Plasma Compound

Alice: Borrwer

Bob: Lender

Tom: Trusted Third Party

TxName: lockEscrowWithInputThenSwap

TxInput1: OwnState(DAI, Alice)

TxInput2: OwnState(ETH, Bob)

Arg1: Sig(Alice)

Arg2: Sig(Bob)

Arg3: Timeout

Output1: CompoundLockedState(DAI)

Output2: OwnState(ETH, Alice)

TxName: unlockEscrowWithInputThenSwap

TxInput1: CompoundLockedState(DAI)

TxInput2: OwnState(ETH)

Arg1: Sig(Tom) <= What we need is the guarantee that Tom DO sign this Tx: Repeated Game Structure is needed for that

Arg2: Timeout

Output1: OwnState(DAI, Alice)

Output2: OwnState(ETH, Bob)

TxName: timeoutEscrow

TxInput1: CompoundLockedState(DAI)

Arg1: Sigh(Bob)

Arg2: Timeout

Output1: OwnState(DAI, Bob)

Note:

Multi token Plasma means finite pair of lending.

Defragmentation would be severe.

What if Tom(TTP) and Bob(Lender) are colluded for rejecting unlockEscrowWithInputThenSwap?

Alice loses DAI and ETH remains

Collateral rate can be eq with trust of Tom.

What if we regard this system as repeated game?

As per Folk theorem, Tom is trustworthy

This repeated game logic is usable for Plasma Operator and other trusted third party

How much fee Tom need?

Tom is to be required some license? -> maybe no(yay!)

Operator = TTP = Bob

Ignoring unlock sig attack

#todo I worry that L1 contract must be Plasma-conpatible - in order to map lockedState's exit.

And so this TTP concept must be used as well in L1 (Ben Jones from PG mentioned about L1 mapping contract https://twitter.com/ben_chain/status/1106976647322963969 )

#todo Maximal damage depends on multi lender collusion rather than single lender collusion. So, if the TTP could collude with 100 lenders, we must count x100 kick back expected profit.

#plapps