What will be in place to stop a group of people destroying someones reputation on the platform?
Great question! We actually describe our solution to various forms of reputation system attacks like Sybil, Traitor attacks. Another class of attacks include targeting attacks that happen when an attacker tries to downgrade (also called Slandering attack [1]) or upgrade another user's reputation or even their own account (i.e. self-promoting). This type of attack occurs by submitting a false review or even purposely rating another user once or even more than once (also called Ballot-Stuffing [2]) without truly interacting with the other party. In some cases, it can occur by hiring external entities to execute the attack (similar to a Sybil attack).
This solution to this kind of attack is as follows:
Only users who have completed a transaction can rate each other.
Only allow a single rating per user per transaction.
Throttle the reputation contribution by counter-party (i.e. track repeated transactions made by the same counterparties and limit the contribution that any single counterparty can have on the overall reputation of a seller).
The method used to calculate reputation should not rely on just the number of transactions but include other factors like transaction value, the reputation of the parties involved and transaction recency to name a few.
You'll find much more detail in
our white paper under Section 4.3 Abuse Prevention.
References:
---
[1] Hoffman K., David Z. and Nita-Rotaru C. (2009) A Survey of Attack and Defense Techniques for Reputation Systems, ACM Computing Surverys 42.
[2] Spitz, S. and Tuchelmann, Y. (2011) A Survey of Security Issues in Trust and
Reputation Systems for E-Commerce. Autonomic and Trusted Computing, 8th International Conference, pp. 203–214.
