GP comment already explicitly addressed this:

> Then, separately, you can manually add people you know personally to your "network". You trust them, so anything they trust is also something you might be able to trust.

It doesn't matter how many sockpuppet accounts a marketing company creates if you don't click the "trust" button on any of them, and if the system is designed so that your local trust network only consists of what you trust directly and (weighted) transitively.

I've been wanting[0] more or less the same for a long time now. Ideally the dataset should be public (and distributed) so that you could find bad actors and eliminate them from your network (i.e. when you see a bad result, you can figure out where it came from into your feed), as well as run your own tweaked algorithms. WoT alone is great, but it gets better still when you can tweak the algorithm.

At that point, it doesn't need to be people you know personally; it just needs to be identities with a proven track record. If you can manage multiple identities, you could make ones e.g. for the explicit purpose of presenting a blacklist of bad sites you've found, or for providing a "front page" to a certain subject. Anyone is free to trust your identity or not, and choose exactly how to rank its actions.

This whole thing could be expanded to tagging and other metadata as well as comments, and so on.

[0] https://news.ycombinator.com/item?id=29691303

I'm so glad to see others talking about this.

A system of distributed trust like this is the makings of decentralized moderation. Decentralized curation. I've also had this idea on my mind for a while: https://news.ycombinator.com/item?id=21975294