4 ms·

Besides the tremendous onus laws like this may place on small startups and side projects: Does anyone know how companies are supposed to comply if “user data”

by ptest1 7y ago

Besides the tremendous onus laws like this may place on small startups and side projects: Does anyone know how companies are supposed to comply if “user data” literally cannot be deleted? I’m thinking in the case of blockchain type applications, where one users’ actions feed into another users’ actions, and you can’t deleted user A’s actions without deleting potentially tons of other stuff and destroying the application. Like does this law basically ban GitHub and code collaboration too?

dsfyu404ed 7y ago
>Besides the onus laws like this may place on small startups and side projects: Startups and side projects are a non-issue. They can just comply from the get go. It's the too small to afford compliance but too big to easily change their business model that are going to be hurt by this. However, that brings up the question as to whether those business models should be able to exist profitably in the first place.
- ptest1 7y ago
  What I’m saying is that there is more work for startups / side projects to do, not that it’s impossible. It raises the barrier.
chobeat 7y ago
For the 1) you just don't commit data in the blockchain itself. You can commit an ID that points to external data and drop these external data if requested. This clearly looks like an anti-pattern but putting personal data in an immutable data structure is an anti-pattern too, since those data don't belong to you. What if a bank put the gold of their customers in a concrete pillar instead of a vault? Fuck the back, I guess. I don't see why not.
- ptest1 7y ago
  Again, the GitHub example is a good one. I have some commits in Django core from years ago. What if I “request” my data be deleted from GitHub? What are they supposed to do? If they rewrite the history it will destroy the project; and if anyone can rewrite the history in the future it leaves projects open to hostile actions. I suppose the argument would have to be made that it’s not personal data; it’s an act of public publishing or something. So in this case it’s akin to me publishing a blog and other people quoting it years later. I can delete the original blog but not the reprints in newspapers or quotes. Or that it stops becoming “your” data and becomes instead “the other user’s” once eg the Django project accepts the PR. So you can delete user A’s PR but not the Django project’s now-integrated copy. I think this rationale makes the most sense. After all, someone could still have the repository on their computer and push it back to GitHub again. I realize this is an edge case that doesn’t apply to 99.9999% of companies, but as an engineer I find it interesting! Edit: after thinking this through more, I suspect that e.g. GitHub could argue they comply as long as they delete User A’s repo. Subsequently integrated PRs, etc wouldn’t have to be deleted because they could argue they’re no longer User A’s. I kinda feel like if language to allow this was added to the law explicitly it would be open to abuse, so I suppose this kind of thing has to remain vague and open to interpretation in e.g. the courts if someone is being nasty.
mdavidn 7y ago
Git does have a mechanism to remove data from history. It's ugly, and you end up with new commit hashes, but it works. It was designed for situations in which a developer accidentally adds something sensitive, like a production database snapshot, to source control.
- ptest1 7y ago
  Yeah so you could do a rebase and remove all of the user’s commits, but their work would likely still remain in the subsequent history- integrated as changes in some way. It would be a real mess.
  - zbentley 7y ago
    Users' contributions (their work) and users' personal information (their metadata) are separate in this case. Cases where they are not separate (think digital signatures/signing keys) are a slightly more complicated question to answer.

by ptest1 7y ago

dsfyu404ed 7y ago

>Besides the onus laws like this may place on small startups and side projects: Startups and side projects are a non-issue. They can just comply from the get go. It's the too small to afford compliance but too big to easily change their business model that are going to be hurt by this. However, that brings up the question as to whether those business models should be able to exist profitably in the first place.

ptest1 7y ago
What I’m saying is that there is more work for startups / side projects to do, not that it’s impossible. It raises the barrier.

chobeat 7y ago

For the 1) you just don't commit data in the blockchain itself. You can commit an ID that points to external data and drop these external data if requested. This clearly looks like an anti-pattern but putting personal data in an immutable data structure is an anti-pattern too, since those data don't belong to you. What if a bank put the gold of their customers in a concrete pillar instead of a vault? Fuck the back, I guess. I don't see why not.

ptest1 7y ago
Again, the GitHub example is a good one. I have some commits in Django core from years ago. What if I “request” my data be deleted from GitHub? What are they supposed to do? If they rewrite the history it will destroy the project; and if anyone can rewrite the history in the future it leaves projects open to hostile actions. I suppose the argument would have to be made that it’s not personal data; it’s an act of public publishing or something. So in this case it’s akin to me publishing a blog and other people quoting it years later. I can delete the original blog but not the reprints in newspapers or quotes. Or that it stops becoming “your” data and becomes instead “the other user’s” once eg the Django project accepts the PR. So you can delete user A’s PR but not the Django project’s now-integrated copy. I think this rationale makes the most sense. After all, someone could still have the repository on their computer and push it back to GitHub again. I realize this is an edge case that doesn’t apply to 99.9999% of companies, but as an engineer I find it interesting! Edit: after thinking this through more, I suspect that e.g. GitHub could argue they comply as long as they delete User A’s repo. Subsequently integrated PRs, etc wouldn’t have to be deleted because they could argue they’re no longer User A’s. I kinda feel like if language to allow this was added to the law explicitly it would be open to abuse, so I suppose this kind of thing has to remain vague and open to interpretation in e.g. the courts if someone is being nasty.

mdavidn 7y ago

Git does have a mechanism to remove data from history. It's ugly, and you end up with new commit hashes, but it works. It was designed for situations in which a developer accidentally adds something sensitive, like a production database snapshot, to source control.

ptest1 7y ago
Yeah so you could do a rebase and remove all of the user’s commits, but their work would likely still remain in the subsequent history- integrated as changes in some way. It would be a real mess.
- zbentley 7y ago
  Users' contributions (their work) and users' personal information (their metadata) are separate in this case. Cases where they are not separate (think digital signatures/signing keys) are a slightly more complicated question to answer.