Niantic's apps collect a surprising amount of data about where users go | Hacker News Reader

Niantic's apps collect a surprising amount of data about where users go

kotaku.com

114 pointskaikai55 comments

I worked at Niantic (originally on Pokemon Go) since the day they officially spun out from Google until a few months ago.

I am speaking solely for myself when I say that, While the company has plenty of flaws, it is very serious about privacy protections. It was drilled into every engineer from day one that we were dealing with really sensitive data. Not just any old data - kids were a target audience and so we were dealing with realtime location data from children under 13.

Niantic took a lot of time and effort to ensure that this data was deleted, obfuscated or has as much precision removed as possible, as quickly as possible.

I don’t like how Niantic handled some things, but their stance on privacy was never one of them.

(Again, this is my personal opinion. I definitely don’t speak for Niantic.)

I was very active in the Ingress community around that time and while I appreciate their efforts, I don't think they were very successful in this regard. Using IITC (3rd party Ingress planning suite) we could track people's paths pretty well, to the point where you could easily see how others ran their routes and could intercept them if you wanted to say hi, or something more nefarious. I know at least one woman in the community changed their account because of a stalking problem.

I didn't totally mind the lack of location privacy since the game made it pretty clear you could run into people if you really wanted, but I wouldn't give them flying marks on the privacy issue. For Pokemon Go I'm sure they were far more careful, but Ingress really felt like the Wild West.

IITC was explicitly a violation of the ToS and Niantic made efforts to foil it and ban those identified as using it until player outrage made them give up. The nature of the game also was antithetical to the notion of privacy since you could predict player movements just on the basis of field layouts. Anyone playing in an area also quickly learned the approximate locations of home and work for other players solely on the basis of habitual game activity. I mean, the whole point of the game was geolocation, it seems odd to me that anyone would have some kind of problem with that for Ingress.

On the last point, Go has the same situation, where playing the game to its full extent requires giving up a lot of information to be publicly displayed.

Gym ownership in particular is needed to get in game coins (required fo basic things like extending pokemon storage), and will reveal a lot about someone’s life patterns to anyone willing to put efforts in knowing.

It gets worse with “friends”, it’s a very powerful leveling up mechanism, but gives the same insight in a more granular and extensive way.

From what I can read you've never been stalked by a creep…

I think the point of the comment is that by playing the game, you broadcast pretty sensitive data. It would be like if you were constantly uploading geotagged photos for everyone to see, and being surprised about people learning your location.

I really hate that this sounds like 'victim blaming', but if a threat vector for you is being stalked by a creep, and you don't want people to know where you are, then maybe voluntarily constantly broadcasting your location to the public isn't the best idea?

What if you innocently play the game and THEN get a stalker, who uses game data to identify work, home, etc.

If anybody can get a stalker, then nobody is safe playing this game... ?

That's a matter for police then. Any thing involving location data and semi-public information could land you a stalker.

Anything can happen to anyone. Whether to care about it at all, involve the state or solve it solve it yourself depends on probabilities involved.

So privacy doesn't matter? And the police are the right solution for all invasions of your privacy? Do I get the local police involved if remote attackers use personal information to drain my bank account?

"The police" (local?) are a solution for certain kinds of problems, but that's it.

Consider restraining orders and all the victims these orders failed to protect. Far more effective is a large dog or other adult humans.

> Niantic took a lot of time and effort to ensure that this data was deleted, obfuscated or has as much precision removed as possible, as quickly as possible.

Define "as quickly as possible". How long after the fact can Niantic tell that I was at Burger King for lunch?

If the answer is anything more than a week, I don't think they are really trying. Pokemon Go does need to keep some data longer because it's used in-game: when and where you caught a Pokemon that you still have; whether you have ever visited a given Pokestop (but not when); how long you've had Pokemon in a given gym. But the game never needs to save actual paths you followed.

They store the fact that someone was there are a particular time but they strip the userid immediately.

They have a GPDR data-request page if you are interested in seeing exactly what data they store about you.

> They store the fact that someone was there are a particular time but they strip the userid immediately.

This appears untrue, according to the article: "In five days of gameplay, Niantic kept 2304 location records for one player. [...] When we asked them about their propensity to eat Burger King for lunch, they were surprised that we knew that"

This is only possible if the userid is still linked to these location records.

> They have a GPDR data-request page if you are interested in seeing exactly what data they store about you.

I know. That's what the featured article used to get the data referenced above.

Wow. It’s not often you see making someone eat their own words executed so well. Professionally handled.

