9 ms·

So, here's the weird catch-22. Or perhaps, the illusion of choice. There are two people who are in a photo. One wishes Facebook to use face recognition, the ot

So, here's the weird catch-22. Or perhaps, the illusion of choice. There are two people who are in a photo. One wishes Facebook to use face recognition, the other does not. Facebook will run face recognition on the photo regardless, but how will they know who wants to be recognized, and who doesn't? Simply, they have a model of your face already trained. And they recognize you in photos--you have no say whether or not they will. All this option does is hide the notifications when they do recognize you.

bamboozled 9y ago
I don't think they care about that unfortunately; This has been a long running concern of mine. You're involved with it even if you choose not to be. Doesn't feel right.
denzil_correa 9y ago
Privacy Leakage via Social (Co-Location etc.) Information - an important and active research area.
- acomjean 9y ago
  That’s an interesting point. Not on Facebook but they know who I am and who I know already (they sent me an invite to join years ago with a list of about 10 people I might know. I knew 9 of them...
electrograv 9y ago
Usually these systems are implemented via a generic face detector that first runs on the whole image, to locate all the faces that are present. After the faces are located, they’re matched against a set of candidates (which I would guess is an enumeration if all your Nth order friends [who don’t opt out of this], to reduce computational complexity). This isn’t the only possible way to do it, but I believe it’s the most common and effective approach. So it’s not at all unlikely that when running the identifier, they simply don’t match against people who say no to this. They’d never be recognized at all, and the system would just identify their face as an “unknown” identity. I cannot say for sure whether or not this is how they implemented it, but it seems likely.
- bjt2n3904 9y ago
  Anecdotal, but years ago I uploaded a picture of a friend. He's a bit of a hermit, and doesn't have social media accounts. The picture scrolled by several times that week, and Facebook asked me to tag who this was. (This was perhaps... 2010?) I uploaded perhaps twenty or so photos, but that was the only one they asked about. I haven't seen this for a while, but I used to get my friends pictures scrolling by. (Not even my own!) "Tag Becky in this photo? Yes/No". Clicking yes confirms what Facebook already knows. Clicking No tells them... what? That they were right, and it is Becky, but I don't want to tag her? What does not responding to the prompt mean?
  - bitL 9y ago
    If you want to fight it, tag them as wrong persons to mess up their training dataset. I am sure they would love that! :) Maybe even start a Facebook group "False positives" where people would do this for fun and to "stick it to the man" lol
    - bjt2n3904 9y ago
      I'm sure they're on the lookout for that, as well. If they have enough confidence in their model, I'm sure they'd raise an eyebrow at something that only has 30% confidence, say "thanks", and then simply flag my account to not be trusted to train their models.
      - bitL 9y ago
        That still sounds like win. If it gets momentum, they won't be able to recognize any new additions due to them being not trusted. Physique changes significantly roughly every 10 years as well, so if all they have are outdated photos... OK, Deep Reinforcement Learning might be used on reconstructed 3D models from younger photos to simulate aging process, but it might not be very accurate even if looking very realistic.
      - badwolf 9y ago
        Similarly to "suggested edits" for pages/businesses. You have an accuracy score as an editor based on how many of your suggested edits get accepted/approved. I'd be willing to bet my hat that they have a similar scoring for approving suggested tags - don't click them/ignore them complete for a few days and you'll probably start seeing that suggesting a lot less.
dalore 9y ago
Well your friend took the photo presumably and uploaded it to Facebook. Let's assume that the friend does know about face recognition and does want it enabled for their photos. Surely they have the right? What about all the face recognition going on in all my Google photos that are automatically uploaded to the cloud? Everyone of the people in my photos has no say in what I do with a photo I took.
jacquesm 9y ago
Which is why I do not want anybody to make any pictures of me, especially not if they are active on social media. There are exactly two pictures of me online, the one is 30 years old, the other about 10. The quickest way to get me pissed off is to point a camera at me.
- Larrikin 9y ago
  Do you spray paint all of the CCTVs on your daily route?
  - jacquesm 9y ago
    Do you always ask dumb questions? It's one thing to have CCTV active in places where you visit, quite another to have the social sharing lemmings plaster your face all over the web for all to see and datamine. What I want and what I get are not always the same things, but I'll do my level best to influence that which I can influence. Here, two older blogposts that may give you an idea of where I'm coming from: https://jacquesmattheij.com/if-you-have-nothing-to-hide https://jacquesmattheij.com/if-you-have-nothing-to-hide https://jacquesmattheij.com/trackers https://jacquesmattheij.com/trackers For the downvoters and flaggers: do you really believe parent asked his question in good faith?
    - sctb 9y ago
      > For the downvoters and flaggers: do you really believe parent asked his question in good faith? That's not implied by downvoting or flagging you for breaking the guidelines ("Do you always ask dumb questions?"). The thing to do with troll comments is flag, not feed. https://news.ycombinator.com/newsguidelines.html https://news.ycombinator.com/newsguidelines.html
- Joeri 9y ago
  Out of curiosity, aside from emotional distress, what harm are you afraid of?
  - 9y ago
cycrutchfield 9y ago
Do they have a model of your face already trained, or are they just using a generic facial recognition neural network to compute an embedding vector and then finding identities that are close to that vector in L2 distance or whatever?
- bjt2n3904 9y ago
  How do they discern the difference between someone without an account (ie: a new face), a poor result from the network, and someone who's opted out? Unless the default for everyone is "opt-out", which I highly doubt.
  - cycrutchfield 9y ago
    Maybe they use some sort of locality-sensitive hash to efficiently gather a set of (approximate) candidates that are close by in that embedding space, and then use some heuristics to further narrow down who the most probable match is.
Ajedi32 9y ago
Think about it this way: the photo itself already contains all the information necessary to identify you, and algorithms already exist which allow that information to be easily extracted. No privacy setting in the world will change that, short of the "delete" button on the photo itself (or maybe a way to blur your face). Therefore, this setting logically _cannot_ change what information Facebook has about you; it can only change how that information is displayed.
- SiempreViernes 9y ago
  If you have a book you certainly own all the information in it, but if you are never allowed to open the book and read its content (your brother wrote it but was then terribly embarrased about it and made you swear nobody would be allowed to read it) then how can you put the information to use?
  - Ajedi32 9y ago
    Facebook's servers are computers, not people. A human might read that book and be unable to control how they react to that information (maybe you might look at your brother differently if you knew what he wrote). In contrast, a computer could read the book, do a bunch of processing on it, run it through a few neural nets, check your brother's privacy settings, see that the book is marked as private, then erase all its "memories" of the book from existence; the result being absolutely no different from if it had never "read" the book at all. So even if it were true that "Facebook will run face recognition on the photo regardless", that would be irrelevant as long as it does nothing with the information obtained via that algorithm. And if you don't trust Facebook to honor your privacy setting and do nothing with the information, you shouldn't trust them to not run the facial recognition algorithm in the first place.
- bo1024 9y ago
  Good point, but the setting might change how that information is used, by Facebook or its users or advertisers (or subpoenas, etc).
- EGreg 9y ago
znpy 9y ago
> how will they know who wants to be recognized, and who doesn't? If they have enough data to recognize you, they will. They just won't tag you.
- bjt2n3904 9y ago
  That was meant to be a rhetorical question. ;) But I'm glad you see this too!
hrie282 9y ago
Welcome to an economy that is run like a game theory experiment not to service public economic interests Quantophrenia in action
kobayashi 9y ago
All of your comment is accurate. In addition, this is not news in the sense that this is not _new_. Anecdotally, this feature has been active for many for months, if not years.
yorby 9y ago
If they don't tell anybody who they recognized in the pictures, it should be fine since you will never know? They do a lot more analysis in the background that they never publish anyways

