3 ms·

> The implicit message was clear: “Yes, we have all of your data, but the fact we have all of your data is a good thing, because it allows us to make your life

> The implicit message was clear: “Yes, we have all of your data, but the fact we have all of your data is a good thing, because it allows us to make your life easier.” This is why a) free/libre software and b) Moore's law are important. Trading privacy for convenience is a real trade you can make right now, and nobody benefits from pretending it isn't--the first step to fixing that problem is to admit it exists. But there's no inherent reason this has to be done "on the cloud" (which is just a marketing phrase for "on other people's computers"). As an industry, I would like to see us putting more computing power into end users' hardware and more intelligence into freely available software so that more of this work can be done on people's personal computers without giving their data to amoral corporations.

oblio 8y ago
Moore's Law died 5 years ago. And services completely negate Open Source.
gnode 8y ago
It's not really about it being cloud based. It's about having a central trusted party with access to private data which has public utility. As an example, Tesla collects lots of private data about their users' driving, in order to train their self-driving tech. Doing this in a free software solution (if you consider the training to be part of the software) would be less practical, as it would require training journeys to be publicly available. The way out of this situation is to develop competitive solutions which are less private data dependent.
- kerkeslager 8y ago
  > Doing this in a free software solution (if you consider the training to be part of the software) would be less practical, as it would require training journeys to be publicly available. I think most free software folks would agree that the training data doesn't have to be public. I do agree that there's some problem to be solve here, though. Part of the point of free software is to be able to trust that the code you are running is serving your interests. Right now I don't know of any proofs of concept of manipulating what an AI does by manipulating the training data, but it's a very plausible attack vector.
  - gnode 8y ago
    > I think most free software folks would agree that the training data doesn't have to be public. That seems antithetical to me. The trained model is the program, and the training data is in-essence the source code. > manipulating what an AI does by manipulating the training data You don't even have to do this. You can manipulate the model directly. Writing neural networks manually isn't difficult. You can't do more with that than conventional programming, but its sufficient to add malicious behaviour.
    - kerkeslager 8y ago
      > That seems antithetical to me. The trained model is the program, and the training data is in-essence the source code. No, I think that trained models and training data are entirely new concepts. I can see the similarities between trained models/programs a and training data/source code, but there are obvious privacy concerns with training data that don't apply to source code. Calling the training data the source code is a leaky abstraction. You could, for example, equally argue that the trained model is the source code, and the program which operates on the trained model is an interpreter which runs that source code. Even if you refuse to concede that there are differences between training data and source code, you'll note that I am talking about free (libre) software, not open source. There are lots of cases where free software doesn't mean opening all the source up. Free software is about freedom, not about open source. For example, while there's plenty of source code in settings files, I don't think there's anyone arguing that we need to check in our settings files on open source projects. The point is to put the power in the hands of users. Putting users' data out of their control is the antithesis of that. > You don't even have to do this. You can manipulate the model directly. Writing neural networks manually isn't difficult. You can't do more with that than conventional programming, but its sufficient to add malicious behaviour. That's an interesting point, and may provide a workaround making training data public, i.e. you can generate the model and then make a representation of the model public, rather than the training data. However, there's an assumption here that the training data can't be reconstructed from the trained model--if a study of this exists, I'm not aware of it.
      - 8y ago
segfaultbuserr 8y ago
And c) cryptography is important. Or maybe also, d) P2P and federation is important? (a) ensures, although only partially, the computing freedom of the user, (b) ensures, although only partially, the computing power of the users, and (c) ensures, although only partially, the ability of the user to selective reveal information, and/or reserve control of them, when it's in hands of a 3rd-party. Unfortunately, even with (a), (b), and (c). I doubt if a general solution exists, as I said in https://news.ycombinator.com/item?id=19861630 https://news.ycombinator.com/item?id=19861630.
Notorious_BLT 8y ago
The average user isn't willing to pay more for their devices just to avoid the seamless sharing of their data with google. There's a reason people stopped buying desktop computers and are relying on mobile devices and cheap laptops for more: its cheaper and they don't care about the hidden "costs" of their data being shared. In addition, sure, people could absolutely create local, non-cloud ways to accomplishing a lot of this. But who is footing that bill? Where's the value proposition to anyone who has the manpower
- kerkeslager 8y ago
  Just because something doesn't make anyone money doesn't mean it doesn't have value.

