Of the systems listed in the paper, the only one I'd really heard of was BrightID, and it still seems like potentially the best. The FAQ on their website is a good introduction[0], but the paper seems rightly cautious:
"To control for Sybil attacks BrightID runs GroupSybilRank, a modification of the SybilRank algorithm, to estimate the anti-Sybil score of the network participants based on affinity between groups. Proposed to be used as the official BrightID anti-sybil algorithm, the effectiveness of this algorithm in the presence of multiple attack vectors, remains to be proved."
Unfortunately, just proving personhood is only the first step in deciding someone's reputation, but it seems like a good basis to build some proper decentralized trust systems from, for example [1].
The problem is that the more successful the system becomes (i.e. it's used for more valuable use cases) the more incentive there is to attack it. There needs to be a dynamic analogous to Bitcoin where the resources that go into securing the network grow along with the value of the network.
"To control for Sybil attacks BrightID runs GroupSybilRank, a modification of the SybilRank algorithm, to estimate the anti-Sybil score of the network participants based on affinity between groups. Proposed to be used as the official BrightID anti-sybil algorithm, the effectiveness of this algorithm in the presence of multiple attack vectors, remains to be proved."
Unfortunately, just proving personhood is only the first step in deciding someone's reputation, but it seems like a good basis to build some proper decentralized trust systems from, for example [1].
[0] https://www.brightid.org/faq
[1] https://adecentralizedworld.com/2020/06/a-trust-and-moderati...