6 ms·

> "There's only one way to be safe. And that's to collect only minimal amounts of data for minimal apps." No. The only way to be really safe is to go live in a

by amirmc 11y ago

> "There's only one way to be safe. And that's to collect only minimal amounts of data for minimal apps." No. The only way to be really safe is to go live in a cave, not to have any friends and never go anywhere near any tech ever again. I'm sure you can tell that I don't consider that to be much of an option. It's unhelpful to label 'data collection' as the source of the problem as 'collection' happens everywhere in our interactions with the world. There are valid arguments and discussions to be had about who really owns that data. It may well be about me, but that doesn't necessarily mean that it's mine. I'd say the majority of people are only just beginning to understand that a discussion needs to take place even though technical folks have known this for a long time. One solution to the 'ownership' problem is to make it possible for everyone to run their own infrastructure. i.e. have their own 'backend' that apps/services can be installed into, which their end-devices can then connect to. This is already possible for the technically savvy but a lot of work is needed before I feel we can trust such systems (I'm working on approaches to this [1,2], based on unikernels). In the meantime, I applaud business models that are not built around profiling (which is the crux of the issue cf. advertising). They're the only ones where the incentives have any hope of aligning. [1] http://nymote.org/blog/2013/introducing-nymote/ http://nymote.org/blog/2013/introducing-nymote/ [2] http://amirchaudhry.com/brewing-miso-to-serve-nymote/ http://amirchaudhry.com/brewing-miso-to-serve-nymote/

higherpurpose 11y ago
Or researching fast fully homomorphic encryption so they can one day be able to mine data while preserving strong user privacy.
hobs 11y ago
My thoughts are basically: + "Run your own infrastructure" is currently a pipe dream for what is 99.9% of people. + Companies selling data when they go bankrupt or M&As changing the nature of what they do with the collection happens regularly. + Discussing ownership is fairly simple as well, "Does the company already have your data?" + Any business not based around profiling has a additional options to make more money by selling their (your) data. Why isnt excess data collection a problem? Is it because its very very hard to solve and if we pass laws bad actors will continue to flaunt them? Is it because we are past the point of no return?
- castell 11y ago
  > Is it because we are past the point of no return? If you look at the computer history, we already had several cycles from single computer to timesharing to mainframes/supercomputers to PC to thin-clients/mobile to (?). There has always been a time when single powerful computers were in the majority and then there has been times when thin-clients with big server structures were the majority. The Internet supports also the model of more decentral computers and everyone could host their data on their own devices. It's currently just not the right time and there is no business model behind it, so no one push for that solution at the moment.
- wyldfire 11y ago
  > + "Run your own infrastructure" is currently a pipe dream for what is 99.9% of people. Or not even a dream, because those 99.9% prefer not to run their own infrastructure even if it's easy and cheap. However, "local apps" (ones which don't require infrastructure) could be appealing if they could achieve parity with the enormous-backend apps.
  - un_ 11y ago
    I just find it hard to run my own infrastructure. I tried setting up an e-mail server once, and the e-mails I sent from it were directly routed to the Spam folder of any Gmail account I tried to send them to. I tried DKIM, but the e-mails were still sent to Gmail's spam folders. Eventually I gave up on self-hosted e-mail.
Silhouette 11y ago
Thank you for posting a well-balanced summary of the situation. I think the biggest difficulty with balancing privacy concerns against other factors -- or even encouraging debate about and awareness of the issues among non-technical friends and family -- is that an item of data is itself neutral. It is how that data is used or combined with other data that may or may not be in any given party's interests. Sometimes the exact same technology or data could be used for very good purposes, useful and generally harmless purposes, or hostile purposes, depending on the context. For example, consider automated number plate tracking of motor vehicles. If your child has been kidnapped and a witness caught the plate of the kidnapper's vehicle, you're going to appreciate the police being able to find and intercept that vehicle as quickly as possible. If you're an urban planner responsible for keeping transport infrastructure as efficient as possible in the face of a rising population, an aggregated data set showing how real travellers want to move around your city could be very useful, helping you make decisions that improve the system for everyone. If you're that same urban planner but on the side you're working with a load of jewel thieves and abusing your access to historical movement records to figure out when specific wealthy residents are usually away from their homes so the thieves can break in and then abusing your access to real time tracking to confirm that the residents really are out or warn if they come home early, that's not such a happy ending for the data subject. Analogous issues arise with many kinds of personal data, including more sensitive areas like financial or health data. In the modern world, with vast databases and powerful data analysis tools and effectively instant communications and effectively unlimited storage, sometimes seemingly innocuous data can also give away a lot more about you than you might want or need. Things like what you bought at a store, or who you were tagged with in photos on social media, or a recording of you walking across a street on CCTV, can be used to determine many apparently unrelated things about you with relatively high (but, significantly, not complete) reliability, again sometimes quite sensitive ones that you might very much prefer to keep to yourself. Consider the store loyalty programme that determines a girl is pregnant from her purchasing patterns long before her partner or parents know. What about the person outed as gay because they were tagged with a whole group of openly gay people on holiday? Oh, by the way, unlike their friends in the photos, the first person lives in a country where homosexuality is still frowned upon and being open about it has real consequences. Also, gait analysis said you looked nervous as you walked into the airport, but it couldn't tell whether you just read a news story about a plane going down or you have inside knowledge of a terrorist threat, so unfortunately you won't be flying today. Just wait until automated text analysis software reaches the point that it can effectively de-anonymise posts like this one, and suddenly a lot of people who thought posting under a pseudonym was going to hide their criticism/whistle-blowing/advocacy of some controversial subject realise they weren't as safe as they thought and those comments are now permanently recorded on some public web site. We therefore need to move beyond black and white assessments like "data collection is dangerous" or "social media sharing is fun". What matters is not just what data is collected, but also who has access to that data, what they are allowed to use it for (including for how long it's stored, what else it can be combined with, and the like), and crucially, how these things can effectively be controlled or regulated to ensure that everyone is playing by the rules when data is data and once someone has it for one purpose it can readily be used for another.