oblio 8y ago

Moore's Law died 5 years ago. And services completely negate Open Source.

gnode 8y ago

It's not really about it being cloud based. It's about having a central trusted party with access to private data which has public utility. As an example, Tesla collects lots of private data about their users' driving, in order to train their self-driving tech. Doing this in a free software solution (if you consider the training to be part of the software) would be less practical, as it would require training journeys to be publicly available. The way out of this situation is to develop competitive solutions which are less private data dependent.

kerkeslager 8y ago
> Doing this in a free software solution (if you consider the training to be part of the software) would be less practical, as it would require training journeys to be publicly available. I think most free software folks would agree that the training data doesn't have to be public. I do agree that there's some problem to be solve here, though. Part of the point of free software is to be able to trust that the code you are running is serving your interests. Right now I don't know of any proofs of concept of manipulating what an AI does by manipulating the training data, but it's a very plausible attack vector.
- gnode 8y ago
  > I think most free software folks would agree that the training data doesn't have to be public. That seems antithetical to me. The trained model is the program, and the training data is in-essence the source code. > manipulating what an AI does by manipulating the training data You don't even have to do this. You can manipulate the model directly. Writing neural networks manually isn't difficult. You can't do more with that than conventional programming, but its sufficient to add malicious behaviour.
  - kerkeslager 8y ago
    > That seems antithetical to me. The trained model is the program, and the training data is in-essence the source code. No, I think that trained models and training data are entirely new concepts. I can see the similarities between trained models/programs a and training data/source code, but there are obvious privacy concerns with training data that don't apply to source code. Calling the training data the source code is a leaky abstraction. You could, for example, equally argue that the trained model is the source code, and the program which operates on the trained model is an interpreter which runs that source code. Even if you refuse to concede that there are differences between training data and source code, you'll note that I am talking about free (libre) software, not open source. There are lots of cases where free software doesn't mean opening all the source up. Free software is about freedom, not about open source. For example, while there's plenty of source code in settings files, I don't think there's anyone arguing that we need to check in our settings files on open source projects. The point is to put the power in the hands of users. Putting users' data out of their control is the antithesis of that. > You don't even have to do this. You can manipulate the model directly. Writing neural networks manually isn't difficult. You can't do more with that than conventional programming, but its sufficient to add malicious behaviour. That's an interesting point, and may provide a workaround making training data public, i.e. you can generate the model and then make a representation of the model public, rather than the training data. However, there's an assumption here that the training data can't be reconstructed from the trained model--if a study of this exists, I'm not aware of it.
    - 8y ago

segfaultbuserr 8y ago

And c) cryptography is important. Or maybe also, d) P2P and federation is important? (a) ensures, although only partially, the computing freedom of the user, (b) ensures, although only partially, the computing power of the users, and (c) ensures, although only partially, the ability of the user to selective reveal information, and/or reserve control of them, when it's in hands of a 3rd-party. Unfortunately, even with (a), (b), and (c). I doubt if a general solution exists, as I said in https://news.ycombinator.com/item?id=19861630 https://news.ycombinator.com/item?id=19861630.

Notorious_BLT 8y ago

The average user isn't willing to pay more for their devices just to avoid the seamless sharing of their data with google. There's a reason people stopped buying desktop computers and are relying on mobile devices and cheap laptops for more: its cheaper and they don't care about the hidden "costs" of their data being shared. In addition, sure, people could absolutely create local, non-cloud ways to accomplishing a lot of this. But who is footing that bill? Where's the value proposition to anyone who has the manpower

kerkeslager 8y ago
Just because something doesn't make anyone money doesn't mean it doesn't have value.