Modefi
  • Introduction
  • Oracle Solutions Suite
    • Decentralized Aggregated Oracle
    • On-Demand Oracle
      • On-Demand Oracle - Technical Manual v0.1
        • The On-Demand Oracle System
        • Types of Users
          • Data Request Creators
            • Requesting Data
            • Setting Times
            • Cancelling Data Requests
            • Disputing Results
          • Validators
            • Account Management
            • Staking (and Unstaking)
            • Providing/Endorsing Data
            • Disputing Results
            • Receiving Payment
          • ODO Custodian
        • Algorithms
          • Computing Request Costs
          • Depositing and Withdrawing Coins
          • Staking to Endorse Data
          • User and Staking Slot Tiers
          • Timing/Lateness
          • Bumping
          • Withdrawing
          • Endorsing
          • Payment
          • Slashing
          • Reputation
          • Staking Bonuses
          • Disputes and Resolutions
          • Coin Credits
          • Account Transfer
      • On-Demand Oracle - High-Level Overview
    • Oracle Marketplace
  • Defi Dashboard
    • What is the Modefi DeFi Dashboard?
  • Token
    • Tokenomics
      • Token Distribution
      • Token Stats
      • Token Emission Schedule
    • Token Sale
    • Token Utility
  • General Information
    • History of Oracle Based Hacks / Exploits
      • Synthetix $1 Billion Exploit
      • Trader Exploits bZx Oracle for $330,000 Profit
      • $100 M Liquidated on Compound Following Oracle Exploit
  • Blockchain Basics
    • What is a Smart Contract?
    • What is an Oracle?
  • FAQ
    • Staking on Fantom
    • Staking on Binance Smart Chain
  • How-to's
  • Smart Contract Addresses
  • Links and Socials
  • Media Kit
  • Disclaimer
  • Terms and Conditions
  • Privacy Policy
Powered by GitBook
On this page
  1. Oracle Solutions Suite
  2. On-Demand Oracle
  3. On-Demand Oracle - Technical Manual v0.1
  4. Algorithms

Reputation

Reputation is a way to reward the long-standing users who consistently provide correct endorsements. With no slashes, a user’s reputation is equal to the amount of data they have provided summed across all datasets, so providing and endorsing larger datasets results in increasing a user’s reputation faster. A user’s level is directly related to their reputation. By increasing their level, a user unlocks the ability to stake for more datasets simultaneously. The relation is direct in the sense that a level 5 user can stake for 5 datasets and a level 8 can stake for 8 datasets at one time.

Reputation is lost whenever someone is slashed. There are two levels of reputation slashing.

The first is for cases where the user provides the wrong data. This type of slash is the most severe type, and it can quickly drain the reputation of an account with successive slashes. The slashes occur based on an exponentially-increasing factor. The first slash reduces the user’s reputation by half, the second slash by a factor of four, then next by a factor of eight, and so on, doubling in severity on subsequent slashes. The slash factor does gradually decrease, however, so if a user is slashed and loses half their reputation, and then provides enough consecutive correct data to lower the factor, and then makes a mistake again, they again lose half of their reputation rather than three quarters. In this way, rare mistakes will not render an account unusable.

The second case is used for late users and cases in which the user provided no data. Providing no data could potentially be due to events outside of the control of the user, so in this case, the reputation loss should be less severe. A similar argument applies to late data. For both these cases, the reputation loss is equal to the percentage of stake that is lost and then scaled based on the user’s level. For users from level 1-4 there is no mitigation to the loss of reputation. From level 5-9 the amount lost is reduced by a factor of two. For users from level 10-14 the reputation lost is reduced by a factor of three, and for max level users at level 15, the reputation lost is reduced by a factor of four.

Even though the loss of reputation has been decreased for late and unfilled endorsements, the faults are still recorded, so if a user who has provided a lot of late data provides the wrong data, the previous transgressions will be counted and the slashing could be very large.

PreviousSlashingNextStaking Bonuses

Last updated 1 year ago