3 ms·

I don't like your project. Data aggregation is powerful but very dangerous and I feel that you are handling it carelessly. I also get a strong Borg vibe here.

I don't like your project. Data aggregation is powerful but very dangerous and I feel that you are handling it carelessly. I also get a strong Borg vibe here. In the current state, people don't know how to handle their data correctly. You are just making it easier for some people siphon other's people data even more. And having their data played against themselves. Most people are not quants. And becoming one is not something trivial. Having people play unguided with their own data is like giving uranium to little kids. They will have a great fun !!!

karlicoss 6y ago
> making it easier for some people siphon other's people data even more I guess you mean someone could easily steal your data, etc? I agree, it's a tradeoff, would be interesting to explore how to make it more secure. > Having people play unguided with their own data is like giving uranium to little kids Well, handling Uranium is a bit different. Can you elaborate on some specific examples? I've seen people misusing statistics, and drawing misleading conclusions, sure. But even more people are using anecdata and broscience, which I think is worse.
- GistNoesis 6y ago
  Not even stealing some data is needed. Just sharing publicly some data, may leak information that can be used against you. Bots do build profiles and will happily ingest any information you give. If your top ten "most interesting Slate Star Codex" contains certain keywords, they will flag you in their database. >Can you elaborate on some specific examples? -People having sport injury from those sport performance training app. Their knowing of their data made them push-themselves past their limits. -Little kids getting bad marks, and assuming they are just bad and not persevering. Their knowing made thing worse. -Hypochondriacs becoming ill from the stress of all the illnesses they think they have. -People doing weight loss badly focusing too much on the scale. -Musk tweets :) -People censoring themselves from knowing that having an unpopular position will cost them karma points. As soon as you make a system self reflective, you impact it in profound ways. You generate some reinforcing feedback loops. That tend to either break the stability resulting in chaos, or tend to lock it into a specific behavior from which it's hard to escape.
  - karlicoss 6y ago
    Ah I see what you mean. Yeah, absolutely, an aggergate system would enable such behaviours. I guess it comes down to a personal level of paranoia. And to the old question "Is ignorance bliss?", for which people had very different answers long before the software.
adwi 6y ago
It’s infantilizing to imply people are incapable of handling the responsibility of having access to the personal data they’re creating, as if it’s better to just trust it with private companies that leverage it to monetize us. I’m sure achieving whatever your sufficient level of “quant” isn’t trivial (congratulations presumably) but a command line interface took shared with the Hackernews community is fairly self-selecting. I think this is a super interesting project, looking forward to seeing how this gets developed!
jborichevskiy 6y ago
> you are handling it carelessly From the post: > Everything is local first, the input data is on your filesystem. If you're truly paranoid, you can even wrap it in a Docker container. > There is still a question of whether you trust yourself at even keeping all the data on your disk, but it is out of the scope of this post. > If you'd rather keep some code private too, it's also trivial to achieve with a private subpackage. This seems as safe, if not safer, than the average service where a rogue employee can read/write to your data with little to no visibility. > You are just making it easier for some people siphon other's people data even more. The data is already being siphoned off of them -- usually through vague legal agreements and opaque security infrastructure they can't audit. At least with this model, a user can see the whole scope of data a service is collecting and act accordingly: whether by limiting the data collection as needed or by switching to another service.

karlicoss 6y ago

> making it easier for some people siphon other's people data even more I guess you mean someone could easily steal your data, etc? I agree, it's a tradeoff, would be interesting to explore how to make it more secure. > Having people play unguided with their own data is like giving uranium to little kids Well, handling Uranium is a bit different. Can you elaborate on some specific examples? I've seen people misusing statistics, and drawing misleading conclusions, sure. But even more people are using anecdata and broscience, which I think is worse.

GistNoesis 6y ago
Not even stealing some data is needed. Just sharing publicly some data, may leak information that can be used against you. Bots do build profiles and will happily ingest any information you give. If your top ten "most interesting Slate Star Codex" contains certain keywords, they will flag you in their database. >Can you elaborate on some specific examples? -People having sport injury from those sport performance training app. Their knowing of their data made them push-themselves past their limits. -Little kids getting bad marks, and assuming they are just bad and not persevering. Their knowing made thing worse. -Hypochondriacs becoming ill from the stress of all the illnesses they think they have. -People doing weight loss badly focusing too much on the scale. -Musk tweets :) -People censoring themselves from knowing that having an unpopular position will cost them karma points. As soon as you make a system self reflective, you impact it in profound ways. You generate some reinforcing feedback loops. That tend to either break the stability resulting in chaos, or tend to lock it into a specific behavior from which it's hard to escape.
- karlicoss 6y ago
  Ah I see what you mean. Yeah, absolutely, an aggergate system would enable such behaviours. I guess it comes down to a personal level of paranoia. And to the old question "Is ignorance bliss?", for which people had very different answers long before the software.

adwi 6y ago

It’s infantilizing to imply people are incapable of handling the responsibility of having access to the personal data they’re creating, as if it’s better to just trust it with private companies that leverage it to monetize us. I’m sure achieving whatever your sufficient level of “quant” isn’t trivial (congratulations presumably) but a command line interface took shared with the Hackernews community is fairly self-selecting. I think this is a super interesting project, looking forward to seeing how this gets developed!

jborichevskiy 6y ago

> you are handling it carelessly From the post: > Everything is local first, the input data is on your filesystem. If you're truly paranoid, you can even wrap it in a Docker container. > There is still a question of whether you trust yourself at even keeping all the data on your disk, but it is out of the scope of this post. > If you'd rather keep some code private too, it's also trivial to achieve with a private subpackage. This seems as safe, if not safer, than the average service where a rogue employee can read/write to your data with little to no visibility. > You are just making it easier for some people siphon other's people data even more. The data is already being siphoned off of them -- usually through vague legal agreements and opaque security infrastructure they can't audit. At least with this model, a user can see the whole scope of data a service is collecting and act accordingly: whether by limiting the data collection as needed or by switching to another service.