Update elections' minimum untrusted score
On 2022-12-12 this referendum proposal was executed to, among other things, lower the min_untrusted_score to deal with a mass slashing incident in Kusama.
This proposal aims at updating the current min_unstrusted_score in Kusama back to sensible values. The new minimum untrusted score was calculated by averaging the last 30 election scores and halving them (starting from election at block #17416892).
When enacted, this proposal will change the current untrusted minimum score in Kusama from
{
minimalStake: '1,722,238,583,242,009',
sumStake: '2,171,296,385,410,646,127',
sumStakeSquared: '62,184,223,048,402,160,114,561,549,513,680,866'
}
to
{
minimalStake: '2,957,640,724,907,066',
sumStake: '3,471,819,933,857,856,584',
sumStakeSquared: '78,133,097,080,615,021,100,202,963,085,417,458'
}
Why is this change important?
The MinimumUntrustedScore configuration defines the minimum election score that will be accepted by the elections pallet from off-chain staking miners. Increasing the minimum untrusted score ensures that weak election solutions will not be accepted. The proposed election score is more sensible and secure given the current state of the network.
Comments (2)
Proposal Passed
Summary
0%
Aye
0%
Nay
Aye (240)0.0 DOT
Support0.0 DOT
Nay (31)0.0 DOT
How can I independently verify that these proposed settings are indeed more sensible and secure? Thanks
The proposed election score was calculated by halving the election scores of the last 30 elections (at the time of opening this proposal). You can check the last election scores based on the
ElectionFinalizedevent emitted by the election provider multi-phase pallet.We want to leave room for low election scores as the score is fluid based on the current nomination/stake in the chain but also because the staking miners may reach different solutions. However, we don't want the score to be too low, as the overall economic security of the elected set of validators can be at risk. This said, the minimum untrusted score is more of a backstop heuristic than the result of a well-known formula.
We're also working on an on-chain mechanism to adjust the minimum untrusted score automatically https://github.com/paritytech/substrate/issues/11965
These figures seems like they were derived in a very sensible manner. I hope this can tighten things up without risking another major slash.