maybe ?

> Tell that to the Bretons, Alsatians, Occitans, Corses, Catalans; not to speak of the outre-mer colonies.

They were told that, that's why they mostly assimilated. https://upload.wikimedia.org/wikipedia/commons/7/75/SpeakFre...

> France Maybe ?

Tell that to the Bretons, Alsatians, Occitans, Corses, Catalans; not to speak of the outre-mer colonies.

>there's no good word for "pertaining to a country"

In English, the question then becomes "Where are we going to import this word from?"

Dealing with large institutions can be even more difficult than governments!

We recently awarded some grants to people for them to work with an open source scientific package we've been developing.

One of the groups actually gave the money back because it was so hard to get their University to deal with them being awarded money, it wasn't worth it for the amount involved.

I agree that it is maddening. The same thing happens when purchasing hardware. Companies that do B2B transactions don't always support billing to myriad small entities within a single-named organization with a single purchasing department.

There are ways in which it makes sense. The company generally wants to interact with a legal entity, not a research group.

A hybrid approach, that happens occasionally at the University of Washington, is that a department negotiates a discounted license for a software license and then opens the license agreement to broader swaths of the university.

When I did my PhD, we just created a lab website on it's own domain and applied for a Google apps for education account for that domain and got it. That's been our labs main collaboration tool ever since and it's been six years now.

> It's like expecting a foreign company to involve their federal government in purchasing a software license.

Google is familiar with that business model, at least when it comes to dealings that the Chinese government might take an interest in.

There's also many different IT departments on a campus in my experience. I've never worked in a physics department that interacts with campus IT. It's either great or terrible.

One of the criticisms of when Google Maps API was basically a free for all was that it was suppressing the ability of startups to exist by devaluing products; it seems like with these pricing changes Google is giving the competition an opportunity to swoop in.

A business model based on data hoarding will never last long-term. As soon as one open source player steps up that's 60% as good, it's over. Like Encarta or Britannica vs. Wikipedia.

Lasting value comes from the software in between the database and the view. In the case of encyclopedias or maps, the software is the most trivial part.

I really think Google Maps will die. It's just unlikely that for the rest of history, no one will improve OpenStreetMaps to the point that it is competitive.

You need a massive amount of capital to create a redundant data set that has a death clock, doing nothing good for the world but extracting some small rent created by antitrust laws. No thanks.

Traditional encyclopedias didn't lose because of their intellectual property model, they lost because they were trying to charge money for their product, so a free as in beer competitor was able to take all their market share.

A maps data set will require constant updating and the quality of the data set will be determined by the amount of work put into it on an ongoing basis. The question is which number is bigger: the monetary value of the time people are willing to donate making contributions to openstreetmap, or the amount of money that Google makes on map ads and can therefore spend improving the data quality.

There are parts of the world that have much better OSM coverage than Google maps, e.g eastern Indonesia. I don't think OSM is ever going to beat Google in for instance New York City but it absolutely can and does beat Google in parts of the world that are not profitable for Google to map. That's the problem with a for-profit map: consistency.

The exact same problem exists with volunteer-maintained maps. In fact it's even worse there because at least with a commercial map, quality is basically correlated with where the users are. With a volunteer-maintained map, quality is correlated with where the volunteers are, which is not necessarily going to be where the users are.

Thousands of volunteers specifically map remote areas they don't live in through the Humanitarian OpenStreetMap team

Every local municipality already maintains GIS data, I imagine that some day these systems will automatically integrate into a public, open database like OSM.

Sidenote, the FOAM project is doing some interesting work into geospacial data on-the-blockchain.

At a regular (bi)monthly/quarterly OSM meetup we were joined by representatives from 2 or 3 local municipalities. It seems they might be interested in unification to reduce costs, but I didn't ask for details.

> I really think Google Maps will die. It's just unlikely that for the rest of history, no one will improve OpenStreetMaps to the point that it is competitive.

From reading the analyses at https://www.justinobeirne.com/, plus just generally following the tech industry, I'm convinced Google Maps will never die, at least not in any timeframe relevant to this discussion.

Between autonomous vehicles at Waymo and Google's desire to provide intelligent suggestions to their Android and web users, they need to have as comprehensive a data set as algorithmically possible.

I suppose it's possible that OSM could become good enough that Google would only need to supply a proprietary layer on top of it for their own needs, but given that they can collect better intelligence about roads and buildings than anyone else through their unwitting spies, I doubt they'd want to give up control over the full stack.

>As soon as one open source player steps up that's 60% as good, it's over.

Does this imply that Linux is <60% as good as Windows because it's not over?

Linux only lags behind in desktop/laptop use. (Dominates servers and smartphones.)

Yes, absolutely, in the case of desktops, it is <60% as good as Windows or macOS. I say this as a nearly lifelong desktop Linux user. The userspace desktop software gets almost no attention compared to the kernel/systemd which actually matters for business. The major DEs like KDE and Gnome3 are full of exploits and memory leaks. The alternative to using a DE is to use a tiling window manager and spend hours a day re-implementing what would have been one-click operations on Windows/macOS. This is not to mention the lack of non-development-related software targeting desktop Linux.

Will this change in the future? I think it is very likely, but the time horizon is key. There are thousands of possible realities where Microsoft or Apple fails--political problems, economic events, bad management, some black swan startup out of nowhere. All it takes is for one of these to trigger for mindshare to shift to Linux (or BSD or Fuchsia--any free one for that matter) and bring it over the threshold.

(But this is kind of irrelevant to the point about data hoarding business models, since Windows/macOS is pure software.)

As someone who dual-boots, I would say Linux is a better OS, today, than Windows or Mac. Sure, there are areas where it sucks more than the others, but the same can be said of either of the other ones for different areas.

As someone who also dual-boots. I would say that Windows is definitely the better of the OSes. I've tried, so hard, to make it work. It's been almost a decade now since I switched to Linux OSes. But, software/package management is absolutely horrible in the OSS world these days. It's so bad that actually downloading + installing MSI/zip files from the software provider is a better experience and more reliable. Linux has lost almost all the benefits from having a package ecosystem.

These days, everything on linux is either random potentially insecure PPA repos, shell scripts loaded from a server via curl and piped to sh with sudo access, or "snaps" (whatever the fuck they are). That all ontop of wildly outdated packages in the supported repos. No wonder everyone rants and raves about "docker", now we know why things moved into that space.

I know, this is my opinion and it's biased. Yes I may not have done things properly, or there are other ways of doing things, etc etc.

I really can understand how you have gotten to such a bad place with packaging. What are you doing that needs lots of PPAs?!

Snaps are actually a great security and packaging advance that decouples package dependencies from a common library set, while maintaining a mechanism to patch and upgrade otherwise dangling library deps. It also starts down the long road to confining software, because what you run shouldn't automatically assume all your privs. Basically iOS style apps for Linux.

People who pipe web content to a privileged shell are on their own at present. Arguably they should be warned, but that assumes taint tracking at every level of the operating system, and no production OS has that. If the software is a dependency shit show then putting it in docker (and then into a VM) is a better compromise.

Windows is already deprecating the installer model for the app store and cloud concept, they just didn't make it stick the first time round.

Snaps have been the single provider of crash, breaking the machine or killing ny battery i have ever seen on linux. It is to the point that if all i can find is a snap, i prefer to boot windows to deal with it.

Snaps are a good idea in theory but do not work in practice.

Well, I'm surprised. Snaps have really simplified things from my perspective. Any you'd like to call out? Particularly they shouldn't be able to 'break the machine'.

I have had multiple time uninstalling a snap that installed its own X stack. And discover that it had been swapped for the whole OS to use this one.

No more graphical stack.

This happened to me with not only X. It seems to me that installing is tested. Removing is not. It was also badly documented last time i checked.

Happened with discord last time.

