Live data from Hacker News

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

nianticlabs.com

321–330 of 365 posts

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

#321

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"). The…

After skimming the article I don't understand what the Large Geospatial Model is supposed to be, what it can be used for.

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

#322
post #285

Earlier quoted context omitted.

Hiding shady or unexpected stuff in the TOS is illegal in the EU and other countries for example. So just because some companies behave amoral, that doesn’t mean we just have to accept hundreds of pages of legalese being able to dictate us.

I don’t think there is something amoral here. Niantic explicitly sends players to take videos of places for rewards. It’s not like it’s done in a sneaky way. Being somehow surprised they actually plan to do things with the data they have you gather is a bit weird.

No, it isn't. Stop normalizing this behavior. There was no consent. You expect that you are playing a game, not working for them for free.

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

#323

Earlier quoted context omitted.

Sarcastically, no one should. Unsarcastically, a lot of people believe user data belongs to users, and that they should have a say in how it's used. Here, I think the point is that Niantic decided they could use the data this way and weren't transparent about it until it was already done. I'm sure I would be in the minority, but I would never have played - or never have done certain things like the research tasks - h…

> 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.

> I can understand that people believe this, but why do they do? Nothing in our society operates in a way that might imply this.

Sure, but that disconnect between what people think and how things work is almost fully general over all subjects.

I've seen people (behave as if they) think translation is just the words, but that leads to "hydraulic ram" becoming "water sheep". People who want antibiotics for viral infections, or who refuse vaccines (covid and other) claiming they're "untested" or have "side effects" while promoting alternatives that both failed testing and have known side effects. I've seen people speak as if government taxation only exists because the guy in charge of taxes is, personally, greedy. I've heard anecdotes of people saying that you can get people to follow the rules by saying "first rule is to always follow the rules" and directly seen people talk as if banning something is sufficient to make it stop.

The idea that it's even possible to do make a model like this from the user data, is probably mind-blowing to a lot of people.

The naïve assumption most people seem to have is that computers do only what they, personally as end-users, tell them to do, and that they're as slow as the ad-riddled web front-end with needlessly slow transition animations placed there to keep user engagement high — rather than the truth, that software primarily does what the operator of the service wants it to do, and that it's absolutely possible for a home PC[0] to hold and query a database of all 8 billion people on the planet and the two trillion or so different personal relationships between them.

When GenAI images were new, some of the artists communities said "That content generated can reference hundreds, even thousands of pieces of work from other artists to create derivative images"[1], rather than millions of images, because the scale of computer performance is far beyond the comprehension of the average person. The fact that the average single image contributes so little to any given model that it can't even represent its own filename, even moreso.

And so it is with stuff like this: what can be done, cannot be comprehended by the people who, theoretically, gave consent that their data be used in that way.

[0] Of course, these days most people don't have home PCs; phone, perhaps a tablet, they may have a small low performance media server if they're fancy, but what we here would think of as a PC is to all that as a Ferrari etc. is to a Honda Civic.

[0] https://www.furaffinity.net/journal/10321622

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

#324
So what they are doing is not different from previous "VPS" systems, its how they are doing it.

What is a "VPS" At its heart, Visual Positioning Systems are actually pretty simple. You build a 3d point cloud of a place, with each point being a repeatable unique feature that can be extracted from an image (see https://blog.ekbana.com/extracting-invariant-features-from-i...) Basically a "finger print"/landmark of a thing in real life that can be extracted from an image reliably.

To make that work, you need to generate a large map of these points: https://www.researchgate.net/figure/Sparse-point-cloud-Figur... Which basically involves taking lots of pictures with GPS tags on where they are. Google has the advantage of street view, Niantic has it's game. Others had to pay a bunch of people to go round a city with cameras.

Once you build that pointcloud (which isn't actually that easy, you can't do it all at once, and aligning point clouds is hard.) you can then use trigonometry to work out where a picture is. This is called "re-localization" which is a stupid name. The hard part is the data management. There are billions of points in the world, partitioning the database so that you can quickly locate a picture is the hard part.

