Web of Trust in IPFS

For many different applications (such as a distributed mapping of foreign to IPFS hashes), it would be desirable to have a Web of Trust. The basic principle is outlined in FMS’ trust FAQ. It could be built on top of IPFS as an external plugin, but the question is whether it should, or if that should be handled inside IPFS due to the wide range of uses. Your thoughts on this? What are IPFS’ specific design goals? To do as much as possible inside or outside IPFS?

An example of another use case would be leveraging IPFS’ deduplication to cross-pollinate the same instance of a file in torrent 1, torrent 2, and IPFS, improving swarm health for all parties. It could also be used in other cases where you can validate something reasonably fast, but at the same time want to protect yourself from spam and DoS attacks.

Here is some more information of the Freenet implementation of Web of Trust:
Using pubsub, you should be able to do away with the hardcoded seed nodes, but you’ll still need some dedicated nodes to provide introduction puzzles. It wouldn’t make any sense to put their generation in IPFS though.