Okay that is seriously bad. Looking at https://github.com/snapcrafters/discord/issues they have several unresolved issues like that, seems this is unmaintained.

Ubuntu needs to more clearly explain whether the app is confined, and force apps to use an API to change the system environment to allow rollback, multiple packages of the same function etc.

You should try something other than Ubuntu. Fedora or Arch are my standard recommendations these days.

Oh, you think MSIs are sticking around? Package manager insanity is taking over Windows slowly too.

> The major DEs like KDE and Gnome3 are full of exploits and memory leaks.

Can you give any supporting evidence for this?

> The major DEs like KDE and Gnome3 are full of exploits and memory leaks.

Sry, but that is just not true.

When it comes to an OS, being 90% as good would still mean a 10% drop in productivity.

The margins are a bit different here.

I think it implies that Windows is not a business based on data-hoarding.

For the average consumer, (i.e. not the average HN user, where this is far more debatable), yes.

For the use case of running whatever win32 executable is promoted on zdnet.com (or pcworld.com or whatever is out there that appeals to computer users that aren't developers) any given day? Yes.

Do they actually promote non-game win32 executables these days? I thought current fashion is mostly services with mobile and web apps, and sometimes Electron ones?

Windows is not over? I see Microsoft as similar to IBM now. They will make a lot of money from their Rolodex, possibly more than they ever made before, but they no longer own a relevant platform someone would write new software for.

Perhaps I wasn't clear enough, when a platform is dying its owner:

1. Tries to capture all profit from the edges.

2. Extracts value "unreasonably" both to grab what they still can and to create an explanation for the death. This could be high fees, ads, or extraction of market intelligence.

3. Begins embracing the new platform with compatibility layers.

4. Starts making 2 offers on everything. One with their legacy and one with the rising platforms.

5. Treats their nonfounding CEO as significant and someone to listen to.

Plenty of companies make this transition successfully (I chose IBM as the example) and make huge profits from not having to charge uniform prices across the entire market. But the discussion was the windows platform being replaced by Linux, not Microsoft profits.

Are you aware of the video gaming industry

Or the enterprise software industry

Are you aware of IBM in the enterprise software industry? They manage to do almost all of that with their own software and open source rebundled.. No OS/2, almost no 3rd party programmers who are interested in their platform.

That is the future of Microsoft and the non-future of Windows.

Or, to a certain extent, the graphical industry?

Yes, windows developers with experience are going to go work in the game industry..

This thread seems to be forgetting that Google Maps didn't just replace the atlas (for looking up names of things) and the dedicated GPS (for routing / directions), they also replaced the radar detector (OK that's Waze, but same people and same data), and the phone book.

I don't know about you, but I haven't bothered to look at any of the various "yellow pages" sites (to find a phone number or address), nor at Yelp (to find business reviews) in many years. And to make matters worse, I'm actually one of the volunteers feeding free labor / data to Maps, because I post reviews and make edits to business information. What can I say? They're really good at what they do.

Apple Maps is way more than 60% there in terms of quality but I still don’t find it very compelling for the same reason I won’t switch to DuckDuckGo - a palpable difference in quality gets me paranoid and makes me double check a lot of stuff.

That instinct to double check makes me go “Oh well let’s just use Google.”

Fortunately, the quality of Google searches has been declining at such a rate that you will soon be double-checking on DDG, instead.

It has already passed that point for me, given that Google is a) subject to censorship, on a per-country basis b) likes to return search results that don't have much to do with your query, apparently because that's what "most people" want c) records every query you make from your IP address, with no way to turn it off if you aren't logged in to a Google account d) the results I get from DDG are generally what I need.

Don't they get their results from Google, though?

No, I think it's Bing and their own index.

It’s pretty much just Bing. Saw posted a bit ago that if you DDG “what is my user agent”, the website preview text will show that it’s Bing. Which means DDG is just whitelabling Bing’s index.

I use bing exclusively. Sometimes I don’t find what I want, and switching to google is as simple as swapping out the authority. The query syntax is the same.

Apple maps has gotten a lot better than release but it still has a long way to go imo. It's probably fine in the bay area at this point, but boy does it do some dumb shit in LA. It's worse than waze, and waze is worthless (an unprotected left across a six lane boulevard is just not happening, among other similar boneheaded moves).

Google also spoils me with multi modal options, although the transit estimates seem to be a complete stab in the dark and there are quirks like invisible buses. The bike timing is also useful, but the routing is usually junk and profoundly unsafe.

Living here has taught me that all these apps generally suck, just google maps seems to suck slightly less.

Last time I was in SF (last August) Apple Maps couldn’t even correctly locate the Walgreens stores downtown. I so much want it to be good, but Apple refuses to take it seriously even in their own back yard.

I'm currently testing out https://beta.cliqz.com/

If you use Brave, you can easily choose your search engine with `:g` for google. You can also set your own shortcuts for any search box on any website.

It's all possible with Firefox and DuckDuckGo.

you have to use the shift key to type a colon though or if you're in mobile you have to long press. same with duckduckgo's exclamation mark.

I much prefer the Firefox method where you add the search shortcut to a browser bookmark and then if you sync your bookmarks to the mobile app.

I use gg for Google, gi for Google images. dd for duckduckgo and di for ddg images etc

custom search keywords have been available in firefox for at least 15 years and chrome since it was released

Google Maps has a multi-layered business moat that makes competition really difficult

* Massive staff and fleet of vehicles on the ground with multiple sensors mapping the entire planet. Capital costs here are enormous * Massive network and bandwidth with caches in just about every major IX to deliver ultra high performance maps * Major established contracts with satellite mapping vendors to get the best data as early as possible * Major brand recognition with many users using the app and knowing it to be the best of the best

I would never want to compete here, but a major institutional player could

You're talking about competing as a consumer map/gps, which is not what the blog post is talking about (or the parent comment, I assume).

Rather, GMaps used to be the default choice if you wanted to build an app on top of a map (in this case, looking at the effects of nuclear bomb detonations). The blog post explains why GMaps is no longer a reasonable choice, which (as the parent notes) opens up room for competitors.

> ”Google Maps has a multi-layered business moat that makes competition really difficult”

OpenStreetMap is going from strength to strength. Maps startups no longer need to duplicate Google's on-the-ground mapping efforts. Now, they can make use of open data.

Mapbox is perhaps the most well-known, but there are many commercial services based on OSM data: https://wiki.openstreetmap.org/wiki/Commercial_OSM_Software_...

When Pokemon Go switched from Gmaps to OSM, the quality drop was dramatic and noticeable. I want OSM to be a real competitor, but I think it represents a perfect example of how difficult it is to compete with Google Maps.

It counts as "Good Enough", but man, it's disappointing to see how inferior Pokemon Go maps are these days. Roads that don't exist, buildings in random places with roads going through them, and map data that seems manipulated for game advantage (by players contributing to OSM for gameplay reasons, not for accuracy reason) ... and this is in the midtown area of a top 10 US city.

> "Roads that don't exist, buildings in random places with roads going through them"

This is unfortunate and surprising. In my experience (mostly UK/Europe) the quality of street data in OSM matches, and in some cases exceeds that of Google Maps. (Points-of-interest data does not, however).

If you do notice errors and omissions, fixes can be made in seconds by getting an OSM account and clicking the "edit" button on OpenStreetMap.org. In recent years the development of the "ID" editor has significantly lowered the barriers to entry of editing OSM.

> "map data that seems manipulated for game advantage"

Perhaps OSM needs some sort of Wikipedia-style spam detection/prevention if this is happening on a wide scale.

If you do notice errors and omissions, fixes can be made in seconds by getting an OSM account and clicking the "edit" button on OpenStreetMap.org. In recent years the development of the "ID" editor has significantly lowered the barriers to entry of editing OSM.

This is the same argument that keeps every year from being the year of Linux on the desktop.

