Vision: Path to removing reduce from NPoS protocol

[kianenigma]

The reduce.rs algorithm is the one that is responsible for causing, in almost all cases, one's stake to end up backing one validator, even if they nominate many. Note that this happens entirely post-election, and has no impact on the active validators, and their stake.

There are multiple issues with this:

1. It prevents nominators against further distribution of their rewards, hedging behind multiple validators.
2. It is super hard to explain that "in NPoS, you should nominate many validators to get the maximum benefit, but your stake will only go to one, but you should do it nonetheless, because it is post-election".

The main reason we have this is twofold:

• The solution is smaller if reduced, so easier verification, and less payload to submit.
• Rewards will be easier, given that there are less reward payouts to claim in total.

Once we have a scalable reward, slash, solution submission+verification system, there is no reason to have reduce anymore.

The linked issues/PRs for that would be:

• Bounded staking: Tracking issue for bounding pallet-staking storage items #255
• Multi-block submission: Tracking Issue for multi-block election + dedicated election system parachain  #461
• Having commit-reveal based submission will also help, but is optional Commit-reveal based solution submission for EPM substrate#13129
• scalable rewards: [NPoS] Unlimited number of nominators eligible for Rewards payout substrate#13498 / New staking reward system #439
• scalable slashing (not tracked in an issue).

What we can already do, if feasible, is to tweak the reduce algorithm to be configurable. The current behavior is to reduce 100% of the nodes, but we can then instruct it to reduce 80%, and gradually less and less.