feat(protocol): need one real ZKP every N blocks #13673
Conversation
|
client matching PR: |
| // For this block, real ZKP is not necessary, so the oracle | ||
| // proof will be treated as a real proof (by setting the prover | ||
| // to a non-zero value) | ||
| evidence.prover = address(1); |
There was a problem hiding this comment.
@adaki2004 I should have let you review this change, as it may impact the tokenomics? if 90% of blocks's the proof times are very short, like 12-60 seconds, will the tokenomics still work?
There was a problem hiding this comment.
Well, the mechanics heavily depends on the proof time. In case we will be able to find a good parameter (kinda average) it could work yes!
Depends also what is the difference between 'real' proof time?
30 minutes vs. the dummy ones (12-60s) ? If so, it might be very tricky to make it work. If the diff is not that drastic, i assume we need to find the sweet spot (=very close to the average).
Soon pushing the simulation code, you can try, but obviously it does not have this scenario possible. (Can try to incorporate tho next week to see how it behaves within these new circumstences)
There was a problem hiding this comment.
Any reason not to submit these at the expected proof time? It's only for testnets so seems good enough even if we could manipulate the prover fee.
There was a problem hiding this comment.
We can certain try to submit these proofs with a roughly average proof delay, but I just want to understand the nature of the impact of such a really small proof delay.
There was a problem hiding this comment.
If most proofs keep being submitted below the target time the prover fee/reward will go to zero (except if the actual proof being submitted is a LOT longer than the target time to try and keep in balance somehow, but that would be very suboptimal)
There was a problem hiding this comment.
See simulation results here:
#13657 (comment)
There was a problem hiding this comment.
@davidtaikocha please make sure you don't submit these oracle proofs too soon!
There was a problem hiding this comment.
@dantaik It can come quick but would be good to have them as 'strictly' automated as possible so that we can calculate the rest properly (e.g.: every 10th block comes somewhere 1-2mins)
To reduce the prover cost of testnets, a new config
realProofSkipSizeis introduced, assuming it's 10, then we only need real ZKP every 10 blocks (block 10, 20, ....). For other blocks, the oracle provers will be considered as the real ZKP and will allow those blocks to be verified.@davidtaikocha this may affect the prover code.