"My computer crashed!" "Oh, just become a Linux kernel developer and fix it yourself."

Comparing a text edit on a website to becoming a Linux kernel developer is like comparing picking apples to running an industrial agriculture company.

Most people who just want to eat an apple (even an apple a day) don't have bandwidth to pick them. Offer them the option of picking apples at an orchard for free vs. buying a bag of apples at a grocery, and most people will go buy the bag.

The same happens with an OS. Most people don't want a free labor-intensive solution. They want something that just works, runs their software and stays out of the way. People don't want to think about what OS they use.

Yes, typical users want a "just works" solution. But yc-news commenters aren't typical users. In the time it takes to post a comment on this forum, anyone here can just fix the problem themselves using OSM's very simple point-and-click editor.

Contrast to Google Maps, where the process of getting a change made can take weeks or months - if the fixes ever show up at all.

There really is a lot of power in having a map of the world that we can all "just edit".

IMO, HN readers can't be the target audience of a mapping product that's being used as a backbone of so many large projects. You have to make it painfully simple and give people a real incentive to do it.

The act of contributing an edit to OSM is vastly simpler than fixing a kernel bug. But OSM does still have the massive barrier to entry in that if you are using an application built on OSM data, it is usually not easy to realize this and understand that OSM is where you need to go to fix a map error, unless you are already familiar with OSM.

> it is usually not easy to realize this and understand that OSM is where you need to go to fix a map error

The OSM licence requires that people using OSM data attribute OSM, i.e. that their users are aware that the data comes from OSM. In theory this means that the final user should know that OSM is where they should go to fix data issues.

But, alas, it's often not followed that well...

If you're not already familiar with what OSM is, the copyright notice is really not going to be enough to alert the user to the fact that OSM works differently from commercial map data suppliers and can easily accept user contributions.

So when my grand mother has an issue with her app, she’s sure to notice that it says “open street maps” in the corner and go to the website, make an account, and contribute more accurate data? I don’t think so.

Really, it’s not even an age thing. I work in tech. I visit HN on the regular. OSM, Google, and Apple all have inaccurate data about my daily commute to work. They all say a path that exists doesn’t. Have I don’t anything about it? Absolutely not.

Imagine the average Pokémon Go user; if they even read the boilerplate that stood between them and playing the game, how many of them understood what role OSM data played in their experience, realized what the word “open” meant, and correctly concluded that they could do something to influence/correct the situation vs just thinking the game had a bug or glitch?

Also 99/1 rule. Inaccuracy can be not a significant nuisance/worth the effort/known to 1% who contribute.

The point is that nobody is going to drive a hundred miles out of their way to pick apples.

This is exactly why Linux is such a good development tool.

No idea why you're getting downvoted since anyone who plays Pokemon Go can attest to this. It has gotten unbelievably bad since they switched to OSM.

I'm far from a power Pokemon Go player, and I live in a city that's likely to have very good map coverage, but those caveats asides I didn't know that this even happened.

I agree that there are disadvantages in OpenStreetMap. In other parts of the world, Google maps is worse. However, for showing which parts of a city would be dead by a nuke, you don't need to have every road laid out accurately.

The moat's draining because more of this data is becoming commodity.

OpenStreetmap data is quite good. I can run a local instance with all the data and the tile server and a web server in a docker image in maybe ~10 min. Source: did this a few years ago during a particularly boring meeting.

What do you mean when you say "all the data"?

The full planet takes "a while", even on a huge computer:


I recall something like an 8gb download that uncompressed to a postgresql db.

The thing benchmarked in my other comment loads an OSM dump into PostGIS (in 1 particular format, translating the OSM data from an editing/versioning friendly format to something more useful for map making as it goes). The current full planet dump is 84 GB:


8 GB gets you a large region still, depending on how much mapping has been done:


It’s kind of ironic that Google Maps is so bad at some things then. Walking & Cycling directions are especially bad - I get much better directions from OSM than I do from Google Maps.

For driving directions GMaps is much better. I presume that’s where the majority of their userbase is.

Google maps mostly avoids the cardinal sin of most navigation apps: the unprotected left across the multi lane road in heavy traffic.

In my experience GMaps route optimizations don't account for elevation or road size in my country. When I drive outside city I strictly use OSMand for routing.

GMaps driving directions failed HARD for me last night.

I had to make a trip during rush hour. I was already going to be cutting it close on timing as this is a 12 minute drive that can take 15-20 minutes at rush hour. I plugged in my destination and GMaps routed me a way I would rarely take but didn't seem too odd. One block past my last alternate route, I saw the lights ahead. There was a 4 car crash with at least 5 emergency vehicles on scene about 1/2 mile ahead. GMaps had no clue and I had no choice but to inch through about 10 light cycles to reach the turn Gmaps wanted me to take.

The 15 minute trip took me 35 minutes. Thanks for nothing.

You mean, there was an edge case where Google didn't have real-life data for a specific accident available for you? This is an extreme case that we wouldn't even expect to exist a few years ago, (and then got annoyed by passive surveillance) but now gmaps is failing for missing it? Is this the new norm?

Did you mark the accident yourself when you saw it?

He shouldn't need to mark it. Maps can tell that its users are spending 20 minutes to get through a single intersection.

If you have enough users with gmaps on that segment, it will happen. Can we guarantee there was enough in that situation?

After how long? 10 seconds? 2 minutes? 10 minutes?

This seems unreasonable.

For real-time data, Waze is much, much better.

If it weren't already aware of the accident so it could warn you, you could have let Waze know about it and, at the least, helped out many others who were coming after you.

A few days ago I had a rideshare driver unable to find me because Google kept telling him to take a non-existent road through the middle of a building.

You wouldn't compete on the things that Google does well, but rather on the things they don't. Not saying it is easy. Price and user interaction is something I believe could be improved upon. Also maybe crowd sourcing info.

Other pain points include bike routing and unreliable (but still more reliable than any other app I've tried) transit times.

"mapping the entire planet"

that's not true though. I know from doing HOTOSM that there are quite a lot places where there is only a road or two going through it only google maps but there is actually a city/town there. there are some counties where Google Street view cars have never even visited.

all of these services sound like things a bunch of different startups would compete to provide. doesn't google outsource stuff?

People said the same thing when Google Apps (now G Suite) dropped the free trier. I didn't notice anything happening.

I think that's not a great argument… If the free Google Maps API put your company out of business, it didn't have a great product in the first place. If these companies did not provided added value, what was the justification for their existence in the first place?

Providing an embeddable map solution with developer friendly API used to be a great product. The fact Google disrupted the market with a free solution doesn't mean the product wasn't great up until that point, it just means someone also saw that product idea and undercut them on price.

Source: I was using mapping tools before Google Maps came to market.

If the free Google Maps API put your company out of business, it didn't have a great product in the first place.

So when a book store giving away books for free puts another book store out of business, that second book store didn't have a great product?

If the store is able to continue to give away books for free, then obviously the first store that had to charge for the books isn't generating enough value.

Unless the end-game for the 2nd store is to price out every single book store, then obtain a monopoly. But then anti-monopoly laws should come into place.

But if the end-game for the 2nd store is to subsidize their cost via a different business - then it's the same as a newspaper using classifieds to subsidize their subscription business. You can't then argue that another newspaper that charge for subscriptions goes out of business is unfair.

Most people will take free and good enough over a superior product any day. Google basically killed the market for mapping products and now have started charging everyone for theirs when used commercially.

That maps api was used by all kinds of hobbyist websites. Once they pulled the plug on free, all those sites bad to scramble to find alternatives.

What is your business that can survive arbitrary increases in costs?

I develop an application at a university that allows users to share and lookup accessibility information on public places that uses Google Maps and the Google Places API. We have run in to exactly the same issues as described in this post and we have had a few meetings on what to do about it. I would like to switch to OSM but I haven't been able to find a quality replacement for the Google Places API, which is critical to our application.

