Calculating EVM account integrity or "realness"

1. Ideas & Brainstorming
1

As I start demoing my software to people I often come across the question: how do you prevent a malicious user from manipulating the triple position data to make their address look “trustworthy”? I am talking aside from Intuition trust claims (I’ll leave that for a whole different discussion)

Off the top of my head these are some relevant heuristics:

  • Age of account’s first transaction
  • Number of account transactions
  • Balance of account
  • Activity / balance on other chains (with same address)?
  • NFT ownership(?)

I’m curious what other pieces of data could be used in heuristics to signal that an address is less or more trustworthy. Thoughts?

2
  • Linked accounts (Twitter, Discord, etc.) - the Sophia guys have the code for account linking if needed
  • Reputation across those linked accounts
  • Criteria used by sybil resistance products like Gitcoin Passport and Authena
  • ‘Trust Circles’ - which can be derived from the above, and from Intuition attestations/reputation. I know you mentioned wanting to leave this for a separate convo, but IMO, this is ultimately how you prevent ‘manipulation’ - everything else is gamifiable and reductionist. And the ultimate form of this model is - you don’t have a single 'truth’ or ‘score’ for things → you have relative, subjective trust, where the trust is based on the perspective of the observer. For example, if I am not connected to anyone you’re connected to, and you’ve made an attestation, then it should be weighted far lower than if someone I have explicitly said that I trust has attested to something. There’s also a path here towards universal scores - wherein you look at an entity’s contextual reputation within the context being observed (say, Food) - and see if they have a high reputation in this context (based on the claims they’ve made / claims about them - super recursive, bc then you need to look at the reputation of the entities making the claims about them) - and then weight things according to this contextual reputation.