Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Niantic plans a “Large Geospatial Model” trained on Pokémon Go player data

nianticlabs.com

Post reply on HN
Print view View original

251–260 of 365 posts

Re: Niantic plans a “Large Geospatial Model” trained on Pokémon Go player data

#251

Post by dgfitz »

virodoran wrote: post #242
dgfitz wrote: post #11

Earlier quoted context omitted.

Pokemon Go is built on the same engine as Inverness I think its called. When it launched they even used the same POIs. I think this was ~5-7 years before PGO launched. Edit: I said inverness and meant ingress. Apologies.

Pokemon Go was launched on the Unity game engine in 2016. Ingress was using a different game engine at the time, and wasn't rewritten into Unity until several years later. Even the backend/server side was significantly different, with them needing to write a shim to ensure compatibility during & after the move to Unity.

Surely the game engine has little to do with logged telemetry data. No?

Re: Niantic plans a “Large Geospatial Model” trained on Pokémon Go player data

#252

Post by vachina »

CaptainFever wrote: post #7

This title is editorialized. The real title is: "Building a Large Geospatial Model to Achieve Spatial Intelligence" > Otherwise please use the original title, unless it is misleading or linkbait; don't editorialize. My personal layman's opinion: I'm mostly surprised that they were able to do this. When I played Pokémon GO a few years back, the AR was so slow that I rarely used it. Apparently it's so popular and commo…

Google branded AR glasses. Not any AR glasses.

Re: Niantic plans a “Large Geospatial Model” trained on Pokémon Go player data

#253

Post by fromMars »

interroboink wrote: post #212
PittleyDunkin wrote: post #138

Earlier quoted context omitted.

> Unsarcastically, a lot of people believe user data belongs to users, and that they should have a say in how it's used I can understand that people believe this, but why do they do? Nothing in our society operates in a way that might imply this.

> Nothing in our society operates in a way that might imply this. I beg your pardon? Consider just about any physical belonging — say, a book. When I buy a book, it belongs to me. When I read a book in my home, I expect it to be a private experience (nobody data-mining my eyeball movements, for example). This applies to all sorts of things. Even electronic things — if I put some files on a USB stick I expect them to…

Before smart phones or the rise of the internet your information was mined by credit agencies for use by banks, employers and other forms of credit lending.

Credit cards and Banks sold your data to third parties for marketing purposes.

Payroll companies like ADP also shared your data with the credit agencies.

This is not a new phenomenon and has been the currency of a number of industries for a while.

The only thing that has changed is the types of data collected. Personally, I think these older forms of data collection are quite a bit more insidious than some of the data mining done by a game like Niantic for some ml model.

I have a lot more control over and less insidious consequences from these types of data collection. I can avoid the game or service if I like. There isn't much I can do to prevent a credit agency from collecting my data.

Re: Niantic plans a “Large Geospatial Model” trained on Pokémon Go player data

#254

Post by try_the_bass »

IsopropylMalbec wrote: post #101
try_the_bass wrote: post #91

Earlier quoted context omitted.

> which is totally unfair, every niantic player should have access to all the stuff because they collectively made it I don't understand this perspective. While all players may have collectively made this model possible, no individual player could make a model like it based on their contributions alone. Since no single player could replicate this outcome based on only their data, does it not imply that there's value…

What you say is fair but if an individual's data doesn't matter, what happens when they ask to have their data deleted under GDPR. is there a way to demux their data from existing models?

While your example isn't exactly coherent (I don't think GDPR would cover photos/videos taken by the user, unless maybe the user was in the photo/video?), presumably they could just train the model again without that user's data. I doubt the end result would be that much different

Re: Niantic plans a “Large Geospatial Model” trained on Pokémon Go player data

#255

Post by reissbaker »

I'm confused by both this blog post, and the reception on HN. They... didn't actually train the model. This is an announcement of a plan! They don't actually know if it'll even work. They announced that they "trained over 50 million neural networks," but not that they've trained this neural network: the other networks appear to just have been things they were doing anyway (i.e. the "Virtual Positioning Systems"). They tout huge parameter counts ("over 150 trillion"), but that appears to be the sum of the parameters of the 50 million models they've previously trained, which implies each model had an average of... 3MM parameters. Not exactly groundbreaking scale. You could train one a single consumer GPU.

This is a vision document, presumably intended to position Niantic as an AI company (and thus worthy of being showered with funding), instead of a mobile gaming company, mainly on the merit of the data they've collected rather than their prowess at training large models.

Re: Niantic plans a “Large Geospatial Model” trained on Pokémon Go player data

#256

Post by mightyham »

sunshadow wrote: post #221
alwayslikethis wrote: post #175

Earlier quoted context omitted.

IIRC Google maps basically does not make money. I wonder if there can be a government deal to subsidize it on the condition that the data be open sourced.