We could switch to using OSM + Places, to at least partially alleviate our problem, except that Google requires any Google Places API results that are displayed on a map to be displayed on a Google Map, which really locks us in.

Is anyone aware of a good Google Places API alternative, including any paid self-hosted databases?

What type of accessibility information are you getting back from the Places API? OSM seems like a great fit for this since information is tagged in a structured way: https://wiki.openstreetmap.org/wiki/Disabilities

Feel free to email me, this is my area of expertise. For projects like that I can often help pro-bono.

We actually aren't using the accessibility data from the Places API, we are collecting and displaying custom data that is provided by users (detailed ratings, reviews and pictures from regular users, and highly detailed building assessments from trained users).

We only need the places API for having a comprehensive list of nearby places. Other services had too few places in general, but didn't necessarily lack information about the places that were available.

> OSM seems like a great fit for this since information is tagged in a structured way:

OSM both is, and isn't, structured. It's free form tagging system, where anyone can start adding new tags whenever they want, without any gatekeepers.

I just spent some more time looking at OSM's place data, and I'm not sure if more has been added since last looking at it, or if I wasn't using it correctly, but it seems to have more than I remember.

I see that a couple of popular restaurants that have been around for at least a decade are still not shown, but maybe that is anomalous? I'll have to re-consider OSM.

If you're collecting accessibility data consider adding it to OSM or making it available publicly for others to add :)

> I would like to switch to OSM but I haven't been able to find a quality replacement for the Google Places API, which is critical to our application.

I know a lot about OSM, but next to nothing about Google APIs like this. Can you explain what you're looking for, what your requirements are? Maybe OSM can help you...

> Is anyone aware of a good Google Places API alternative, including any paid self-hosted databases?

I think you can achieve this with wikidata / wikibase : https://addshore.com/2016/05/geospatial-search-for-wikidata-...

You will have to populate it yourself though.

Maps has gotten progressively worse in most ways over the past decade, and it drives me crazy. I asked a google employee about this who told me a lot of the changes I hate stem from a refactor around 2014-2015. Obviously there was some big culture shift on the team though...

Google maps is

* Much, much slower than it used to be. My machine in 2012 was able to drag the map around and everything would load just fine.

* Harder to read/understand, especially street names. A lot of the time I have to zoom in 100% and scroll up and down a street to see what the name of the street I'm looking at is.

* Much worse at text parsing e.g. I can no longer loosely type something like "10th and grove to Jim's Hardware" and get directions. In fact most of the time I can't even type in a major intersection and get what I'm looking for.

* Obnoxiously spammy: they push new "features" on me all the time that I don't want to use (including a recent popup mid-directions when I turned on location tracking because I was lost on a busy, complicated highway intersection. It felt pretty dangerous, getting confused by a popup in the map like that mid-intersection), and the phone version asks me to turn on location tracking EVERY SINGLE TIME I OPEN THE APP.

* Addresses seem to be getting replaced by some weird google maps-designed address format in Colombia (maybe other countries?) that are useless to humans

* While every good review site seems to fail for whatever reason (tough to monetize, and fighting spam/fraud is also hard?), Maps seems to have become the de facto review site in many places. But the UX is so godawful I can't even wrap my head around it. For example, search for a restaurant, and you get one (ONE) possible filter in browsers: star rating. There's also a "more filters" button which brings you to a separate view where star rating is the only available filter. On the phone app, sometimes I can select "open now," and a few other things about 20% of the time I use the app, and it's not clear why those options are gone the rest of the time...

I think it's a real shame, because I remember in, e.g. 2012 I thought Google Maps was an incredible revelation. It was so much better than MapQuest, it was free, fast, and just worked. If only they had open-sourced the old version before the refactor...


Anyway, I hope someone writes a nice open-source view layer that sits on top of OpenStreetMaps someday. Like a wikipedia for maps

You made me realize that I share all these grievances but the degradation has been so relatively slow that I never really thought about how worse it had become, boiling frog style.

I've definitely experienced the "wait why won't it show me the name of that street I'm clearly focused on" problem multiple times. It's especially ironic because I remember that once Maps readability was first class and an example to follow, IIRC mainly because they had a relatively simple but elegant algorithm that would aim to maintain a relatively constant information density over the area of the map at any zoom level. So if you had a very dense area only the very important labels would show and the rest would be hidden until you zoomed in, while in more sparsely populated areas you'd have a lot more detail pop up even at relatively high zoom level. That was a good compromise.

In hindsight it's completely obvious but I remember that when Maps started that was relatively innovative, I remember that many predecessors tended to emulate paper maps more closely (probably because that's what people expected back then) and were a lot more cluttered as a result.

>I think it's a real shame, because I remember in, e.g. 2012 I thought Google Maps was an incredible revelation

That seems late to me, are you sure that's the right date? I remember spending hours zooming in and out of Google Maps (and Google Earth) in the mid-to-late 2000's. Having a full map of the world to explore was mindblowing to me. Apple Maps launched in 2012!

You're right! 2012 is definitely too late... it feels weird to be old enough that things I vividly remember happened ~15 years ago

>Harder to read/understand, especially street names. A lot of the time I have to zoom in 100% and scroll up and down a street to see what the name of the street I'm looking at is.

To add another nitpick, I often want to see the name of the street I’m standing on, especially in San Francisco which doesn’t like spending money on street signs it seems. But GMaps usually requires zooming way out or panning a non-trivial distance to find that. All the other nearby streets on the map have clearly visible names without having to do that, just not the one I’m one. It happens regularly enough to be annoying.

Adding to the chorus, but the decisions on what street names to show is simply insane. Opening the map right now on Android and scrolling around, almost none of the major streets near me are named, and tons of tiny streets are. Why would anyone do this? What AI got programmed that learned to make these decisions with no oversight?

Apple Maps does this too. Or it'll show the street names in tiny, tiny font and then keep the tiny font when you zoom in. "Hello, I'm zooming because I can't read the font, you arseholes!" (although I will grant it's impossible to distinguish between "can't read font" and "need more detail" zooms but still.)

More annoying, to me, is how much of a struggle it is to determine the names of streets that a navigation route follows.

I'm the type that prefers to look at a map to see where I'm going, and it drives me absolutely insane to have to pan up and down my route just to try to determine the name of a street I need to turn down.

For other use cases, I can somewhat empathize with the complexity of determining what street names to show. And over-cluttering with street names wouldn't be much better. But for a navigation route, the logic should be dead simple: throw a street name within the viewport if that street is part of the route.

(Yes, I fully realize I can switch to step by step directions. Maybe it's because I'm more of a visual person, or maybe I'm just old and used to using maps for directions, but I just can't stand using that view to figure out where I need to go.)


Every time I open the app, half the screen is taken up by some "Explore your neighborhood!" nonsense. Makes it very difficult to actually use the map, which is the only reason I use the app. 99% of the time I'm in my neighborhood, which I've already thoroughly explored to my satisfaction, and I'm just trying to find some single specific item.

I've simply set my phone to stop updating the Google Maps app and stick with the ancient version the phone originally shipped with, because my phone's screen hasn't grown in resolution to counteract all the screen space Google Maps now wastes.

That works until Google releases an incompatible change on the API. Then you either upgrade or are out of the app.

Worse is that the app will capture any link to the web maps, and AFAIK, you can't change your mapping app on Android.

I find it a bit surprising that such an old version of the app still works today, but the fact that they haven't broken the backend in over 3 years means I'm not too worried that they'll do so before my hardware bites the dust. (I'm expecting the shutdown of 3G to be what finally kills this non-VoLTE phone, though I am likely to stop using it as my primary phone before then.)

