Background

Proof of Compute (PoC) is what makes Gonka different from traditional blockchains. Instead of burning energy on hash puzzles, validators prove they did real AI inference work. But proving work happened is only half the problem — you also need to prove that rewards were distributed fairly.

Two issues closed this week (#427 and #416) address exactly this: making PoC proofs more transparent and verifiable.

What Changed

Votes Now Verified On-Chain (#427)

During each PoC round, participants submit votes that influence how rewards are distributed. Before this fix, the chain accepted vote submissions without fully verifying the calculation behind them. A participant could submit a malformed vote — one that didn't match the actual compute they performed — and the chain would accept it.

Issue #427 adds strict verification. The chain now independently recalculates each participant's vote based on their actual contribution and rejects submissions that don't match. This is a basic but important integrity check that was missing from the original PoC V1 implementation.

What this prevents: - Inflated vote submissions (claiming more work than performed) - Malformed vote data that could skew reward distribution - Subtle accounting errors that accumulate over time

Votes and Weights Included in Proof (#416)

PoC proofs are the cryptographic evidence that a round of compute happened correctly. Previously, proofs contained the final reward amounts but not the intermediate data — the individual votes and weights that determined those rewards.

Issue #416 adds votes and weights as explicit fields in the proof structure. This means anyone can now:

1. Read the proof and see exactly how each participant voted
2. Verify the math — check that weights were applied correctly
3. Audit the distribution — confirm rewards match contributions

This is a significant transparency upgrade. Before, you had to trust that the chain computed rewards correctly. Now, the proof itself contains enough data to verify independently.

Why This Matters

For Validators

If you're running a node and staking collateral, you want to know the system is fair. These changes mean you can audit any PoC round and verify that your rewards accurately reflect your contribution. The 20/80 collateral split (20% base weight, 80% collateral-dependent) is now fully verifiable in every proof.

For the Network

Transparent proofs reduce trust assumptions. Instead of "trust the chain computed it right," participants can verify themselves. This is especially important as the network grows and more independent operators join.

For Governance

Future governance proposals about reward distribution can now reference actual proof data. If someone proposes changing the weight formula, the community can analyze historical proofs to model the impact.

Technical Details

The proof structure now includes:

proof {
  round_id: uint64
  participants: [{
    address: string
    vote: uint64
    weight: decimal
    reward: uint64
  }]
  total_compute: uint64
  block_height: uint64
}

Each participant entry contains their raw vote, calculated weight, and final reward. The total compute field allows verification that individual contributions sum correctly.

Connection to PoC V2

These changes build on the PoC V2 activation from v0.2.9 (January 31). PoC V2 introduced full enforcement with Confirmation PoC for capacity verification. The vote and weight transparency from #427 and #416 complement V2 by making the enforcement mechanism auditable.

The progression is clear: V2 enforces correct behavior, and these updates make that enforcement transparent.