Parasitic Immune Augur

Assumptions:

• REP market cap before and after a fork are the same for the true universe
• REP market cap goes to 0 after the fork on the false universe

Simple Single Forking Market Scenario:

• 1 binary market
• $100 in open interest locked in contracts
• $1,000,000,000 in parasitic open interest
• $500 REP market cap (5x OI)
• 200 REP token supply
• On Market Fork:
  • $100 locked in contracts divided up among 200 REP holders so each gets $0.50
  • 50 REP-true minted for true share token holders (REP supply in universe will be 250 @ $500 market cap, so minted tokens are worth $100)
  • 50 REP-false minted for lie share token holders (REP supply in universe will be 250 @ $0 market cap, so minted tokens are worth $0)
  • REP holders vote with their REP with 150 going to lie and 50 going to truth (because parasitic open interest controls REP in this example)
  • At end of fork REP holders who migrated to REP-true get minted an extra 150 REP-true, REP holders who migrated to REP-false get minted an extra 50 REP-false.
  • $1,000,000,000 parasitic open interest resolves as lie if they follow the reported outcome. They would need to follow market cap to detect which one the world felt had real value.

Formula for minting tokens to outcome share holders: tokens_minted = open_interest * rep_supply / (rep_market_cap - 2 * open_interest)

Interesting Effects

• In a fork, the oracle never actually resolves the market. It just gives people REP in each forked universe and the broader market will ultimately decide which REP is actually worth something. This means any parasitic open interest (any open interest that isn't locked in Augur contracts and paying fees) is left out in the cold in a fork. Essentially, parasitic open interest isn't actually secured by the oracle, because there is no final oracle decision!
• At the time of the fork, the system transforms into a colored coin system, but unlike other colored coin systems it "recovers" out of it at the end and returns to a system that is returning externally valuable assets to users rather than just colored coins.
• Because we no longer have to deal with parasitic open interest, we can probably reduce the 5x multiplier.

Downsides

• When a fork occurs, all open markets at the time of the fork convert from their normal denomination (e.g., DAI) to REP denominated. Any new markets that get created will return back to their normal denomination but the REP denominated markets will remain so until they resolve. This not only introduces some amount of complexity in the code (though solvable I think by tracking assets internally in a wrapper contract), but it also means people that intended to make a bet denominated in the market's denominated token end up holding a long REP position along with their bet.

Rationale

• The tokens minted formula assumes that the sum of the market caps across all post-fork universes will equal the market cap of the pre-fork universe minus the locked assets that are about to be distributed to REP holders. The tokens minted will be of equal value to the the locked assets that were given to REP holders, so the outcome share holders will not take any loss due to the fork.

Things to Figure Out

• Forks are disruptive, and having one may result in a decrease in confidence in future REP value. Should we give a little bonus to outcome token holders so there is some room for REP price fluctuation?
  • What is the upper limit bonus we can give that won't break incentives? At least 1.25% of REP needs to be destined for REP-lie for a fork to happen, and if the REP targeting mechanism is working properly then this means 1.25% of 5 * open_interest, we need to make sure we don't mint more value than 6.25% of open interest. However, that assumes there are no other incentives for attacking the system.
• How much can we reduce the REP market cap targeting requirement by?
  • Since the final oracle doesn't actually make a decision, there is no longer a financial incentive to force a fork. REP market cap only needs to be high enough such that minting 99.9...% of supply would end up being worth the same amount as all of the open interest, or 1x multiplier.