Thank you for reminding me I can do that... going to give it a try and see what I'm missing/forgotten, and what features I might miss.

Wish there was a VCS-ish way to get an intermediate version.

Why is it so hard to find the name of the street I’m on in Google Maps. It often takes a minute or so of zooming and panning to reveal that.

It seems to me that Google Maps is far more focused on being a suggestion engine than a map now.

Yes, and this even happens with major highways! I'll be trying to pick a ~200-mile route, and I have to play zooming gymnastics just to see the difference between I-84 and I-95!

Okay, I should just know the difference with roads that big. And okay, generally East/West interstates use even numbers representing what % of the country lies to the South while North/South use odd numbers representing what % lies to the West. But I shouldn't have to know that when I'm wondering how to avoid the accident on "that big yellow stripe that passes through D.C."

Sometimes I'll use a mapping app designed for outdoors activities, because it is so much easier to read than Google Maps.

I had no idea about the interstate numbering system. This is awesome to know, thanks.

It's a cool system. The numeric rules are (though there are lots of exceptions):

* Primary interstates are one or two digits. Shorter spurs, loops, and connectors are three-digits.

* East-west primary interstates are even-numbered, north-south are odd.

* Major arterials are divisible by five. (The longest interstates are I-90, I-80, I-40, I-10, I-70, I-95, and I-75.)

* Odd routes increase in number going from west to east.

* Even routes increase in number from south to north.

* The last two digits of a three-digit interstate usually identify its main parent interstate. (For example, the I-310 spur connects I-10 to US 90.

Even routes increase from south to north (e.g. I-10 runs through Arizona, I-90 runs through Washington).

Oops, fixed.

Another fun tidbit is that any 3-digit interstate is an auxiliary route of the 2/1-digit mainline. The first digit determines the type of route, with odd being a spur and even being a bypass or beltway. So I-405 is a bypass of I-5, and I-110 is a spur off I-10 connecting it to the Port of LA.

I-110 is also a spur connecting I-10 to downtown Pensacola, FL, Biloxi, MS, and downtown Baton Rouge, LA.

And apparently there's a I-405 in Seattle. (All the I-110 spurs I've driven on.)

> And okay, generally East/West interstates use even numbers representing what % of the country lies to the South while North/South use odd numbers representing what % lies to the West.

The idea that the highways are designed and numbered to represent the proportion of land area of the contiguous US seems specious. Source?

Edit: the interstate highway going north from Kansas City is 29, but obviously much more than 29% of the US is west of I-29.

The semantics aren't exactly meant to be "% of land area", but they do represent a rough coordinate system[1], from (5, 8) in the SW corner of the contiguous US (San Diego) up to (95, 90) in Boston.

Interestingly, US highways follow the opposite system[2], going from (1, 2) in the NE corner of the country (Maine) to (former) (101, 80) in San Diego.

[1] https://en.wikipedia.org/wiki/Interstate_Highway_System#Numb...

[2] https://en.wikipedia.org/wiki/United_States_Numbered_Highway...

Those us routes came first and the interstates are purposely numbered backwards of it to avoid confusion.

Numbering from west to East and south to north and using evens and odds to mean East-west and north-south makes sense.

But that doesn’t imply anything about proportion of country, however the measurement is defined.

Sure it's anecdotal, I just heard it from my parents as a kid.

And obviously it's not exact, just look at how close I-94/I-90 are around North Dakota and compare it to I-90/I-80 or I-80/I-70. And a lot of the "East-West" interstates go diagonally or even North-South in places.

But I doubt they made the roads by laying rulers across the map (except maybe in Kansas or Oklahoma), and it does seem like they generally increment from 0-100 in each direction with odd/evenness decided by the primary axis.

Wikipedia has a lot to say on the matter, but I got bored reading all the special cases like how they number loops that connect two highways: https://en.wikipedia.org/wiki/Interstate_Highway_System#Numb...

Thank you I had no idea how the numbers were chosen until now!

That‘s arguably what many people use it for, though.

Regarding the street names. I usually approach the service from these 2 directions:

- Address or name of place is known. I use Maps to find out where it is and how to get there.

- Trying to find places in a certain area (Laundry, restaurants, shops, etc)

Only very rarely do I need to know a street name specifically. Not saying it‘s not important for other people though.

This also happens with rivers! It's incredibly annoying.

Adding lake names to the problem.

> Much worse at text parsing e.g. I can no longer loosely type something like "10th and grove to Jim's Hardware" and get directions. In fact most of the time I can't even type in a major intersection and get what I'm looking for.

That's the part I don't get. Google is all about search. Their main engine is so good it's creepy. and yet, for me, it is unable to parse a perfectly well formed street address, written as recommended by the postal services.

I know it is a hard problem, but that's Google we are talking about. I also live in France (i.e. not the US) but our street addresses are not so different, and Google has a presence and generally works well here.

For example search something like "building X, 123 street, postal code, city" next to your current location and it will find a shop called X on the other side of the country. Search "123 street, postal code, city" and it will find exactly what you want, to the letter. Why? Is it that hard to throw off a small bit of information to perfectly match everything else instead of searching for an insignificant random detail? Probably harder that it looks but if there is one nontrivial thing Google can do, that's it.

I get very annoyed copy-pasting addresses into Google Maps because it completely fails on some forms of zip codes and countries. Just deleting from the end of the address until Google shows the right result works.

I can't find any examples of this not working. Can you give an example? The GP's complaint of "x and y to business" works perfectly well for me. E.g. "55th and 5th to barclays center" works perfectly.

> * Harder to read/understand, especially street names. A lot of the time I have to zoom in 100% and scroll up and down a street to see what the name of the street I'm looking at is.

This one is hugely aggravating, but there's another one that annoys me more. Try searching for "food" or "restaurants" at an intersection with 10-20+ options, Google will completely hide plenty of results. Just show me where I can eat at nearby...

I don't understand how anyone can prefer Google maps's style (probably lack of knowledge of alternatives). It's pretty much unreadable, extremely low contrast, doesn't show like 20% of streets even if they're there and fit easily on a selected zoom level, and thus is useless for orientation/printing for walking types, like me.

Thankfully I only ever have to use Google Maps if the website integrates them into the contact section, and we have much better local mapping website/app.

Same zoom level:

Gmaps: https://megous.com/dl/tmp/fbcf80d0eda5570c.png Local app: https://megous.com/dl/tmp/5cbe3d860874717b.png

I can use one to comfortably find my way in the city center even without zooming, and it's not the Google Maps.

I can look at the better map, figure out where I am and where I want to get to, figure out my next few skips and turns and forget about the map for the next 5-10 minutes.

I don't need to constantly zoom around or use GPS at all. Actually, I wonder if Google made such a garbage map style just to force people to fiddle with their devices more, while they're navigating around the city, and to make them use GPS, because it's really hard to orient yourself and plan if you lose so much detail when looking at lower zoom level for overview.

I'm guessing most people using gmaps are looking for something like places to eat or landmarks/attractions then use the navigate feature to get there and so it focuses on presenting that information.

Well that's my point. Stupid map that forces you to enable GPS, and fiddle with the device constantly, instead of presenting useful information to plan a walk for several streets and enjoy/be aware of the city.

Harder to read/understand, especially street names. A lot of the time I have to zoom in 100% and scroll up and down a street to see what the name of the street I'm looking at is.

This is so bad, incredibly bad. The legibility of street names is also extremely sensitive to lighting conditions (at least on iOS) which is a rare accomplishment in this age of screens that adapt their brightness in response to ambient light.

The color design of their iOS application is also terrible.

So many elements in the navigation UX are a similar color.

The worst is the Current route (light blue) and alternate routes (grey). Both of these colors are fairly similar and have very little contrast compared to the background. I have a hard time picking out the difference between a road (grey) and an alternate route (grey, slightly thicker line).

