WoT typically involves computing some trust level within a context. I think we need specific trust attestations and use DVMs (and manual observations, why not) on what to trust someone on

Remember that likes, zaps, etc can all be faked