How is this a surprise? The original nitanic red-v-blue game was totally about geo-location.

Interesting read but the hook of surprising doesn't grab me. What amazes me is that I can't find a page to do experimental A/B testing in these users for $.

Because it is a paid-for app instead of a gratis one. People expect they "paid off" the privacy invasion that way. Turns out, they did not.

That's an interesting assumption of people, and I can see where they might come to that, but the logic is unfortunately flawed. Just because publishers/developers might feel the need to monetize user data on free apps to make money, doesn't mean they won't also do it for paid apps to make more money.

Betting a business is going to leave money on the table because it's the nice thing to do is not a winning strategy on the consumers part.

> Because it is a paid-for app instead of a gratis one.

That's not correct. All Niantic games are free to play. In-app purchases are available, but in reasonably densely populated areas you can play just fine without them, there is enough "stuff" to collect in the wild.

Private data of people with a demonstrated willingness to pay for things is the most valuable kind. So it’s kind of the opposite: by paying you make data theft more likely, not less.

I guess it's suprising if you've never looked into it, but even back when their only game was Ingress, it was a gamified way of giving them location data that they can sell to help optimize mapping software.

I guess I had my first real wake-up call about how much data is being collected and used was when I got an Android update, and google maps had pre-filled entries for my home and work addresses based on where I spent most of my time at certain parts of the day.

Yes the entire purpose of Ingress was to make meat modules walk to under-documented locations, take pictures, and indicate points of interest. The meat modules seemed to think it was "fun". Hard to explain.

Was a player, but to this day I still can't figure out why "Go there, take pictures, and 'capture' a location" can be THAT beneficial when it come to map software optimization.

I know the game certainly records my location which maybe good for discovering walk route, but other data they've collected seems (I guess) useless. How can picture of sculptures improve map quality? Kind curious.

A level 40 player (deeply committed to get that high) creates a stop with a photo of a statue in his back yard fountain so he could cheat. You could wait until multiple people submit it, or just allow it and just see how many people visit it. Either way, you get validation data on it.

Conversely, a player creates a stop for a new statue that was just placed outside of a government building in the last few weeks. This stop gets visited hundreds or thousands of times a day. You've now identified a hub. This hub can be used to validate the existence of other items of interest in the near vicinity.

It makes more sense now, but I still got questions.

You see, when you created an in-game location, you're inviting players to go there, regardless whether or not the location is popular in the real life. Dedicated players may take a detour from their normal day-to-day route just to hit that location to gain in-game advantages, which will generate data noise if the map software wants to use players data to improve their product for normal non-player users.

Also, let's don't forget products like Google Maps and Bing Maps are already collecting users locations. And in addition to the location, it also knows where the user was, where the user wants to go and how the user would go there (Take bus, drive, walk etc). That, is high quality data (Better than a Niantic game could provide) that generates profit directly.

All the reasons combine, I think I still have some doubts about that business model.

A couple points here.

First, you get data about how pedestrians can reach all the truly popular locations. If you also get extra data about how pedestrians can reach various irrelevant locations, okay, that doesn't have value but that's not harmful; what matters is that all kinds of niche landmarks that would be interesting to tourists are somewhere in your data.

Second, people making a detour from their normal day-to-day route to reach some location is the whole point - these are local people with knowledge of how their city works; you don't want to measure how tourists usually get from landmark A to landmark B; you'd want to see what shortcuts an optimizing local would take when getting from landmark A to landmark B - which is different than simply looking at random people walking habits with no intent to optimize the route; the gamification provides an incentive to optimize routes and allows you to harvest the knowledge of that optimization.

If someone tells google maps that they want to go from A to B, and google maps tracks how they got there, then it doesn't harvest any information about the best route because the user doesn't know it, they wanted a recommendation and likely tried to follow it even if it's very suboptimal, so you'll just get a reprocessed version of the data you already had (and gave to that user). On the other hand the data from Niantic is useful so that Google maps can make a better recommendation.

Third issue is that the Niantic process also allows you to detect undesirable routes. If Google maps directs a tourist from A to B through a shortcut that's passable but unpleasant in some manner, then they'll take that route; and if they send another one there, they'll also go there, because they don't know better and the alternatives are (probably) not obvious. You'll only get a signal of people not going there if it's really bad e.g. impassable.

Niantic, however, can detect that most people who want to go from A to B (because of game incentives) but who know all the routes from A to B (instead of asking for directions like the google maps usecase) are intentionally taking a longer route for whatever reason - which is again useful data for improving Google Maps recommendations.

