3 ms·

Rohan makes a valiant and useful attempt to expand the over-simplified notion of digital privacy. Anything that throws more light on this area is welcome. Howev

Rohan makes a valiant and useful attempt to expand the over-simplified notion of digital privacy. Anything that throws more light on this area is welcome. However I feel that some of the distinctions are incomplete or need highlighting more strongly. The word 'tracking' shouldn't be used to stand-in for "absence of privacy". For example, I may want to be tracked in every detail and might buy a GPS tracker. However it should remain under my exclusive control. If I find it's defective because it's treacherously uploading my data somewhere I didn't ask it to, that's a breach of privacy. The suggestion that techniques for web browsing might be generalisable to wider privacy doesn't hold up well. The main focus is mitigations (evasion and reduction) against cross-site identity leakage. Active obfuscation, avoidance, spoofing, dazzle, camouflage and decoying isn't covered, nor are threat actors or actor position. For example my ISP or device vendor may be a greater threat than a website (doubly so when the device and site are owned by the same entity eg. Google.) It is oft said that privacy means different things to different people, but this is not the same as saying people have different use cases and needs, and is rarely unpacked by socio/psychological analysis (different expectations and ethical judgements may exist within the same use-cases and needs). Also, someone "being okay" with a violation of privacy is not a sufficient indicator. Objective harms exist and they don't go away because the user is ignorant or convinced, or coerced to make "acceptable trade-offs".

dcow 4y ago
I read this as an attempt to explore the tension between people who want total anonymity and people who just don't want their personal information abused. Sure, there are companies that abuse your information and yeah that's a violation of privacy. But the way I am reading the argument, the solution to one company violating your privacy isn't necessarily a surefire: we should take one more step towards total anonymity. Rather, it depends on the threat model. > Also, someone "being okay" with a violation of privacy is not a sufficient indicator. Objective harms exist and they don't go away because the user is ignorant or convinced, or coerced to make "acceptable trade-offs". Curious what you'd consider an objective harm. For example, I've been in discussions where any stable identifier is objectively harmful because it could be used by a service to track you. Therefore we can't do things like mutual authentication since your cert has a globally unique name, allow signatures in the browser because your key is a global identifier, behave normally at layer 2 (thanks Apple, I can't manage devices on my network anymore because they all randomize their MAC), or find my nearest Lowes because now my traffic exits 5 states away because ~~Apple wants to own my data~~ of my big scary ISP who can't be trusted. Sadly the only conclusion I can come to is that these problems are social and, though technology may have introduced new means to abuse people's privacy and it's natural to seek a technology solution, ultimately require legal solutions. For example, "deleting" IP addresses from the internet is not an acceptable solution in the fight against privacy abusers. Nor is telling services they can't collect information that could identify you unless they need it because, well, everyone needs it at some level and it turns out everything can identify you in the right context. Punishing people who abuse others' privacy is the solution. That, or, technology would need to let me select a la carte exactly which identifiers I want to allow and which I don't and we'd have to live with most everyone never touching those toggles and blissfully remaining tracked and targeted. Frankly I'm sick of privacy nut technologists (who no doubt are genuinely concerned for everyone's safety on principle) continually pushing everyone towards "tracking evasion" when all that most people really care about is surface level "tracking resistance". That much, at least, I resonated with in the essay.
- nonrandomstring 4y ago
  > Curious what you'd consider an objective harm. To dodge the risk of opening a can-o-worms I don't have time for on a Monday morning I'll just say watch for the chapter "Harms" in Ethics for Hackers when it's out, or ping me a PM if you fancy being a proofreader for that chapter. > Sadly the only conclusion I can come to is that these problems are social I just replied to another similar remark. They are simultaneously psychological, social, technical, legal and political, and must be addressed on all fronts to reduce long-term harms upon the many to the temporary advantage of the few. > Punishing people who abuse others' privacy is the solution. It's a solution amongst many which is necessary but insufficient to cover all harms. An example already given in comments; of Nederlands collection of religious data prior to the Nazi rise, is apropos anachronous and displaced threats. > Frankly I'm sick of privacy nut technologists (who no doubt are genuinely concerned for everyone's safety on principle) continually pushing everyone towards "tracking evasion" I hear that. There's a real risk of alienating users and losing the psychological campaign for better opsec by harping on a "mother knows best" tune. For example, GDPR cookie popups seem to have had that negative effect. > That much, at least, I resonated with in the essay. Me too. I very much like Rohan's writing. It sparks great discussion.
  - dcow 4y ago
    > For example, GDPR cookie popups seem to have had that negative effect. Yes, but I don't think it's limited to "visibly" negative user experiences. Not being able to sign a document in the browser using an HSM-backed hardware key because vendors won't add that functionality because some nut on the mailing list is concerned about "supercookies" also is, arguably, harmful to users because it deprives them of stronger security applications (that might also lead to a stronger tracking resistance stance nonetheless by nature of the ability to sign and encrypt messages at the application level) in pursuit of a fuzzy goal of never letting anybody have a stable identifier, ever. That seems, in the very least assuming the supercookie concern is legitimate and not a vendor play at retaining platform control, something that can only be decided individually by the user and not the browser vendor. Another example is TLS 1.3. In order to have better privacy against unfriendly networks encrypted SNI pretty much kills content filtering at the last mile network level (can still do it on device, which is Apple's solution). This might be a great win for Joe who wants to browse Ashley Madison at the coffee shop away form prying eyes. But it's not great for parents who depend on content filtering solutions to help keep their kids safe on the internet. So, who really, is a technology like that harming? > To dodge the risk of opening a can-o-worms I don't have time for on a Monday morning I'll just say watch for the chapter "Harms" in Ethics for Hackers when it's out, or ping me a PM if you fancy being a proofreader for that chapter. I emailed you via book2021@. This, though, is really the crux of it for me (sorry I rambled about a bunch of peripheral things). It seems that ethically the idea of "remembering a customer to draw inferences about how they'd behave" is rather natural to how humans operate. Historically, I can't even tell you I don't want you to do that because how I conduct myself in public is, well, public. So on the individual scale I think a lot of literature needs to be developed around building an understanding of what exactly is wrong, regarding profiling in public, to do without a person's consent. Is it wrong to write down someone's name, for instance? Is it harmful to be so good at profiling someone based on immediately available information that you might learn some things about someone even they didn't really care to discover? It certainly seems invasive from a user perspective but then again so would a good shaman, psychic, magician, or psychologist who build careers on playing the statics and profiling games. In short, I totally agree that these things are very often unwanted. But I don't think I've made the jump to them being broadly ethically harmful on an individual level, yet. At scale is where I see these problems becoming much more obviously harmful. If you can track an entire population you might also have the tools to influence an that population. If you're influencing them in a harmful way then that probably constitutes as ethically harmful. Perhaps we need to figure out how to limit or separate these powers at a global level. Maybe tracking firms should not be allow to sell information to political organizations? Maybe they should all just be shut down? Anyway if you get time to further develop that point I'm all ears as ethics, especially in computer science is fascinating.

