Proposal: Incentivize Sub P-Reps to run a Node by changing how Sub P-Reps become Main P-Reps
A network is only as strong as it’s nodes. We believe that currently on the ICON network there is too little incentive for Sub P-Reps to run a node.
How does it work at the moment?
If a Main P-Rep fails, the Sub P-Rep with the most votes takes over and becomes a Main P-Rep for the present term. That means Sub P-Reps that are down in the ranking (#30+) have very little incentive to run a node because the probability that multiple nodes fail in a row in one term is extremely low.
How it might work in the future
Instead of selecting Sub P-Reps only based on the total amount of votes assigned to them, the protocol could either:
- Choose a random Sub P-Rep
- Use a weighted probability function to chose a Sub P-Rep
A Sub P-Rep is chosen purely based on randomness, regardless of how many votes it has.
Pro’s: This would give the same probability to all Sub P-Reps of being chosen, so all would be equally incentivized to run a node.
Con’s: The amount of votes a Sub P-Rep has would be totally ignored.
Weighted Probability Function
Each Sub P-Rep’s probability of being chosen is based on it’s votes in comparison to the total amount of votes of all Sub P-Reps (that means excluding votes from Main P-Reps!).
At the moment (using the current MainNet votes), this would looks something like this:
Pro’s: In comparison to total randomness, this would account for the votes a Sub P-Rep has received and therefore increases the probability of being chosen as a Main P-Rep.
Con’s: Sub P-Reps with very little votes still might not run a node because the probability of being chosen could be very small (<1% below rank #50).
If they get chosen as Main P-Rep and they don’t run a node and therefore miss 600 consecutive blocks, they get penalized and a portion of their (and their voter’s) stake gets burned. 🔥 — Regardless of the proposed method above.
Overall it can be said that at the moment the Weighted Probability Function method would cause more incentivation for Sub P-Reps from #23 to #39 compared to the Random approach.
If the calculations are done assuming 100 P-Reps (therefore 88 Sub P-Reps) are running nodes on the MainNet, the probability for Random falls from 2.27% down to 1.28% per Sub P-Rep of being chosen.
⚠️ This is meant to kickstart the conversation about approaches how we can solve this issue and by no means a be-all-end-all. We are also going to post this in the ICON Community Forum. Feel free to take part in the discussion!
block42 is a blockchain solution provider developing applications on top of most recognized blockchain frameworks like e.g. Ethereum, Hyperledger, ICON and many more.