5 ms·

I've become totally superstitious (convinced, even) that phones are listening to more than just ultrasound. There have been too many uncanny situations where y

I've become totally superstitious (convinced, even) that phones are listening to more than just ultrasound. There have been too many uncanny situations where youtube or google have recommended results that were far too specific to the conversation happening moments before. For example, one time I was telling my girlfriend about fourier transformations, something she'd never heard of before. She typed "Fo" into her phone. "Fourier transormation" it suggested. Maybe this was due to the IP address we were connected through. That alone is creepy enough as it is. Sometimes I wonder if it isn't something more though. Facebook messenger and google voice search are my number one suspects. But moreso, all it takes is a single person in a room full of phones to have installed an app that's using the microphone unethically, and suddenly everyone's conversation is being turned into marketing/spy data. Anybody who's connected somehow to that location is going to be statistically associated ("implicated") with this data.

scottlegrand2 9y ago
No common friends, but college girlfriend recommended as potential friend on Facebook is my creepiest data mining moment ever.
- anchpop 9y ago
  If she searched for you, it can make you show up in the recommended friends list
- H4CK3RM4N 9y ago
  I think that works based on location. A few years ago there were reports of people who shared a psychiatrist being recommended to each other, based on their frequent visits to the office.
- moioci 9y ago
  What about friends who are two degrees of separation? I suspect your social graphs are closer than you might realize.
- elFarto 9y ago
  I've had this with someone who I emailed (completely separate from Facebook) being recommended to me. The only logical conclusion I could come to is that Facebook was mining their emails for contacts.
- legolas2412 9y ago
  She might have been stalking you..
krmboya 9y ago
Also experienced something like this. Had a discussion with a work colleague on a particular topic, and on getting home and going to youtube, the top recommended videos were all on that topic
IanCal 9y ago
These are like modern day ghost stories. Do you really think they're running constant speech recognition on all phones, apparently unnoticed so far, just to recommend YouTube videos?
- craftkiller 9y ago
  I agree that we should remain skeptical, however, one important distinction between this and ghost stories is this is already fully technically possible. All that's needed is the will, the raw processing power, and the battery life to support it. No one could ever create a proof of concept for ghosts.
  - macawfish 9y ago
    This comment is an argument that it's possible, not a claim that it's happening. It'd be much, much easier for apps like FB messenger to transmit sparse samples of low bitrate audio to a server for processing there. 1 hr/day @ 8kbps ~ 110 MB/month. As a matter of fact, with "deep learning" and whatall, it doesn't seem like the audio would even need to be intelligible to human ears for it to still be useful in some marketing algorithms or to generate guesses of interests correlating to the audio. Those interests might be presented as an ad, search suggestion, etc. Why would google use something like this for search recommendations? Because it's training data, that's why! What I'm arguing is that even though conventional "speech recognition" is processor intensive and unlikely happening on a mobile client side, it's just not that far-fetched that small snips of low bitrate conversation data might be used to tune marketing prediction models on the server-side. In training general purpose models, huge amounts of data could be generated by only a little bit of data/user. Where would this sort of thing cross a privacy line? I'm not sure. Personally, it kinda creeps me out regardless. Seriously, listen to these 8kbps opus samples: http://www.opus-codec.org/examples/ http://www.opus-codec.org/examples/ . If they're intelligible to human ears, I can't imagine what some clever learning algorithms could do with that and much much less. Here's a random anecdote about how much data FB messenger can use: https://www.reddit.com/r/androidapps/comments/49yhlh/facebook_messenger_background_data_usage_4gb_in_a/ https://www.reddit.com/r/androidapps/comments/49yhlh/faceboo...
  - 9y ago

scottlegrand2 9y ago

No common friends, but college girlfriend recommended as potential friend on Facebook is my creepiest data mining moment ever.

anchpop 9y ago
If she searched for you, it can make you show up in the recommended friends list
H4CK3RM4N 9y ago
I think that works based on location. A few years ago there were reports of people who shared a psychiatrist being recommended to each other, based on their frequent visits to the office.
moioci 9y ago
What about friends who are two degrees of separation? I suspect your social graphs are closer than you might realize.
elFarto 9y ago
I've had this with someone who I emailed (completely separate from Facebook) being recommended to me. The only logical conclusion I could come to is that Facebook was mining their emails for contacts.
legolas2412 9y ago
She might have been stalking you..

krmboya 9y ago

Also experienced something like this. Had a discussion with a work colleague on a particular topic, and on getting home and going to youtube, the top recommended videos were all on that topic

IanCal 9y ago

These are like modern day ghost stories. Do you really think they're running constant speech recognition on all phones, apparently unnoticed so far, just to recommend YouTube videos?

craftkiller 9y ago
I agree that we should remain skeptical, however, one important distinction between this and ghost stories is this is already fully technically possible. All that's needed is the will, the raw processing power, and the battery life to support it. No one could ever create a proof of concept for ghosts.
- macawfish 9y ago
  This comment is an argument that it's possible, not a claim that it's happening. It'd be much, much easier for apps like FB messenger to transmit sparse samples of low bitrate audio to a server for processing there. 1 hr/day @ 8kbps ~ 110 MB/month. As a matter of fact, with "deep learning" and whatall, it doesn't seem like the audio would even need to be intelligible to human ears for it to still be useful in some marketing algorithms or to generate guesses of interests correlating to the audio. Those interests might be presented as an ad, search suggestion, etc. Why would google use something like this for search recommendations? Because it's training data, that's why! What I'm arguing is that even though conventional "speech recognition" is processor intensive and unlikely happening on a mobile client side, it's just not that far-fetched that small snips of low bitrate conversation data might be used to tune marketing prediction models on the server-side. In training general purpose models, huge amounts of data could be generated by only a little bit of data/user. Where would this sort of thing cross a privacy line? I'm not sure. Personally, it kinda creeps me out regardless. Seriously, listen to these 8kbps opus samples: http://www.opus-codec.org/examples/ http://www.opus-codec.org/examples/ . If they're intelligible to human ears, I can't imagine what some clever learning algorithms could do with that and much much less. Here's a random anecdote about how much data FB messenger can use: https://www.reddit.com/r/androidapps/comments/49yhlh/facebook_messenger_background_data_usage_4gb_in_a/ https://www.reddit.com/r/androidapps/comments/49yhlh/faceboo...
- 9y ago