So, I’ve been vaguely workshopping an idea for a while and I was curious to hear what y’all thought: decentralized spam prevention, via post/comment level incentives (tentative name: Spam Controlling Augmented Reputational Stake, or SCARS. No comment on whether that was a backronym or an acronym). Let me explain what that means – in short, require some small stake for someone to post or comment on a decentralized network, which they lose only if the message is found to be spam or advertising. In order to ensure a lack of people removing stakes for frivolous things, the spam evaluation is done by people on totally separate forums after the message is anonymized – anyone can perform this sort of evaluation as long as they themselves stake some slightly larger amount (to ensure you don’t get censorship problems – this is a tweakable knob and might be set network by network). Spam evaluations themselves are cast up to higher validators who can cause people to lose their bonds by providing contextual proof or a decentralized vote that a comment marked spam was not, or vice versa. The aim of the system is to force spammers/advertisers to pay some level of fair market rate for the negative externality they inflict on others.
Interesting. My big concern is how do you anonymize the data properly? You can anonymize the structured part (e.g., the user initiating the post), but what about the unstructured part (the message itself)? Either the prose, or certain words in the message could leak identifiable information, which validators would be privy to.
Basically, if this can be effectively machine-automized, reducing the need for human votes (and therefore - subjectivity), then I think this could be valuable.
The way you describe it sounds a lot like a Token Curated Registry design. The only difference would be only individuals who are part of a registry (after staking some amount) would be able to post comments / advertise. Anyone would be able to challenge what they post and it would be up for a voting. I’d recommend you check out Adex if you haven’t seen it yet.
One problem with TCRs as is is around voting and how to prevent collusion. We recently did a meetup around Enigma & Private Voting and how this can be applied to the TCR setting.