dcow 4y ago

I read this as an attempt to explore the tension between people who want total anonymity and people who just don't want their personal information abused. Sure, there are companies that abuse your information and yeah that's a violation of privacy. But the way I am reading the argument, the solution to one company violating your privacy isn't necessarily a surefire: we should take one more step towards total anonymity. Rather, it depends on the threat model. > Also, someone "being okay" with a violation of privacy is not a sufficient indicator. Objective harms exist and they don't go away because the user is ignorant or convinced, or coerced to make "acceptable trade-offs". Curious what you'd consider an objective harm. For example, I've been in discussions where any stable identifier is objectively harmful because it could be used by a service to track you. Therefore we can't do things like mutual authentication since your cert has a globally unique name, allow signatures in the browser because your key is a global identifier, behave normally at layer 2 (thanks Apple, I can't manage devices on my network anymore because they all randomize their MAC), or find my nearest Lowes because now my traffic exits 5 states away because ~~Apple wants to own my data~~ of my big scary ISP who can't be trusted. Sadly the only conclusion I can come to is that these problems are social and, though technology may have introduced new means to abuse people's privacy and it's natural to seek a technology solution, ultimately require legal solutions. For example, "deleting" IP addresses from the internet is not an acceptable solution in the fight against privacy abusers. Nor is telling services they can't collect information that could identify you unless they need it because, well, everyone needs it at some level and it turns out everything can identify you in the right context. Punishing people who abuse others' privacy is the solution. That, or, technology would need to let me select a la carte exactly which identifiers I want to allow and which I don't and we'd have to live with most everyone never touching those toggles and blissfully remaining tracked and targeted. Frankly I'm sick of privacy nut technologists (who no doubt are genuinely concerned for everyone's safety on principle) continually pushing everyone towards "tracking evasion" when all that most people really care about is surface level "tracking resistance". That much, at least, I resonated with in the essay.

