---
tags: Meetings
---

# Stream Agreements 5/18/2022 Questions

## Stake Slashing

### Question 1
- when, how much etc. (at least in the "default policy")
- right now: slash all stake if leaving a running bounty
- possibly soon too: slash if leaving too early; don't slash leaving a running bounty if been brokering for long enough
- also in v1: adminKickPolicy: slash for misbehaviour

### Answer 1

- In general, the model usually was to penalize all funds unless the broker waited for the specific amount of time before leaving, but this was also meant to be something that could be changed given different uses/opinions
- Z Thoughts: this types of policies we all left intentionally under defined in early versions because its exactly the kind of thing that you cannot totally predict the consequences of, you need to make judgement calls and/or run experiments were use minor variations of the rules (in experiments with real users) to see which ones yield the desired behavior. My generally recommendation would be simply provide some degree of parametrization in the contracts so they can be tuned in production. But more concretely, now that you need to implement something, assuming its not done parametrically, the clarity of the rules matters almost as much as the rules themselves. people need to know what the rules are before they engage so they aren't surprised

## Auto-compounding

### Question 1

- Uniswap v2 auto-compounded, uniswap v3 keeps liquidity and rewards separate. Can you think why that is? Regulatory/business reasons, or technical?

### Answer 1

- This source has some good information: https://docs.uniswap.org/protocol/guides/liquidity-mining/liquidity-mining-overview
- It may be something to do with the fact that an NFT is being used and possibly should not be modified but rather have rewards considered outside of it
- In general it seems that V3 is a bit more complex
- Z Thoughts: I suspect there is a relatively first principled difference that can be distilled from the dynamical system --but my intuition that auto-reinvesting leads to capital accumulation on some brokers, with little incentive/opportunity to distribute more widely
- Z Thoughts: If we assume that the amount of stake on a particular broker exhibits diminishing returns, you want the staker to be nudged toward going out and finding more good brokers to stake on, rather than just re-deploying all their returns on the same broker

### Question 2

- Did we have a specific reason to have separate stake+allocations vs just "pool shares"?

### Answer 2

- Within the mathematical specification it is good to break it out so that we can have metrics such as return on stake where we differentiate types of funds
- If it works out better for the smart contract to collaspe then it shouldn't be an issue

### Question 3

- Of course here slashing also important: what's the proper way to slash pool shares? Maximum slashing amount / "safe" part of shares? How much?

### Answer 3

- With keeping track of stake, it is feasible to just slash the staked amount so that anything earned may still be taken and the penalty is constant
- There are further refinements that could be placed i.e. slashing amount given a threshold time where the slash is (% of threshold time not fulfilled) * stake

### Question 4

- game theoretically, everyone should withdraw their allocations and add to the stake all the time
- but if everyone does it, result would be "auto-compounding"
- also: auto-compounding would result in the same allocations as if NO ONE does this move
- so: auto-compounding would result in the same allocations as current system AND not incentivize pointless transactions
- what were the downsides of "auto-compounding"?
- Auto-compounding really means there could be only ONE quantity ("pool shares"?) instead of two (stake and allocation), and buying/selling pool shares would be same as increasing/decreasing stake => simpler implementation

### Answer 4
- It depends on if stake drives returns or it is really just a barrier for entry to ensure brokers have skin in the game
- If it is simply a requirement that X stake be deposited, this will be used as a safeguard and no one will need to update stake, it will just stay to ensure the broker is doing the work
- If stake does factor into the returns and is not just a safety mechanism, then the easier solution ought to be utilized
- The primary question is whether amount of stake influences a broker's revenue or it just for them to have skin in the game

### Other thoughts:
- in one-bounty universe auto-compounding probably is the game theoretical correct behaviour
- but there are many bounties, so a rational broker might want to withdraw their allocation and stake it on another stream that pays better