New permission requirements make this an automatic hard pass - recommendation should be revoked.

Edit in response to the developer:

I think you're missing the point.

The issue isn't that Mozilla suddenly gave RYD access to the whole internet. It's that Mozilla has now made explicit data collection that was apparently already happening, but wasn't previously disclosed.

And "anonymous user ID" doesn't really solve that. It's a persistent identifier tied to YouTube activity. That's pseudonymous, not meaningfully anonymous.

More importantly, the extension being open source only tells us what gets sent. The RYD backend is still a black box. Once that data leaves the browser, we're being asked to trust RYD not to retain, correlate, profile, monetize, or sell it - particularly after we've learned the collection was previously undeclared.

The fact that the data is useful to RYD doesn't mean it's necessary for the extension's basic advertised function. Showing a dislike count doesn't inherently require building a persistent record of users' YouTube activity.

And the Chrome and Firefox stores don't even have identical data-collection declarations, so "Mozilla just changed the labels" doesn't really explain everything away.

That's why I find the response a bit tone deaf. The concern isn't "OMG, RYD can now read every website." It's "why does a simple dislike-count extension need to collect this much information about my YouTube activity, and why was that collection not clearly disclosed before?"


Edit 2:

The vote/registration pipeline isn't anonymous - it's pseudonymous and persistent, and your own code and privacy policy confirm it:

1) Persistent identifier: generateUserId() creates a unique 36-character ID at registration, which createBrowserCredentialStore() persists in browser.storage.sync. It isn't rotated per session or video.

2) Persistent correlation: performVote() sends { userId, videoId, value } to /interact/vote and /interact/confirmVote. Reusing the same identifier lets your backend associate votes and video activity with the same user over time. That's pseudonymization, not anonymization.

3) Additional identifiers: Your privacy policy says IP address and request time are also collected. Those can further associate the persistent identifier with a real-world network connection.

4) Your own policy treats the identifier as personal data. So describing it to users simply as an "anonymous" ID is misleading. Randomly generated does not mean anonymous when the value is persistent and specifically used to correlate records.

The appropriate description is pseudonymous, persistent user identification, not anonymous collection.