nonrandomstring 4y ago
> Curious what you'd consider an objective harm. To dodge the risk of opening a can-o-worms I don't have time for on a Monday morning I'll just say watch for the chapter "Harms" in Ethics for Hackers when it's out, or ping me a PM if you fancy being a proofreader for that chapter. > Sadly the only conclusion I can come to is that these problems are social I just replied to another similar remark. They are simultaneously psychological, social, technical, legal and political, and must be addressed on all fronts to reduce long-term harms upon the many to the temporary advantage of the few. > Punishing people who abuse others' privacy is the solution. It's a solution amongst many which is necessary but insufficient to cover all harms. An example already given in comments; of Nederlands collection of religious data prior to the Nazi rise, is apropos anachronous and displaced threats. > Frankly I'm sick of privacy nut technologists (who no doubt are genuinely concerned for everyone's safety on principle) continually pushing everyone towards "tracking evasion" I hear that. There's a real risk of alienating users and losing the psychological campaign for better opsec by harping on a "mother knows best" tune. For example, GDPR cookie popups seem to have had that negative effect. > That much, at least, I resonated with in the essay. Me too. I very much like Rohan's writing. It sparks great discussion.
- dcow 4y ago
  > For example, GDPR cookie popups seem to have had that negative effect. Yes, but I don't think it's limited to "visibly" negative user experiences. Not being able to sign a document in the browser using an HSM-backed hardware key because vendors won't add that functionality because some nut on the mailing list is concerned about "supercookies" also is, arguably, harmful to users because it deprives them of stronger security applications (that might also lead to a stronger tracking resistance stance nonetheless by nature of the ability to sign and encrypt messages at the application level) in pursuit of a fuzzy goal of never letting anybody have a stable identifier, ever. That seems, in the very least assuming the supercookie concern is legitimate and not a vendor play at retaining platform control, something that can only be decided individually by the user and not the browser vendor. Another example is TLS 1.3. In order to have better privacy against unfriendly networks encrypted SNI pretty much kills content filtering at the last mile network level (can still do it on device, which is Apple's solution). This might be a great win for Joe who wants to browse Ashley Madison at the coffee shop away form prying eyes. But it's not great for parents who depend on content filtering solutions to help keep their kids safe on the internet. So, who really, is a technology like that harming? > To dodge the risk of opening a can-o-worms I don't have time for on a Monday morning I'll just say watch for the chapter "Harms" in Ethics for Hackers when it's out, or ping me a PM if you fancy being a proofreader for that chapter. I emailed you via book2021@. This, though, is really the crux of it for me (sorry I rambled about a bunch of peripheral things). It seems that ethically the idea of "remembering a customer to draw inferences about how they'd behave" is rather natural to how humans operate. Historically, I can't even tell you I don't want you to do that because how I conduct myself in public is, well, public. So on the individual scale I think a lot of literature needs to be developed around building an understanding of what exactly is wrong, regarding profiling in public, to do without a person's consent. Is it wrong to write down someone's name, for instance? Is it harmful to be so good at profiling someone based on immediately available information that you might learn some things about someone even they didn't really care to discover? It certainly seems invasive from a user perspective but then again so would a good shaman, psychic, magician, or psychologist who build careers on playing the statics and profiling games. In short, I totally agree that these things are very often unwanted. But I don't think I've made the jump to them being broadly ethically harmful on an individual level, yet. At scale is where I see these problems becoming much more obviously harmful. If you can track an entire population you might also have the tools to influence an that population. If you're influencing them in a harmful way then that probably constitutes as ethically harmful. Perhaps we need to figure out how to limit or separate these powers at a global level. Maybe tracking firms should not be allow to sell information to political organizations? Maybe they should all just be shut down? Anyway if you get time to further develop that point I'm all ears as ethics, especially in computer science is fascinating.