Just an idea, I think coin age would be very useful in subverting reputation systems -- can't send your own money back and forth to subvert the system and build up reputation.
This will end up with people selling old coins for reputation though.
Any other ideas for preventing gaming of trust systems/reputation systems for e.g. a market?
0 ·
Comments
It's nice but also allows big boys to push up their reputations by colluding and destroying massive amounts of coindays.
What might be fairer is something like total_rep = total_trades * log(total_coindays_destroyed)
The log() is there to not give too much weight to massive txs.
Also, i reckon reputation is a kind of 'inferred trust toward an end'. The person that you trust design a bridge, you wont necessarily also trust to fly a plane. Trust you have is 'calculated' by yourself. I.e. you could ask people whether they trust a third party toward some end, and calculate from responses how much you would trust by implication, as dependent of the trust 'towards the end of determining trust' in the people you ask.
I am not sure how much reputation needs consensus,(and could be on the blockchain for it) any consensus trust source is essentially like an authority. Which would be fine, if there is reason to trust the authority. But that implies that if you use a trust system, it might be better to use consensus reputation via that trust system, or that it would make sense for such a reputation system to join a trust system.
So I was thinking "1000s of faux transactions with the same coins should be worth considerably less than 1000s of transaction with legitimate coins, because the legit coins will have more coin days"
> Doesnt that just promote making many smaller fake entities trusted?
I don't understand, can you elaborate?
The more I think about this, the more I realise that perhaps it's just the case that we'll never be able to stop people from circumventing the system by buying and selling trust.
[1] we might actually have this luxury on freebay after all. New guy enters the market, does 1000 deals in the first day, seems less trustworthy to me than a guy who does 1000 deals over a year.. Seems unfair though, first guy might have just marketed well. I wonder how eBay deal with this.
The idea is that generally if U is some usefulness derived by some freely divisible value V, then U(Va+Vb)≥U(Va)+U(Vb) to avoid people being better off splitting up value. Egalitarianism wants it smaller, but it just doesnt work, best we can do is equal.
Same shit here, but i think i should look at the case where you disadvantage 'small users' in one way but give them an advantage the other way.
In 'identifying the human', we can essentially remove divisibility.