Hence this approach, which is basically "train a model to do it for us" You still get a "VPS", you still need all that data, but they hope that a model will able to optimize for speed.

is it private?

No, the original system isn't private. If they've done their job properly, then nothing identifiable will be in the "map" as thats extra data you dont need. What they do with the raw photos, and the metadata that they contain is another matter.

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

#325

Earlier quoted context omitted.

They have never been a mobile game company and they have said as much themselves on many occasions. They're a data harvesting company. Guess now they're trying to figure out what to do with all of that data.

> They're a data harvesting company. I didn't realize this. I used to submit so many "portals" to ingress T_T

Thank you for your service. The robot who will replace DoorDash workers will prioritize your food deliveries higher.

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

#326
post #322

Earlier quoted context omitted.

I don’t think there is something amoral here. Niantic explicitly sends players to take videos of places for rewards. It’s not like it’s done in a sneaky way. Being somehow surprised they actually plan to do things with the data they have you gather is a bit weird.

No, it isn't. Stop normalizing this behavior. There was no consent. You expect that you are playing a game, not working for them for free.

I absolutely agree with you that this should not be the norm. The fact is that "they" absolutely do it and even give you "rewards" for your behaviour and actions in the free game. Reminds me of a certain opiod crisis, but now it is combining software with the human phyche almost directly.

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

#327
post #191
post #20

This is pretty cool, but I feel as a pokehunter (Pokemon Go player), I have been tricked into working to contribute training data so that they can profit off my labor. How? They consistently incentivize you to scan pokestops (physical locations) through "research tasks" and give you some useful items as rewards. The effort is usually much more significant than what you get in return, so I have stopped doing it. It's…

“If you're not paying for the product, you are the product” (I realize you can pay, but are not required to)

Nowadays you are the product either way unless you avoid pay via pirating.

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

#328
post #181

Earlier quoted context omitted.

> All companies should be truthful, forthcoming, and specific about how they will use your data, but… I'm fairly sure, if you read the terms-and-conditions, it probably said that the company owns this data and can do what they want with it. > There’s no reason to hold out for a company to pay you for your geolocation data because none of them offer that service. Well, it can make perfect sense (to some people) to hol…

> terms-and-conditions I would argue that's being legally truthful, but not practically truthful. The company knows there are ways they can ensure their consumers are aware of the truth. And they know that burying it in Ts and Cs isn't one of them.

I'm inclined to agree with your distinction in general. But not in this particular case:

Everybody knows, even without actually bothering to read the terms-and-conditions, that they will say that the company owns all the data. Letter and spirit agree.

In some sense reading the the T&Cs might actually be detrimental to your understanding: you might misinterpret the carefully lawyered language to conclude that there are certain limits to what the company can do with your data. But they are probably way better than you at interpreting legal terms they crafted themselves.

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

#329
post #322

Earlier quoted context omitted.

I don’t think there is something amoral here. Niantic explicitly sends players to take videos of places for rewards. It’s not like it’s done in a sneaky way. Being somehow surprised they actually plan to do things with the data they have you gather is a bit weird.

No, it isn't. Stop normalizing this behavior. There was no consent. You expect that you are playing a game, not working for them for free.

Of course there was consent. There is even an explicit EULA listing in plain writing that you are actually collecting data for them that people have to agree to before playing.

That people suddenly wake up to the fact that they were dumb for providing labour for worthless virtual gifts doesn't magically allow them to claim it was abuse post-fact.

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

#330
post #117

Earlier quoted context omitted.

> I have been tricked into working to contribute training data so that they can profit off my labor You were playing a game without paying for it. How did you imagine they were making money without pimping your data?

Niantic made 700 million dollars last year, mostly selling virtual game items.

But some numbers pusher somewhere saw an opportunity to make even more money and write good quarter number, padding themselves on the shoulder for a job welll done, without ever wasting a thought about any such unimportant thing as ethical implications...
Post reply on HN