I imagine, as well as user verification (mentioned down-thread) that statues provide a known location to remove systematic errors from a users GPS data.

In my limited experience it's not been uncommon for a GPS track to run parallel to the actual track. Having known markers (that are small, and can be treated as points) means traces can be pinned.

Just a guess.

For one they created the dataset that Pokémon Go uses and that game is massively profitable.

I think it doesn't improve directly the maps, but it improves the whole experience of using Google maps. Maybe they could reuse the pics uploaded by ingress users and display them on Google maps.

Maps are usually static snapshots in time. Really good maps would constantly update to show changes in a living world as quickly as possible.

It was, in fact, fun.

Do you use a search engine? Social media? Any modern online game? A connected smart home? A smart phone without its privacy settings altered? A modern flat screen tv? A modern car? All of them do behavioral extraction in one form of another. Often under the guise of entertainment.

To single out Ingress or Pokémon Go seems to ignore the much bigger problem.

And, as an Ingress oldie, it was a lot of fun. We planned for weeks to cover the northern hemisphere, preparing with people in four different continents. I’ve met people and seen parts of the world I would never have met or seen otherwise.

Humans are so competitive that they'll often take pride in being the worst at something, because even that's a perverse way to be at the top of the list. That you can create an arbitrary game, seed it with some people to set up a ranking, and open it up to a bunch of people and they'll go crazy figuring out how to rank themselves using your criteria isn't all that surprising.

Fun anecdote about that: During crunch on a new android device (or release? I think it was a device), the new automatic pre-fill for home/work had decided that many android team members lived at the office and worked at home.

I got this "feature" at Google I/O. It promptly decided I lived at the Motel 6 and worked at the Moscone Center.

That's not related to the game. I got the same thing without ever playing it

An important point is that it doesn't just passively collect location data, it actively moves players to locations.

Yes, there have been instances where people have been trying to put portals into fenced off locations such as army bases or shipping yards. Useful to learn about what's there... and those pictures could be useful in conflicts too.

https://community.ingress.com/en/discussion/comment/19749 https://community.ingress.com/en/discussion/3117/portals-in-... https://www.reddit.com/r/Ingress/comments/alenbc/portal_in_r...

I really cannot imagine being surprised about this

"Wow! That is surprising!"

I gave it a shot. How did I do? It isn't very surprising no, I think the author just added that in to add a nefarious tone to the article.

> “If you have mobile computing and services people are using while moving around in the world,” said Hanke in a 2016 talk to the Berkeley School of Business, “the question we asked ourselves was, ‘Could these services influence how people behave in the physical world? Could the products they’re using cause them to walk a different path, drive a different path, divert from the trajectory that they’re normally going to go on? If you could do that through information services that you’re offering to people, there’s tremendous opportunity for businesses that might want to change the behavior of people to get them to go places where they wouldn’t otherwise go.” He added: “We were also interested in being an advertising company.”

I think it's crystal clear where this is going. Why you would even bother giving the excuse of "we accidentally added code to collect all this random metadata about you, sorry about that" given this is beyond me.

In case you forgot, Hanke was already working for the biggest advertising company in the world when he had that brilliant insight.

An free-to-play mobile game that requires GPS location to work collects location data. Also at 11, water determined to be wet.

Well not only that, but the whole game mechanic is that you have to get close to specific points of interest in order to activate them. Obviously for the game to work, they need to know you are at said point of interest...

It's like accusing the airline you used to knowing where you traveled to...

It's an observation, not an accusation.

And you missed the part where the airline makes most of it's money from passengers paying to be moved around, whereas Niantic makes nearly $0 from anything players give them money for.

>Niantic makes nearly $0 from anything players give them money for.

I am under NDA so all I can say is that publicly available data (sensortower or appanie for example) says Niantic has received literally billions of dollars from in-app purchases from players. This may or may not be correct but the amount is likely not “nearly $0”

Never underestimate how much people will spend on fancy hats (See Fortnight & Dota)

You don't really need insider info to know that they would obviously not be running a game this big if they made "no money".

There's a pokemon stop in the center of the pentagon courtyard area. You can pretty much only get there as a visitor or as someone related to the government / military.

If you want to talk about a really easy way to create a tracking system for senior military officials, just mark their young aides and you'll be able to find them internationally.

The military has been wrestling for years with the problem of troops broadcasting their movements to the world, intentionally and accidentally.

Centuries, indeed.