---
tags: Meetings
---
# 6/16/2022 Update
## Agenda
1. Mathematical Specification Updates
2. cadCAD Updates
3. Catch Up on Example Scenarios
4. 6/14/2022 Discussion Topics
5. Priority Options
## Mathematical Specification Updates
- Specification can be found here: https://hackmd.io/6fN8ITTeTVyMwV3ToeApSw?both
- State variables were iterated on and solidifed for the entities
- All mechanisms were written out
- Mathematical Specification is now in V1 form
- Further improvements and fixes are logged here in the work log: https://hackmd.io/v7nLGvGjSIG6HoR8PiEpMQ
## cadCAD Update
- The 5 example scenarios are implemented within the cadCAD model as tests
- Take away 1: There is still a few things to add in such as a policy that sends the yield to the pool value instead of directly to internal balances of delegators. These are being iterated on
- Take away 2: Some simplifications might be better to expand upon such as the assumption that streams are exogenous from this cadCAD model and we track stake at the aggregate level
- Further feature expansions have been added such as the maximum deposit per epoch option (policy to limit how much a delegator can deposit) and maximum withdraw per epoch (to limit withdrawals)
## Example Scenarios
- Catching up on thoughts regarding last meeting's example scenarios state
- The scenarios have been expanded to cover more uses
- The cadCAD model will be used for verifying and testing that the model holds
- Current plan would be to have a longer list of short-form examples and a smaller selection that we do a long-form breakdown for
- Short form scenarios: https://hackmd.io/QFmCXi8oT_SMeQ111qe6LQ
- Long form scenarios: https://hackmd.io/_Tv-VJdXTo6UNV8NRFM0NA
## 6/14/2022 Discussion Topics
### Juuso Topic
Broker's own stake in a BrokerPool: right now, all capital that a broker deploys comes from delegators; in other words, delegators bear all the risk for broker misbehaviour. Of course the broker itself can also be a delegator (I guess?) and put in a starting capital. Question: should broker's equity be "special" in the sense that it would get slashed first, or at least that prospective delegators would be able to see how much broker has its own bacon on the line? Should there be (from the start in the MVP) different "seniority levels" of equity in case of slashing?
### Thoughts
- This is a good idea, while it may add complexity, it probably is worth the cost of complexity to add safety to the pools
- Would likely function similar to first loss accounts in a fund
- Generally, it probably makes sense to have a separate variable to track broker stake
- When slashing occurs, there would be a waterfall of payments such that first broker funds are taken out, then pool value is decreased
- Also likely there would be an additional action when yield is collected whereby if the broker stake < minimum broker stake, all their earnings get diverted back into the pool as funds and it increases this broker stake account value
- Would yield need to be weighted differently based upon the amount a broker puts in then?
## Priority Options
Recommendation:
IF 6/14 topic implementation is prefered: first start by scoping out and tackling smaller improvements time permitting. Next meeting discuss and if the decision is to expand, re-write the spec to include this and add it to the cadCAD model.
ELSE: Option 1 + Option 2
### Improvement Options
1. Working through the work log and finding divergences between cadCAD model and math spec
2. Creating more example scenarios
3. Improving the current example scenarios
4. Create failure mode scenarios, identifying edges/limitations such as dilution of early delegators for example
5. Finish V1 of policy options sheet
### Expansion Options
6. Beginning cadCAD extensions, writing out a framework for parameter selection under uncertainty
7. Begin implementation of 6/14 topic
### Innovation Options
8. Write scripts for auto-documentation in the future. Basically would be definitions of all the policies, mechanisms, behavioral actions, etc. and how they are graphically related. This would then make it so that one could iterate over the graph relationships and re-create the diagrams and specifications (what type inputs are etc) automatically allowing for easy switching of names, expansions of documentation, etc.