Protocol design
Rewards & curation
Reward concentrated support for good work. Give curators time to find it. Keep the payout fund bounded by its available balance.
Quadratic claims#
A post’s effective positive reward weight, after downvotes, produces a claim on the shared reward balance. The selected historical curve is:
The constant c is not yet selected. The linear term matters for smaller claims; the square term increasingly rewards concentration at larger weights. A larger claim changes a post’s share of the fund. It does not mint an additional reward budget.
The aim is a creator economy with a long tail and a meaningful reward for reaching the top. Zipf-like concentration is an inspiration, not a mathematical guarantee. Stake and coordination can also concentrate support.
Try the claim curve
Same fund. Different claims.
Two posts, one illustrative balance. Change the support and see how the claims compare.
Claims: 30,000 and 80,000. Post B has 2.67× the claim of Post A.
This compares two claims in isolation. It is not a rolling payout forecast: live payouts also depend on the fund balance, recent claim history and other cashouts.
A week of mana, on your schedule#
Positive voting capacity accumulates continuously in a seven-day reservoir proportional to stake. A curator can spend any available amount, including the full reservoir, on one post. There is no daily voting quota or weekly reset.
This lets a small holder save a meaningful vote for something they value. It also lets a whale concentrate a large vote. Unused capacity stops accumulating when the reservoir is full; future mana cannot be borrowed.
Stake transfers, delegation and withdrawal must not create fresh spendable mana. Exact accounting and vote-edit behavior need consensus tests.
Equal contribution, equal curation share#
Remove the early-voting advantage. Two eligible contributions of the same effective positive weight to the same post receive the same curation share, regardless of arrival time or order.
The proposed allocation within a post is proportional to each curator’s eligible contribution. The post’s aggregate claim is quadratic; individual curator shares are not squared. The author / curator split remains open.
Votes must still arrive before the eligibility cutoff. Equal treatment is not permission to vote after payout. Tests must cover strategic waiting, vote changes and interaction with downvotes.
Rolling payouts#
- Fund continuously.A designated portion of the 2% scheduled issuance flows into the content reward fund.
- Cash out posts individually.The inherited baseline is seven days after creation. Weekly mana does not create a weekly payout batch.
- Compare with recent claims.Decay the recent-claims ledger, add claims due in the block, then calculate payouts from the same block’s fund and claims snapshots.
- Keep the remaining balance.Deduct paid rewards. Claims remain in history and decay; they are not removed immediately after payment.
The modeling baseline uses Hive’s 15-day decay timescale. This is gradual decay, not a hard 15-day window. Launch needs an explicit initial claims policy so the first cashouts do not gain an unintended windfall.
Downvotes and contested rewards#
Hybla retains Hive’s downvote mechanic: stakeholders can contest a post’s reward allocation by reducing its net reward weight before cashout. This is part of the response to self-voting, vote circles, purchased votes and other disputed allocations.
Downvotes are not an automatic collusion detector. They require people to identify abuse and spend voting capacity. Large holders can also downvote, retaliate or coordinate. Testing must include honest disputes and false accusations, not just successful defense against an obvious ring.
The separate downvote budget and its interaction with seven-day positive mana need an explicit implementation. No special downvote multiplier, identity gate, self-vote ban or mandatory per-post payout cap has been adopted.
What the models have taught us#
Concentrated votes amplify both genuine popularity and coordinated capture. Banning direct self-votes does not catch an owner’s second account. Counting accounts does not establish independent people. In the tested rolling-fund scenarios, payout caps could encourage splitting claims across more posts and retain money in the fund for later capture.
These are scenario results, not forecasts. The next useful comparison holds the inflation budget and stake distribution fixed, then measures small holders, large holders, self-votes, coordinated rings, targeted downvotes and retaliation. Reward concentration and participant behavior must be observed in a testnet.
Upstream reference: reward calculation and cashout processing, Hive 1.28.8. Hybla’s changes still require implementation.