4 ms·

Isn't the burden on the data controller to ensure that they only transmit data to processors that are setup to handle deletion requests? In this case, that'd me

Isn't the burden on the data controller to ensure that they only transmit data to processors that are setup to handle deletion requests? In this case, that'd mean Pintrest or whoever is required to get their advertising partners or API consumers to setup some kind of webhook or feed for deletion requests.

hodgesrm 8y ago
It's just not very clear how easy that protocol would be to implement. The problem is siloed data without clear identifiers to track sources. For example, it's common in services like Pinterest to have upstream collectors that feed information into a Kafka pipeline to downstream services that use the data as they see fit. They don't necessarily have consistent identifiers that would allow you to determine that person A's data lives in derived form in services 1, 5, and 14. In the limit you might have to scan all persistent data. It gets much harder once the data leave the company and travel to other online services. Depending on the case law and administrative guidance that develops to interpret GDPR more precisely this problem may be simply unsolvable for some businesses.
- neuland 8y ago
  I totally agree, re: possibly being unsolvable for some companies. It'll be very interesting to see how the actual enforcement develops. I suspect the regulators want companies to create these kinds of consistent identifiers where they don't already exist and make sure that they ride along with personal data wherever it finds it's way. There's a comment in the article about an oblique reference in an email to "a tall, bald guy living on East Whatever St." being PII and must be included in data subject review/deletion requests. How do you find that, even if you're scanning everywhere? What about users' computers with excel sheets with personal data? That's how reporting is done a lot of places that I know of. You can make them install an agent to scan data or require all data be put in centralized storage like a network share or corporate subscription to a service like dropbox or something. But, I 100% guarantee that Andy from HR or Jane from marketing going to "forget" to do that or put it in some insane format that the scanner can't handle.

hodgesrm 8y ago

It's just not very clear how easy that protocol would be to implement. The problem is siloed data without clear identifiers to track sources. For example, it's common in services like Pinterest to have upstream collectors that feed information into a Kafka pipeline to downstream services that use the data as they see fit. They don't necessarily have consistent identifiers that would allow you to determine that person A's data lives in derived form in services 1, 5, and 14. In the limit you might have to scan all persistent data. It gets much harder once the data leave the company and travel to other online services. Depending on the case law and administrative guidance that develops to interpret GDPR more precisely this problem may be simply unsolvable for some businesses.

neuland 8y ago
I totally agree, re: possibly being unsolvable for some companies. It'll be very interesting to see how the actual enforcement develops. I suspect the regulators want companies to create these kinds of consistent identifiers where they don't already exist and make sure that they ride along with personal data wherever it finds it's way. There's a comment in the article about an oblique reference in an email to "a tall, bald guy living on East Whatever St." being PII and must be included in data subject review/deletion requests. How do you find that, even if you're scanning everywhere? What about users' computers with excel sheets with personal data? That's how reporting is done a lot of places that I know of. You can make them install an agent to scan data or require all data be put in centralized storage like a network share or corporate subscription to a service like dropbox or something. But, I 100% guarantee that Andy from HR or Jane from marketing going to "forget" to do that or put it in some insane format that the scanner can't handle.