mirror of
https://git.sr.ht/~seirdy/seirdy.one
synced 2024-12-17 22:32:10 +00:00
1.1 KiB
1.1 KiB
| title | date | replyURI | replyTitle | replyType | replyAuthor | replyAuthorURI |
|---|---|---|---|---|---|---|
| Limited tracking and consent | 2022-09-24T11:28:34-07:00 | https://indieweb.social/@Chronotope/109054613809239268 | if the initial collection is very limited it doesn't run afoul of the issues that should require consent | SocialMediaPosting | Aram Zucker-Scharff | https://aramzs.github.io/aramzs/ |
Assuming data is a liability, how limited should data collection be to not require consent?
I think temporary storage (a week or less) of access logs combined with low-entropy binary information (dark mode, is viewport narrower than what I test with, etc) is reasonable for a small operation. This holds if the data collection is clearly documented in a privacy policy, is Tor-friendly, and obeys signals like GPC. These access logs should exclude high-entropy headers like client hints.
Larger operations should store even less since they have the means to correlate information from many sources. ipscrub comes to mind.
The only long-term storage that should happen without consent is of bot traffic.