At night, when it switches to a dark color scheme, everything is a shade of dark blueish grey, with very little contrast.

The colors drive me crazy when looking at forest service roads in the mountains. National Forest land is some shade of green, and the tiny roads are light gray and there is virtually no contrast between the two. So if I want to be able to see the roads, I have to zoom in a lot, and then I can only see a radius of half a mile, so I have to scroll the map constantly to see where a road goes. Absolutely maddening. Presumably some designer is really pleased with how elegant and subtle these colors look, but without contrast it's useless.

It's not just tiny roads that are swallowed by green with the latest updates. I couldn't find my favorite state highway while trying to show it to a friend recently until I zoomed in so far as to lose all context of its route and surroundings.

That low contrast stuff is what bugs me. I often go out into rural areas. Try making out where the (white) minor roads are against the (almost white) background. Now, try that in bright sunlight!

A good mapping app would have different modes for this. Garmin GPS's are much more readable, which is one reason I recently bought one, even though GM in theory is better. GM is better at routing, but it is really hard to see the detail, which is dangerous if you are driving.

Try Waze [1]. Even though they are owned by Google, I find their maps much more readable for driving.

1. https://www.waze.com/en-GB/

The Maps feature I hate most on Android is the "answer a few questions about places you've visited by tapping on these colorful cards."

The UX here is one of the worst I've ever seen. Once you tap an answer you don't get a chance to change it, or even see what the question was you just answered. It's on to the next question.

Many of the questions are insane. Like "Safeway: can you buy meat here?" Hello? Safeway? It's a giant supermarket!

But one design feature constantly leads me to giving wrong answers: every card is a different color, even for questions about the same establishment.

These frenetic color changes overpower the actual business name, so it's easy to get in a rhythm of "Safeway: can you buy meat here? Yes. Can you buy paper towels here? Yes. Can you get an oil change here? No, of course not!"

Oops, what? Did that card say Safeway or Jiffy Lube? Well the card's gone now, no way to go back and check it or fix it.

The color changes are a distraction from feeling confident which business you were answering a question about. With the lack of undo, they make me feel stupid. I like software that makes me feel smart.

I have never seen these, but I wouldn't answer them if I did. Does it not give you a way to just use the map instead?

Oh, it's not something you have to do, and never prevents you from using the map. I think it comes up as an Android notification like "help people by answering a few questions about the places you visited."

I like to help people, so I tried it out for a while. I would probably still be answering questions if they didn't make it so annoying! :-)

I empathize greatly with these complaints - honestly, this is one of those threads I'm kind of happy to see appear on HN since I know some of the feedback might actually make it back to people who can make an impact. I also feel like Google Maps has spent the last few years in a backslide from a really incredible example of the promise of the web to surprisingly frustrating and not useful. I'm somewhat mystified that while the "discover around you" section of the app has been iterated on nothing has been done to improve the actual experience of searching for places on the map. I really hope someone builds an alternative like you suggest - I want to be able to look for bars near a music venue and see the walking distance within the results, for example. Why can't I search for a lunch spot within 5 blocks of the office I'm meeting in and without any steep hills on the way?

Google maps 2012 refactor was what made it amazingly useful to me. The street names are much more readable now, they actually had a decent breakdown of the before and after and how the street names are much more readable overall now that they switched off of a "dumb" text fitting model vs their scaled text and "smart" placement of street names.

I wonder if your area has more trouble than most areas due to street shapes/density? The before and afters for me everywhere I've traveled and where I live are a huge step forward compared to old maps.


I don't note that it's changed much from this 2012 release.

Another gripe, in the US they have gotten into the habit of calling street names by their state route number (that nobody knows), rather than the common name (that is on the street signs).

On the other hand, a couple times Apple Maps has screwed me over badly by trying to route me down closed roads. Google seems to have much better knowledge of road closures.

The one that hurts me the most, search for a location I know is in town, Google decides the best match is somewhere across the US and zooms me there.

Recently searching the name of a town overseas in a browser detects my IP, pops up a local map and returns 'overseas place not found'

It works normally if you bother to include the country, but telling me a place doesn't exist isn't how I'd expect it to behave

> I can no longer loosely type something like "10th and grove to Jim's Hardware" and get directions

I've had the same issue, and I find it bizarre. It's really become a much less natural interaction.

A couple of years ago I implemented the Google Maps location API which at the time said it would give you a consistent location_id for each physical location in the world, a couple of years later and a bug report is still open and it's still not consistent. It's to the point now where I won't touch the Google Maps API anymore because of the price and lack of fixes.

Justin's theory is that the design focuses on a clean (empty?) look for mobile users. And on POIs, not streets.


* I searched for "chinese restaurant". Why is it showing me taco shops at all? I don't care that there are several open this late at night, I want to see my _search results_.

Wow. You just nailed everything that is wrong with google maps and I didn’t even realize it was happening. Frog in boiling water.

Have you tried Waze? It shows you the street you’re on and the incoming perpendicular streets.

> Much, much slower than it used to be.

It may be using OpenGL to display it now, you could try disabling webgl in your browser to check.

The appearance is different when drawn with webgl than when it is disabled so it is fairly easy to see if that is the cause.

One of the real travesties of our time is the active, continual sacking of the libraries of Alexandria inherent in every public website and application which is lost to the sands of adtech, paywalls, and the changing whims of anyone called "stakeholder"

Like, it's a metaphorical question I must give you, but who owns the information that we as a people produce? Take away the considerations of current law, and think about it in a different paradigm. Imagine that we had some impending apocalyptic event heading towards us, and we needed to as a people work together instead of competing. (Just a rough thought experiment, not that exactly, but just come along for a bit of a ride)

In that circumstance, the human population of earth forms this more cohesive collection, and in that frame, a group of people who produce some great work of art and invention and function, and then they or other people ruin it or destroy it, or simply bury it underground. In the frame of humanity as a collective being, that happening is akin to if you have some brilliant idea that you're scared of whether it's good or not, and so you write it down and decide it's not good and throw it away, but it could have been the seed of the creation of cold fusion, or romeo and juliet, or whatever.

The metaphor breaks down quickly, but this societal, special (as in species) loss is such a cancer on us as a people, and I wish there was a way to remove its crippling effects. But then again, it's akin the balance between to the law of nature and selection, and human societal safety nets. It would be easier without it, with a safety net built around everyone, but would we be inspired to such greatness? Or, what things wouldn't have been produced without the incentive of billions? Are you sure that number is zero? Please, don't take my positions as argumentative in the debate sense. This post is part complaint and part thought experiment. I also wonder what brilliant art we've lost because a child died of malnutrition or ~~malattention~~* neglect.

* instead of erasing "malattention" the unnecessary word I just coined†, I'm striking it with pen, so that it exists. Yeah, that's a bit meta and on the nose, but I like it.

† 148 Google Results at time of writing‡, so "independent of, but created after, a few others"

‡ Receipts: https://web.archive.org/web/20191214060156if_/http://web.arc...

By the way, I really wanted to delete this instead of clicking "reply", so: You're welcome/sorry.

This is off topic. OP is about the Maps API. Perhaps submit as a separate post.

If only we were paying for it, then we could legitimately complain.

We are paying for it in:

* personal data & attention

* opportunity cost (b/c their market domination deters investment interest in alternatives)

* advertising costs (each $ that businesses have to spend on advertising is a $ taken from R&D, profits which get reinvested in the economy, and/or consumers' pockets)

Just because you don't pay directly with money doesn't mean there isn't a cost.

Sorry, but this seems like nonsense to me. You are telling me that the curtailing of your attention by an app which you voluntarily install and voluntarily use is a ‘cost’ such that you are a stakeholder in the app?

Just gonna leave this here -- https://markmanson.net/attention

This is a great article. But it’s irrelevant to this discussion.