bamboozled 9y ago

I don't think they care about that unfortunately; This has been a long running concern of mine. You're involved with it even if you choose not to be. Doesn't feel right.

denzil_correa 9y ago

Privacy Leakage via Social (Co-Location etc.) Information - an important and active research area.

acomjean 9y ago
That’s an interesting point. Not on Facebook but they know who I am and who I know already (they sent me an invite to join years ago with a list of about 10 people I might know. I knew 9 of them...

electrograv 9y ago

Usually these systems are implemented via a generic face detector that first runs on the whole image, to locate all the faces that are present. After the faces are located, they’re matched against a set of candidates (which I would guess is an enumeration if all your Nth order friends [who don’t opt out of this], to reduce computational complexity). This isn’t the only possible way to do it, but I believe it’s the most common and effective approach. So it’s not at all unlikely that when running the identifier, they simply don’t match against people who say no to this. They’d never be recognized at all, and the system would just identify their face as an “unknown” identity. I cannot say for sure whether or not this is how they implemented it, but it seems likely.

bjt2n3904 9y ago
Anecdotal, but years ago I uploaded a picture of a friend. He's a bit of a hermit, and doesn't have social media accounts. The picture scrolled by several times that week, and Facebook asked me to tag who this was. (This was perhaps... 2010?) I uploaded perhaps twenty or so photos, but that was the only one they asked about. I haven't seen this for a while, but I used to get my friends pictures scrolling by. (Not even my own!) "Tag Becky in this photo? Yes/No". Clicking yes confirms what Facebook already knows. Clicking No tells them... what? That they were right, and it is Becky, but I don't want to tag her? What does not responding to the prompt mean?
- bitL 9y ago
  If you want to fight it, tag them as wrong persons to mess up their training dataset. I am sure they would love that! :) Maybe even start a Facebook group "False positives" where people would do this for fun and to "stick it to the man" lol
  - bjt2n3904 9y ago
    I'm sure they're on the lookout for that, as well. If they have enough confidence in their model, I'm sure they'd raise an eyebrow at something that only has 30% confidence, say "thanks", and then simply flag my account to not be trusted to train their models.
    - bitL 9y ago
      That still sounds like win. If it gets momentum, they won't be able to recognize any new additions due to them being not trusted. Physique changes significantly roughly every 10 years as well, so if all they have are outdated photos... OK, Deep Reinforcement Learning might be used on reconstructed 3D models from younger photos to simulate aging process, but it might not be very accurate even if looking very realistic.
    - badwolf 9y ago
      Similarly to "suggested edits" for pages/businesses. You have an accuracy score as an editor based on how many of your suggested edits get accepted/approved. I'd be willing to bet my hat that they have a similar scoring for approving suggested tags - don't click them/ignore them complete for a few days and you'll probably start seeing that suggesting a lot less.

