Initial Research Outputs
We’ve just wrapped up and initial research project, and we’re excited to be able to share some of our research outputs.
Recently we wrapped up an initial research project looking at the economics of end-user staking using the Juri Protocol. Specifically, we were looking to figure out which parameters in the system were most important to validate, and the values (or ranges of values) which might lead to stability. In this post, I’m going to share some a few of the things we learned.
If you’re not already familiar with the Juri Protocol, you can read about it in the previous blog post, Staking on Your Health. If you’re not sure why we’re taking the time to examine the system, you can read more about mechanism design in our previous post on Token Engineering.
We found a few parameters seemed to be key in controlling overall state (and hence stability) of the system.
The first was that both overall compliance rates, but also compliance distributions seem important. It’s more complex than just “people, on average, do their exercise in 60% of weeks.” Does that mean every one is compliant in 60% of weeks, or half are compliant in 30% of weeks, and half are compliant in 90% of weeks? The overall compliance is the same, but the distribution affects the system too.
On a slight side note, part of this research included a literature review which revealed some interesting data about populations – the World Health Organisation’s data on exercise levels globally are much lower than even the baseline data from randomised control trials in literature, suggesting that even observing people’s exercise will change their behaviour. I’m curious as to if this will be evident when the observation is by a device like a smart watch instead of a real person you have to look in the eye.
Back on topic, the amount of money in the system, and the way it is redistributed between compliant and non-compliant users is also important. Although there is a lot of literature on monetary rewards for compliance, and loss aversion for non-compliance, there is little on the dual-incentive mechanism we’re proposing, and our simulations to date have helped give us some starting ranges for values which lead to a stable system.
It seems like the reasonable values for the parameters we can control are something like the below.
- Gain for compliance – around 5%
- Penalty for non-compliance – between 10% and 30%
We’ll use our future research to help narrow down those ranges.
This research also seems to show that larger pools are incrementally more efficient. Pools with 200 or more users seem to be much more stable than smaller pools, so we might look into minimum and maximum pool numbers to provide a balance between stability and loading on individual smart contracts.
With any research, one of the key outputs is guidance as to what the next research should cover. I’ve broken this into two branches, which will feed each other.
The first branch is actual experimentation – we’re going to start running experiments with real life participants, using the Juri Protocol. This is super exciting because it means we can start gathering data about how real people (/populations) behave under the dual-incentive structure, and how their compliance varies over time. Part of this initial research also let us propose some experimental protocols and run a power analysis, so we know that our future experiments will have adequate statistical power.
The second branch is improvements to our simulations. The data from our ongoing experimentation will help feed our simulations, so the sampled population matches the real population much more closely. Like any simulation, we’ve also made some (justified) assumptions, and we plan to spend some effort seeing how the system behaves if we relax those assumptions in the future.
I hope that’s given you some insight into what we’re working on at Juri – it’s exciting to be able to share some of our research outputs.