Oh I can complain just fine. Just because a corporation monetizes via a different means than selling software doesn't mean that I forfeit my ability to gripe about it (see Facebook).

They've also taken all the air out of the room such that there's very little ability for competitors to gain traction. So it's not like there's really many real options out there.

I don't really see a problem in complaining about a software product, especially when:

  - It was actually BETTER before
  - We're basically giving them a list of high-priority UX stories to make their product better
(edit: formatting)

I switched from Google Maps to Apple's Mapkit JS. It's awesome. And they respond to bug reports and make quick fixes. https://developer.apple.com/maps/web/

Like, Feedback Assistant reports?

Taking the power e.g. search, the profits from one domain e.g. advertisement and going into another domain e.g. 2-D maps and charging nothing or little until competition has thinned and dominant position has been achieved and then raising prizes tenfold is the sort of behavior regulators find interesting.

Unfortunately US regulators are too sissy (or probably already in bed with the big corps) unlike the old days where Microsoft was getting sued left and right due to antitrust violations.

This is a great write-up - thanks to the author and for their continued work in this space to bring home the consequences of the use of nuclear weapons.

Leaflet is an amazing library and I've found it a very useful and powerful tool with a great plugin ecosystem. Glad they managed to land the page someplace that was less dear to them monthly.

It definitely should not cost 10 cents per visitor to use the maps API to toy around with nuclear scenarios.

The idea that he could apply for a “grant” to escape a poor pricing model is insulting and I think good for him for just dumping Google instead. I look forward to seeing how far the 2019 Google backlash goes.

Too bad the backlash will only be among the modern archmages than hang around HN. The average people whom I know still get a warm fuzzy feeling from The Google. They're largely shielded from the obvious problems the pricing models. As long as the peasants don't have to pay with their pocketbooks, not much will change.

Maybe, but I see a lot of the censorship and demonetization on YouTube getting attention first from conservatives and second from lesser liked candidates like Tulsi, the antitrust investigations in USA and EU, and some of the Facebook and Twitter criticism being rightly shared by Google. I would say “fuck Google” is extending beyond geeks.

We've already got a co-ordinated YouTube Walkout which, whilst small, is international. It might go further… though I'm looking forward more to people properly beating Google by doing what they do better, and without spying on people to do it.

Good ol' internet boycotts...

Jack up the price, give grants, and deduct the grants from your taxes.

Google maps is only profitable on the basis of advertising syndication & data collection. If you're not pumping google information about how people are using your map service or directing people to the sites of google's ads customers, they are probably losing money by serving you.

The issue is similar to the one that ruined their custom search offering: it doesn't make them enough money to justify its existence as anything more than a public service. Unless charity is what Google had in mind when they launched the api, their behavior makes business sense.

Google has 100,000 employees not one can offer up anything in defense of their pricing? That's sad, I'm sure they have some explanation for what they did, and why they did it, and how they want things to evolve from here. Would be nice to hear something even if it was PR laden.

This pricing hike is over a year and old and everywhere I see complaints and about it and not a shred of explanation from them.

I'd expect something like "Yes Google Maps was free for a long time, but it's a very expensive service to run, for the service to exist long term we needed to increase the cost. Here are some resources how to contain your costs. Here is how to apply to be a non-profit."

I have to believe there are hundreds of people at Google who fully understand this has ruined many people's projects, do they not feel some obligation to at least open up a channel of communication? Seems like a no-brainer.

Or are they just so embarrassed and so wracked with in-fighting that they can't stomach even authoring a post?

I'm sure Google employees are probably not allowed to comment on HN without legal approval. BigCos have a lot of red-tape for communication.

> [B]ut they do have a way to request discounted access to the Google Cloud Platform, which appears to be some kind of machine-learning platform

This line in the article made me laugh out loud. GCP is a full blown competitor to AWS, but it sounds like whatever marketing page the author landed on is hyping machine learning so much that they didn't even realize that.

In this post the author speaks a lot to how inaccessible Google is for any kind of support. This is one of the under appreciate reasons NOT to use services provided by the big cloud providers: AWS, GCP, etc.

I recently started a small SASS project for a client and decided that Heroku would be a good fit for the situation. That led me to use smaller independent service providers like Mailgun for email, Mapbox for maps, Transloadit for video encoding. I haven't needed much support but when I have I actually talk to real people (and developers on the other end). It's a very pleasant experience. I get much more of a feeling that they actually care about developers using their products. Plus it feels "good" to be supporting independent shops, instead of giant tech monoliths.

Consolidated billing can be annoying if you decide to go with many smaller, independent providers. At least Heroku covers that well.

IDK about the other cloud providers, but if you have the support on your AWS account for businesses the support is excellent, skilled and fast. We have had support doing TCP dumps and other advanced troubleshooting whenever we've had a problem with their services. Google for their free* services is terrible yes. Which I don't use their services for anything important. I'd imagine they are better with GCP.

there's a lot of evils about AWS, but even their personal/developer support account has delivered good service for me -- they've really dug in and helped me on some tough questions, and not only limited to configuration.

At the office, we have AWS business level support, which has been really excellent for troubleshooting.

Agreed. Their business support has more visibility into your account than you do. They’ve been superb at finding issues fast and explaining things better than I can phrase the question.

You imagined wrong :)

Before using GCP I thought highly of them. Not I actually understand why they are number 3 or 4, well deserved.

I pay a ridiculously small $100/month for AWS support and they get the job done however I want it (chat, voice, email) at any hour of the day. I basically signed up for it so I'd have someone to blame if something went wrong, but they resolve every single issue to my satisfaction.

I don't think it's that small, it 10% of your bill for business support. ($100 if you are spending less that $1k).

And the support is ok sometimes. But when you actually discover an issue on their end they constantly try and flog you off, and then don't get back to you for ages.

I've yet to replicate that latter observation, having worked with AWS as an individual, part of a small company, and (currently) part of a recently-unicorn-statused startup. AWS support has always been pretty acutely responsive. This is obviously better if your company's big enough to shell out for enterprise support, but even the middle support tier (the one presumably being discussed here) ain't bad at all.

I used to use the free tier for the 1 year period, turned everything off, and have an alert set to go off at 1 penny of spend. Even with a lifetime spend of <$1 I still get to talk to real people for support at AWS.

You pay a small monetary amount for now.

Most of these companies are super powers with blatant disregard for humanity. When it comes to the good in your life, the things around you; opportunities, options, community health... they are slowly but surely drinking your milkshake.

That's a little extreme.

I got hooked on Unix when as a dev/DBA I inherited Solaris administration that I wasn't really qualified for. Sun support bootstrapped my skills and taught me alot. In another gig I had similar experiences with IBM.

It contrasted heavily against Microsoft, where even the $$$ Premier offerings are pretty awful, with lots of ridiculous hoops between you and super-skilled SEs.

Companies may suck or not, but that doesn't mean their support organization sucks.

On the contrary, if you find a company with a policy of gaining maximum market share through time on a break-even cost structure, and if the market is very big, you can essentially ride “at cost” through the entire growth/capture era.

If you notice, since inception AWS consistently returns its economies of scale as pricing drops to customers, allowing AWS to capture more market share.

They’re drinking the milkshake of incumbent IT, and you can benefit like a remora riding a shark.

I honestly don't know what you're saying here. Please explain.

It's crazy the extent to which for big companies this experience is different.

The large tech company that I work for has AWS developers& support lingering in dedicated Slack channels on our company Slack where they will help talk you through any issues and keep you updated on feature requests/bugs moving through the Amazon system.

yeah but Enterprise level support with TAM's cost upwards of 15k a month. That's pretty prohibitive for most small startups

You can pay $100/10% a month on Amazon and get good support. They will debug and guide you on. Hiring someone with the expertise you get would cost magnitudes of order more. And if your enterprise is at the level where it warrants 15k a month support, it is still worth it. Considering that still would barely get you only one skilled person hired.