They made 11B$ last year. It has incredible amount of ads. If you haven't noticed, then that means they did a great job. (tip: look for the custom logo pins in the map. Its printing money)

While I have no way to validate this, I highly suspect that the routing algorithm is also subtly manipulated. There is a route I drive with regular frequency that contains a roughly 20 mile section of two mostly parallel roads, one for through traffic and one for local. Every single time I drive through, Google routes me to the local traffic road. I know for certain the local road is slightly slower and it's also simply incorrect. The only way it makes sense is if it's a bug, or my hunch is that Google weights the route a little higher because it goes by a bunch of businesses that pay for advertising.

Re: Niantic plans a “Large Geospatial Model” trained on Pokémon Go player data

#257

Post by viknesh »

eru wrote: post #182
JohnMakin wrote: post #166

Earlier quoted context omitted.

It may surprise you to learn pokemon go is nearly a 10 year old game based on 40 year old beloved IP that when it was released did not exist in the same data hellscape we do today, and even if it did, the attraction of the IP would overrule people thinking about this kind of thing. These kinds of comments are extraordinarily disingenuous sounding, particularly when anyone can spend 3 seconds and figure out their prim…

> [...] when it was released did not exist in the same data hellscape we do today [...] That was fairly obvious at the time. And people used more or less exactly the same language to describe the world back then, too. > These kinds of comments are extraordinarily disingenuous sounding, particularly when anyone can spend 3 seconds and figure out their primary market is literal children. Poke Mongo was popular with peo…

I believe Google explicitly stated that they used data collected from Ingress (arguably a predecessor to Pokemon Go) at the time. It's the reason Niantic was founded. It's hard to take these complaints seriously.

Re: Niantic plans a “Large Geospatial Model” trained on Pokémon Go player data

#258

Post by fromMars »

8note wrote: post #219
thfuran wrote: post #179

Earlier quoted context omitted.

Wanting something to be a certain way is very different from believing that it is. And yes, I would expect any moderately informed and technically savvy user to assume that the company is doing anything they possibly can to profit off of user data.

But you don't expect people to also try to profit off whatever said company is doing?

Sure they can, but niantic offered a free game, I already got some profit out of playing it.

Re: Niantic plans a “Large Geospatial Model” trained on Pokémon Go player data

#259

Post by virodoran »

dgfitz wrote: post #251
virodoran wrote: post #242

Earlier quoted context omitted.

Pokemon Go was launched on the Unity game engine in 2016. Ingress was using a different game engine at the time, and wasn't rewritten into Unity until several years later. Even the backend/server side was significantly different, with them needing to write a shim to ensure compatibility during & after the move to Unity.

Surely the game engine has little to do with logged telemetry data. No?

Perhaps, perhaps not - I have my theories, but is that not what you meant when you said Pokemon Go was built on the same engine as Ingress?

I do think it wasn't until after Pokemon Go launched and they saw the success of it, that they shifted focus to be more of a platform for these types of experiences (see Niantic Lightship). Additionally, I think Unity offered them the opportunity to integrate with ARCore and collect much more detailed data than they would've ever been able to do on the old Ingress engine. In fact, I expect a significant chunk of ARCore functionality was added specifically thanks to Niantic and Unity (in fact, you see Unity mentioned all over the Google Developer docs for it).

Re: Niantic plans a “Large Geospatial Model” trained on Pokémon Go player data

#260

Post by Syonyk »

PittleyDunkin wrote: post #230
interroboink wrote: post #212

Earlier quoted context omitted.

> Nothing in our society operates in a way that might imply this. I beg your pardon? Consider just about any physical belonging — say, a book. When I buy a book, it belongs to me. When I read a book in my home, I expect it to be a private experience (nobody data-mining my eyeball movements, for example). This applies to all sorts of things. Even electronic things — if I put some files on a USB stick I expect them to…

> Consider just about any physical belonging — say, a book. When I buy a book, it belongs to me. When I read a book in my home, I expect it to be a private experience (nobody data-mining my eyeball movements, for example). Perhaps this is just my own brain's degradation, but how far removed from society do you need to be to expect your purchases to not be sold to the highest bidder? This practice is certainly older t…

> Perhaps this is just my own brain's degradation, but how far removed from society do you need to be to expect your purchases to not be sold to the highest bidder? This practice is certainly older than I am.

It depends quite a bit on how you make your purchases.

If your purchases are on a credit card, with a loyalty ("tracking") card or App(TM) involved in the purchase? They're absolutely being sold to... well, probably not the highest bidder, but "all bidders with a valid payment account on file."

If you make a habit of paying cash for things and not using Apps or loyalty cards, and don't have your pocket beacon blaring loudly away on a range of radio frequencies when you shop, I expect a lot less data sales. It's a bit of a transition if you're used to credit cards, but once you're used to it, it's not bad at all, and involves a lot less data collection. I don't mind if the local barista or bartender knows me and my preferences, but I do mind if their POS system is uploading that data continuously.

Post reply on HN