Liz Bancroft-Turner, project manager at IOHK, gives an update on the progress made in the month of June on Project Shelley. Below you will find a summary of the latest video update.
Goal
- the primary goal of the Shelley release is to upgrade the network to operate in decentralized fashion
- it includes the following workstreams:
-Incentives which is about providing stakeholders with monetary incentives to follow the protocol and makes sure the system to run smoothly
-Delegation to allow stakeholders to delegate their right and obligations to sign blocks
-Networking which is about providing the network infrastructure to support decentralization
Project Update
Delegation:
- through their research paper, they have established how delegation will be performed in a way that satisfies requirements
- these requirements are also detailed out in the research paper, such as requirements for stake pools, recovery if keys are compromised, security convenience, reward considerations as well as making sure security of your funds do not get compromised if you delegate
- for the delegation design document, they are working on specifying and detailing how the scheme proposed in the research paper can be implemented in Cardano
- they must make sure the design of delegation is compatible with the envisioned incentives mechanisms
- design elements covered in the design document are things like handling of stale stake (what does the system do if it detects a stake pool or stakeholder stops producing blocks?) or how to share rewards with people delegating to a pool without flooding the system with transactions
- the design document will also specify how leader election works within delegation
- there was also an updated chapter on wallet recovery
Incentives:
- in the Incentives research paper, they proved that using formulas for utility, each Nash equilibrium has a desired form of k pools of equal size
- Nash Equilibrium is a central concept in game theory: where you have a game in which players can choose a strategy and depending on the strategy that each player chooses, rewards are paid to the players
- strategies are selected for each player so that such player or no player can improve his or her reward by unilaterally changing their strategy
- in other words, you have a Nash equilibrium if we have no rational reason to change what they are doing if nobody else changes what they’re doing
- in Cardano’s case, strategies are things like: creating a pool with certain costs and then setting the margins of a certain amount, or delegating a certain ratio of one’s stake to a pool and the rest to another pool
- the basic idea of applying game theory to practical problems is the belief that in reality, things will settle down in a Nash equilibrium
- in our example, it would be bad if there were Nash equilibria with bad properties such as less than k pools or pools of unequal sizes. So the fact that Liz and the Shelley team proved that there are no bad equilibria means that hopefully in practice, the system will indeed end up with k pools of equal size
- this was the main achievement on the research side
- for the design document: after having settled on the general shape of the incentives mechanism, the research team spent the last month fine-tuning, verifying and refining it
- small simplification to one of the formulas for pool desirability was made and the change was verified experimentally
- after some back-and-forth between theory and experiments, the results eventually aligned perfectly
- research team was able to formally prove the correctness of this improved game theoretical model and then the experiments confirmed the mathematical theorems
- this gave more confidence to the team that their mathematical analysis was correct
- from there, the team moved attention to the analysis side, and here, they looked at some of the cases like coalition between players and large stakeholders and examined how the system would behave in the presence of such complications
- while the engineering side examined the question of how to be sure that pool leaders would provide the necessary relaying infrastructure
- as of now, they have decided to rely on social pressure instead of technical measure
- this will be done by requiring pool leaders to publish their relay addresses as part of their pool registration, thus making that information publicly visible and verifiable
Networking:
- the networking stream involves Peer 2 Peer Discovery
- for this part, the team has a new addition who is looking at this
- he is still onboarding and getting familiarized with what was left from the former colleague
- another stream in networking is Delta Q Measurement
- here they have identified a number of real-world use cases that they will investigate to act as examples during the implementation and testing
- this gives them control to make the network usage more or less aggressive in its real-time use of network resources
- the team has identified the key use cases like catching up with the tip of the chain and block broadcasts
- the Delta Q Measurement approach is being used to help minimize latencies for those activities while not creating adverse network performance for other applications
- they are also investigating these use cases on the communication protocol
- this is essential part of the communication protocol design
- they have been looking at the way in which a node downloads blockchain segments from its peers and the way in which it deals with forks
- at the moment, this is done not so efficiently in the current release of the Cardano Settlement layer and it is essential to improve on that
- the Shelley team have converged on a solution and now they are analyzing it to make sure it is faithful to the Ouroboros paper
Additional Graphs
- At this point in the video, Liz presents two graphs to help understand the simulation and experimentation they conduct in their research
- The following two charts show the evolution of pools but with different parameters
- To read these charts, you will need to know that:
-the x-axis is time
-each coloured band is a pool
-the height of a band is the size of the pool - These simulations both used K = 10, meaning the ideal outcome would be 10 equally sized bands (note: in real life, K will be a much higher number like 100)
- the first shows that the ideal outcome was achieved as you can see that at the end of the simulation, there are 10 equally sized pools
- Liz and her team were able to achieve this result in all their experiments
- the second part plots the number of pools over time so it should be at 10 at the end of the simulation on the right hand side
- this second chart shows similar parameters as first graph, but while the first graph shows you a scenario in which pool costs are low with respect to rewards and the stake of the pool leaders has no influence on rewards
- in this version, costs are high and there is a significant influence of leader stakes on rewards
- as you can see that under both set of parameters, they were still able to get the desired results
The Plan
- at the point of this video, Liz and her team are almost complete with the research phase
- they have completed the delegation research paper
- and are just making final adjustments to the incentives research paper
- speculative research by its nature is not an exact science and while the team did not meet their calendar date milestones for the research phrase, they are confident about achieving their overall business objectives
- for the design phrase, they are due to complete the delegation and incentives design document in the next week
- incentives design would then be merged in with delegation design to make one final document
- once the design document is signed off, which they aim for next week, they will detail out the technical implementation plan
- this involves breaking the design down into the user stories and creating tasks for the developers to implement
Testnet Update
- Liz concludes by providing an update on the Shelley testnet and her project management methods for this
- currently, this part of the project is in the initiation phase of defining the objectives, deliverables and work plan for the testnet launch
- Liz follows an Objectives-Delivery-Work Plan (ODW) method
- the team will contribute to the testnet development by translating the objectives into deliverables then turning these into a work plan that has its own workstream phases and activities
- so far, they are working on defining the high-level testnet objectives and outcomes and scoping out the high-level work activities and timeline
- they have also defined schedule risks, secured planning resources, and conducted initial stakeholder analysis and engagement
- having this framework in place will ensure that there is alignment and integrity of the test plan against the users’ needs as well as avoiding the risk of producing a meaningful plan of unfocussed activities
- at IOHK, the teams spend a lot of time in this sort of initiation time, but it is to ensure that users’ needs are addressed to create the best outcomes and experiences