All your suggestions cause us to need to audit not only T3 but also T1 and T2. This greatly adds to the complexity.
This could not happen to just any T1 LP because no other T1 LP is motivated by the T3 reward other than the T3 custodian. If a T3 custodian sells to an independent T1 custodian, that T1 custodian has no reason at all not to put up the funds and get the T1 pool rewards. If the T3 custodian sells to a colluding T1 custodian, that T1 custodian has every incentive to fool the shareholders into thinking the T3 custodian did their job, then walking away and splitting the money with them.
We could also think of ways to fully audit T1 custodians, but it would require some heavy use of API keys on the part of the pool operator to try to identify their individual providers. I’d really rather just do the simple T3 audit (everything on the blockchain) and provide the custodian with a whitelist.
What about following the POS model, when nu needs t1 liquidity, do a lottery on avilable t3 fund. the probability to win is proportional to everyone’s fund-times-days, reward the lucky custodian proportionally to f.t.d used.
Ok, interesting. What if the T3 custodian declines to participate until they are chosen as both T1 and T3? Do they sacrifice their f.t.d? This is a possible solution.
I don’t see why anybody would require to be formally elected as T1 custodian.
The incentive for being T3 custodian is to balance the T1 orders with the T3 funds and to increase the compensation rate (per invested money) for the money that’s invested at T1.
In my view there are at least two types of T3 custodians
an elected T1 custodian operating a NuBot on an exchange (MLP) who broadcasts liquidity
or a participant in an ALP (preferrably with fixed compensation, because that increases the incentive to create orders at sides that ran low)
I see it as there being three types of T3 custodians:
MLPs that run and report all three tiers (NuLagoon). The veracity of the T3 reports is upheld because they are bundled contractually with T1. T1 can be verified directly, so T3 is verified by proxy.
Trusted T3 custodians that are extended credit. These custodians are paid via their own spread at a constant rate/txn. They are kept honest by the profitability of their own contract, which they share in.
Self-funded T3 custodians. These are most desired as decentralized and cheap liquidity that scales well. However, the reward scheme must be complex as T3 cannot be directly verified. @mhps’s concept of fund days destroyed may be a good method for keeping T3 honest.
it will be all sorted out if T1 pays for T3 from T1’s profit. Then Nu doesn’t pay T3 directly. Nu only raises fixed reward fee when T1 liquidity is needed and let more fund come out of the woodworks.