Revamped reward formula that compensates for wasted ticks

[iddo]

Let me also summarize the landscape of our anti-splitting proposals (the links here have further details but they’re non-public for now):

1. Increase k2 and thereby lower the k2pow difficulty: the easier k2pow decreases the expected time and variance of PoST proof generation in each epoch.

   • This approach is bad news because it burdens the verification complexity of the entire network.
     • It’s better in terms of long-term fundamentals (in particular incentive-compatibility) that the individual miner will do more work locally and thereby reduce the workload for all the full nodes.
     • It might be ok as an optional feature (for non-newcomer ATX) after we have distributed verification.
     • More at https://community.spacemesh.io/t/non-trivial-improvements-to-post-ux-and-large-atx-incentivization/377/5

2. Increase the k2pow data size and thereby reduce the k2pow variance.

   • There’s no reasonable (non-VDF style) scheme to achieve this (so far).
   • Even if there was, the effect would be less significant relative to the other proposals.

3. k2pow delegation, which is a win-win.

   • It’s significantly more pleasant for an individual miner to pay for k2pow that has already been computed, so it’s strictly superior for our protocol to support k2pow delegation.
   • The outline of the spec is described at https://community.spacemesh.io/t/proof-of-work-and-post-proof-generation/361/16

4. Reward formula that compensates for wasted ticks, which is superior to our current reward formula (as explained in the OP).

5. Phased PoET, which is a win-win.

   • For example, if there are 14 PoETs such that one PoET starts on each day, then a miner that fails to produce a proof can re-try in the next PoET phase and thus lose only 1/14 \approx 7\% of the reward.

6. PoET with early exit points, which is a win-win.

   • The prudent miner can start from an early exit point, and then her PoST proof generation succeeds with a higher probability.
   • The early exit points are also essential as an anti-drift mechanism.

7. Rational reporting of the storage size: https://community.spacemesh.io/t/tree-free-post-in-more-depth/53/12

   • Possibly by using k2max scheme or variable-quality-with-fixed-k2 scheme.
   • This is a win-win, because it reduces the adversarial advantage relative to the honest miners, which means that we can lower k2 and/or k2pow.