by amirmc 11y ago

higherpurpose 11y ago

Or researching fast fully homomorphic encryption so they can one day be able to mine data while preserving strong user privacy.

hobs 11y ago

My thoughts are basically: + "Run your own infrastructure" is currently a pipe dream for what is 99.9% of people. + Companies selling data when they go bankrupt or M&As changing the nature of what they do with the collection happens regularly. + Discussing ownership is fairly simple as well, "Does the company already have your data?" + Any business not based around profiling has a additional options to make more money by selling their (your) data. Why isnt excess data collection a problem? Is it because its very very hard to solve and if we pass laws bad actors will continue to flaunt them? Is it because we are past the point of no return?

castell 11y ago
> Is it because we are past the point of no return? If you look at the computer history, we already had several cycles from single computer to timesharing to mainframes/supercomputers to PC to thin-clients/mobile to (?). There has always been a time when single powerful computers were in the majority and then there has been times when thin-clients with big server structures were the majority. The Internet supports also the model of more decentral computers and everyone could host their data on their own devices. It's currently just not the right time and there is no business model behind it, so no one push for that solution at the moment.
wyldfire 11y ago
> + "Run your own infrastructure" is currently a pipe dream for what is 99.9% of people. Or not even a dream, because those 99.9% prefer not to run their own infrastructure even if it's easy and cheap. However, "local apps" (ones which don't require infrastructure) could be appealing if they could achieve parity with the enormous-backend apps.
- un_ 11y ago
  I just find it hard to run my own infrastructure. I tried setting up an e-mail server once, and the e-mails I sent from it were directly routed to the Spam folder of any Gmail account I tried to send them to. I tried DKIM, but the e-mails were still sent to Gmail's spam folders. Eventually I gave up on self-hosted e-mail.

Silhouette 11y ago

Thank you for posting a well-balanced summary of the situation. I think the biggest difficulty with balancing privacy concerns against other factors -- or even encouraging debate about and awareness of the issues among non-technical friends and family -- is that an item of data is itself neutral. It is how that data is used or combined with other data that may or may not be in any given party's interests. Sometimes the exact same technology or data could be used for very good purposes, useful and generally harmless purposes, or hostile purposes, depending on the context. For example, consider automated number plate tracking of motor vehicles. If your child has been kidnapped and a witness caught the plate of the kidnapper's vehicle, you're going to appreciate the police being able to find and intercept that vehicle as quickly as possible. If you're an urban planner responsible for keeping transport infrastructure as efficient as possible in the face of a rising population, an aggregated data set showing how real travellers want to move around your city could be very useful, helping you make decisions that improve the system for everyone. If you're that same urban planner but on the side you're working with a load of jewel thieves and abusing your access to historical movement records to figure out when specific wealthy residents are usually away from their homes so the thieves can break in and then abusing your access to real time tracking to confirm that the residents really are out or warn if they come home early, that's not such a happy ending for the data subject. Analogous issues arise with many kinds of personal data, including more sensitive areas like financial or health data. In the modern world, with vast databases and powerful data analysis tools and effectively instant communications and effectively unlimited storage, sometimes seemingly innocuous data can also give away a lot more about you than you might want or need. Things like what you bought at a store, or who you were tagged with in photos on social media, or a recording of you walking across a street on CCTV, can be used to determine many apparently unrelated things about you with relatively high (but, significantly, not complete) reliability, again sometimes quite sensitive ones that you might very much prefer to keep to yourself. Consider the store loyalty programme that determines a girl is pregnant from her purchasing patterns long before her partner or parents know. What about the person outed as gay because they were tagged with a whole group of openly gay people on holiday? Oh, by the way, unlike their friends in the photos, the first person lives in a country where homosexuality is still frowned upon and being open about it has real consequences. Also, gait analysis said you looked nervous as you walked into the airport, but it couldn't tell whether you just read a news story about a plane going down or you have inside knowledge of a terrorist threat, so unfortunately you won't be flying today. Just wait until automated text analysis software reaches the point that it can effectively de-anonymise posts like this one, and suddenly a lot of people who thought posting under a pseudonym was going to hide their criticism/whistle-blowing/advocacy of some controversial subject realise they weren't as safe as they thought and those comments are now permanently recorded on some public web site. We therefore need to move beyond black and white assessments like "data collection is dangerous" or "social media sharing is fun". What matters is not just what data is collected, but also who has access to that data, what they are allowed to use it for (including for how long it's stored, what else it can be combined with, and the like), and crucially, how these things can effectively be controlled or regulated to ensure that everyone is playing by the rules when data is data and once someone has it for one purpose it can readily be used for another.