dalore 9y ago

Well your friend took the photo presumably and uploaded it to Facebook. Let's assume that the friend does know about face recognition and does want it enabled for their photos. Surely they have the right? What about all the face recognition going on in all my Google photos that are automatically uploaded to the cloud? Everyone of the people in my photos has no say in what I do with a photo I took.

jacquesm 9y ago

Which is why I do not want anybody to make any pictures of me, especially not if they are active on social media. There are exactly two pictures of me online, the one is 30 years old, the other about 10. The quickest way to get me pissed off is to point a camera at me.

Larrikin 9y ago
Do you spray paint all of the CCTVs on your daily route?
- jacquesm 9y ago
  Do you always ask dumb questions? It's one thing to have CCTV active in places where you visit, quite another to have the social sharing lemmings plaster your face all over the web for all to see and datamine. What I want and what I get are not always the same things, but I'll do my level best to influence that which I can influence. Here, two older blogposts that may give you an idea of where I'm coming from: https://jacquesmattheij.com/if-you-have-nothing-to-hide https://jacquesmattheij.com/if-you-have-nothing-to-hide https://jacquesmattheij.com/trackers https://jacquesmattheij.com/trackers For the downvoters and flaggers: do you really believe parent asked his question in good faith?
  - sctb 9y ago
    > For the downvoters and flaggers: do you really believe parent asked his question in good faith? That's not implied by downvoting or flagging you for breaking the guidelines ("Do you always ask dumb questions?"). The thing to do with troll comments is flag, not feed. https://news.ycombinator.com/newsguidelines.html https://news.ycombinator.com/newsguidelines.html
Joeri 9y ago
Out of curiosity, aside from emotional distress, what harm are you afraid of?
- 9y ago

cycrutchfield 9y ago

Do they have a model of your face already trained, or are they just using a generic facial recognition neural network to compute an embedding vector and then finding identities that are close to that vector in L2 distance or whatever?

bjt2n3904 9y ago
How do they discern the difference between someone without an account (ie: a new face), a poor result from the network, and someone who's opted out? Unless the default for everyone is "opt-out", which I highly doubt.
- cycrutchfield 9y ago
  Maybe they use some sort of locality-sensitive hash to efficiently gather a set of (approximate) candidates that are close by in that embedding space, and then use some heuristics to further narrow down who the most probable match is.

Ajedi32 9y ago

Think about it this way: the photo itself already contains all the information necessary to identify you, and algorithms already exist which allow that information to be easily extracted. No privacy setting in the world will change that, short of the "delete" button on the photo itself (or maybe a way to blur your face). Therefore, this setting logically _cannot_ change what information Facebook has about you; it can only change how that information is displayed.

SiempreViernes 9y ago
If you have a book you certainly own all the information in it, but if you are never allowed to open the book and read its content (your brother wrote it but was then terribly embarrased about it and made you swear nobody would be allowed to read it) then how can you put the information to use?
- Ajedi32 9y ago
  Facebook's servers are computers, not people. A human might read that book and be unable to control how they react to that information (maybe you might look at your brother differently if you knew what he wrote). In contrast, a computer could read the book, do a bunch of processing on it, run it through a few neural nets, check your brother's privacy settings, see that the book is marked as private, then erase all its "memories" of the book from existence; the result being absolutely no different from if it had never "read" the book at all. So even if it were true that "Facebook will run face recognition on the photo regardless", that would be irrelevant as long as it does nothing with the information obtained via that algorithm. And if you don't trust Facebook to honor your privacy setting and do nothing with the information, you shouldn't trust them to not run the facial recognition algorithm in the first place.
bo1024 9y ago
Good point, but the setting might change how that information is used, by Facebook or its users or advertisers (or subpoenas, etc).
EGreg 9y ago

znpy 9y ago

> how will they know who wants to be recognized, and who doesn't? If they have enough data to recognize you, they will. They just won't tag you.

bjt2n3904 9y ago
That was meant to be a rhetorical question. ;) But I'm glad you see this too!

hrie282 9y ago

Welcome to an economy that is run like a game theory experiment not to service public economic interests Quantophrenia in action

kobayashi 9y ago

All of your comment is accurate. In addition, this is not news in the sense that this is not _new_. Anecdotally, this feature has been active for many for months, if not years.

yorby 9y ago

If they don't tell anybody who they recognized in the pictures, it should be fine since you will never know? They do a lot more analysis in the background that they never publish anyways