Hacker News Reader

Comment by dleslie | Hacker News Reader

I'd rather block all EU IPs than bother to break working technology that is operated and maintained freely.

How do you revoke PII from a distributed information system, like anything using a blockchain or distributed version control? You don't.

View on HN

You can. You just need to think about it and have the correct controls in place.

If you choose to block all EU IPs instead of implementing the most basic data security and retention policies, then it’s for the best that EU users are not able to use your compromised service.

You cannot force peers of any decentralized distributed system to forget data. They can pretend to and appear as compliant peers and yet retain the data.

Then why are you storing PII on them?

Have you ever used git and supplied a username and email as your author? That's PII in a distributed service that cannot be revoked.

I've always felt that git is poorly designed for that reason. At the very least, there should be an "identity block" that commits etc point into, rather than embedding names, emails, and other identity information into immutable commits. Under GDPR, that's how it would've been designed in the first place. Of course, this'll never happen without a complete fork of git.

Once you've got that sorted and you can change/remove identity information, the likes of GitHub have no issue so long as they have GDPR-compliant contracts with any business partners who can access git repos. Obviously, anyone using GitHub who decides to store all identity data forever is, generally speaking, not GitHub's problem, same as someone who noted down the names of all their friends on Facebook isn't Facebook's problem.

> I've always felt that git is poorly designed for that reason. At the very least, there should be an "identity block" that commits etc point into, rather than embedding names, emails, and other identity information into immutable commits.

Misuse of git isn't a problem inherent to git.

Putting your name and email into commits is the default behaviour of git, and git will not work until you do so. This is design, not misuse.

No, putting the name and email you give it, if you give it one, is by design.

So the GDPR is entirely irrelevant because we could just give fake details to companies? And giving correct details to anyone, ever, is in fact “misuse”? That’s not how the law works, nor is it how it should work.

This was specifically about git. However, it does appear that you need to at least provide a user identifier for git. I can understand why that would be useful in a distributed system. It could be opaque identifier, but you're points taken in this case.

This was about git in the context of the GDPR, so a company (say, a git hosting provider) is by definition involved. If you're sending commits to other private developers in some way that doesn't involve a company storing the data, the right to be forgotten/right for records to be corrected/etc in the GDPR is irrelevant in the first place.

That is also PII being used for the purpose it was collected for (identifying a contributor) and I believe falls under Art.6(1)(f) of the GDPR. You would likely have a hard time convincing anybody that you can apply the right to be forgotten to a git repo - especially as that particular processing can be argued to not be requiring consent once you have submitted your commits.

The author details are not necessary for the core function of git; the change itself does not need the PII. Moreover, my concern is general for when PII is in such a distributed system; git is just one example of many.

Reply on news.ycombinator.com