This service is incredibly useful, my newly built EC2 instances rebooted randomly and I couldn't figure out why, so I opened up the chat, and they helped me debug, identify and mitigate the issue.

It turned out to be a bug between the newest Linux kernel available for Amazon Linux 2 and EC2, the mitigation was downgrading the kernel until an update was available.

I've worked in a lot of startups and have never had to pay that whole price until we had decent revenue. The AWS Activate program [1] has been pretty solid for my current company.

There's a big joke going around that AWS is basically a VC that invests in startups with free services and then makes their money back on revenue once they've grown.

1. https://aws.amazon.com/activate/

Actually, I had a great experience with Google Firebase support.

Suddenly, my apps were unable to connect to Firebase, after making sure that it's not me doing something wrong I contacted support and they came back to me in minutes, Together with the support engineer we debugged the issue, turned out that it was a configuration deployment error on their part and got if fixed.

What I am terrified of is having an issue with a free service that I depend on, like Gmail.

Google one is a support plan for consumer services. With some added benefits such as extra drive space. I had a good experience with them.

On this line, shoutout to Vultr for their support. Just today I opened a ticket with a relatively simple technical issue. I got a reply in 5 minutes ... yes 5 minutes, I couldn't believe it. My business with them so far is worth about 2 dollars, they will get many more from me, for sure!

I was using Retool and was looking for a feature.

Mentioned it in the chat window and within 60 seconds I got response, 10 minutes later the feature was pushed to production.

The velocity of startups can be really nice

I have always found AWS support to be pretty fantastic.

We have enterprise support at work with a named TAM who is amazing.

On my personal projects I have standard support and was able to speak to a very very capable tech on a sunday morning within 20 minutes.

Azure are pretty decent too, although they offload a lot of enterprise support to their partner network, but that works well enough.

GCP on the other hand...

Judging by the AWS support role interviews, I must say that they hire fantastic individuals who care about cases and the tech.

I think this is more a factor of percent of total earnings.

For example, my consumer and small business experience with Microsoft support is very negative. But working in an enterprise that always elects for the most expensive support package... I have never experienced better or more personal support than from Microsoft.

> That led me to use smaller independent service providers

I believe in doing this as well. These companies make or break themselves by delivering one product, and so you can trust they're going to do a very committed job of it. In comparison, services from Google seem like they're done by rotating groups of people who don't really care about what they're doing, and there's no longevity in the services unless they're super profitable.

I don't kno about AWS or GCP, but I've used Azure's technical support a few times over the years, and I've always been happy with the feedback, and the competence of the person providing it.

I'd assume GCP's support is poor, but I admit I'm basing that bias solely on my AdAwords support experience, which is and has always been, ludicrously bad.

I find that support for free products is impossible, but support for paid products is actually pretty good. For example, I was able to chat with a google checkout person to cancel an order pretty quickly, and when I had a domain issue, I was able to chat with a google domain person very easily as well.

I guess you get what you pay for.

I paid for the $150/month support option for gcloud and found it very good - they had engineers go over the code I wrote to help debug an issue on three different occasions, for example.

Isn't Heroku using AWS under the hood? So you are still supporting AWS?

You've stumbled across the Fox and Rabbit Product paradigm: https://www.teamten.com/lawrence/writings/fox-and-rabbit-pro...

AWS support is in general nice and helpful. GCP, abysmal indeed.

Since the Google Maps price hike, there are a number of such projects that have got affected. However I am curious why no one seems to be considering Bing Maps (from Microsoft) at all??

Bing Maps actually has a pretty decent free usage quota that should fit most such projects (either free or low cost).

Not even Azure uses Bing Maps. Instead of using their own existing mapping platform, Microsoft created Azure Maps, complete with a totally independent data store, rendering, etc.

Doesn't exactly inspire confidence.

I miss the Bing maps that was innovating and delivering regularly, but I haven't taken them seriously since 2016 at least; this is coming from a former Bing fanboy.

edit - links:

* Azure Maps: https://azure.microsoft.com/en-us/services/azure-maps/

* Bing Maps: https://www.microsoft.com/en-us/maps/choose-your-bing-maps-a...

From their copyrights, it looks like

- Azure uses TomTom

- Bing uses HERE (formerly Navteq/Nokia)

Did Microsoft buy all of Nokia? Or just their phone business?

Microsoft only bought parts. HERE has been its own thing for a while now.


Related - Uber hired/bought many of the engineers and assets involved in Bing Maps' data: https://techcrunch.com/2015/06/29/uber-acquires-part-of-bing...

>But in 2018, Google changes its pricing model, and my bill jumped to more like $1,800 per month.

Presumably, Google has done the market research here and had found that this level of pricing will end up maximizing revenue. It would be interesting to have a generally accepted model for these no money now, money later business models based on some measure of industry lock in. That way it would be possible to do the pricing at the start of a project rather than part of the way through when everyone had forgotten to budget anything.

The Googles of the world get to know how much our projects are going to cost, why can't we at least get an idea?

That presumes a project starts with people who intend to charge from the get-go, but defer that step to later. Rather than being started by folks who love what they're doing, getting funneled into a PM'ed project where the enthusiasm is curbed by "product viability", and then goes through a period of a few years where the original project folks rotate out until no one left is invested in how useful the product is, and a lot of folks are invested in how much money can be made by pivoting to a model that heavily monetizes on scoping down to purely the most used features used by its biggest, financially secure customers.

That reminded me of a project where person leading it estimated that it will be cheaper and faster to use AWS, they only forgot all the hidden fees that AWS adds, most commonly the network fees (which are charged between AZs) or EBS etc.

On top of that an instance that claims to have specific parameters won't deliver the same performance as supposedly equivalent physical machine, so you will need to use bigger instances etc.

Ultimately the project did not have resources to set up DR version which was originally planned and still went way over the budget.

We switched to Mapbox as well, the only downside we faced was the horrendous satelitte maps. They apperantly bought a lot of satellite imagery, but it still looks way way worse compared to Google or Apple Maps. The weird thing is that when zooming in it gets pixelated, even though the maps actually look more crisp when not zoomed in completely.

The streets are pretty offset as well :(

If you're not afraid of breaching Google's ToS you can load Google tiles directly with leaflet. The tiles themselves have no API key requirement. Obviously not a proper solution both legally and technically as the routes are undocumented and can change at any time (tho I've seen only 1 major change in the past ~7 years).

Their documentation seems to be lacking as of late or at least the last time I checked. I was attempting to get geocoding working on a project and was having a hell of time getting the api keys issued with the correct permissions. Swapped over to mapbox and had everything working in a few minutes.

> They clearly don’t care about small developers. That much is pretty obvious if you’ve tried to develop with their products. Look, I get that licensing to big corporations is the money-maker. But Google pretends to be developing for more than just them… they just don’t follow through on those hopes.

That is really true. Google isn't for the small devs anymore, they're too big to care about us. They need to cross that trillion dollar mark and it seems everything's about profits right now.

Lesson: Don't even touch google apis unless you're a really large corporation willing to spend >10k per month. Even if you're a really large corporation, have a plan B, because google will still fuck you over with their APIs changing without much notice.

There is something called "Here maps" that's a pretty good alternative to Google Maps, Apple Maps, and Bing Maps for mobile phones. Obviously that's not what Nukemap is talking about, but for those who are frustrated with the shortcomings of those 3 apps, "Here WeGo" is a free Android/iOS app. It was developed by Nokia and then sold off.

I like it because I can download an entire state with just a few clicks. And the UI isn't cluttered by "explore your neighborhood" nonsense.


Here wasn’t really developed by Nokia; it existed as Navteq before Nokia bought it in the 00’s and was headquartered in Chicago. At some point at least they were a fairly big player in in-car navigation systems.

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