The economy
Kreds rules
What each action is worth, and how to tune it for your team.
Kreds exist to answer one question honestly: who is moving this team forward?
Every number on this page comes from the published economic policy. Nothing here is invented, and nothing here is a suggestion. Where this page and the constitution disagree, the constitution is right and this page is a bug.
Two systems, deliberately separate
Kreds runs two scoring systems that never touch each other.
Contribution Points are recognition. They record that you did the work. They are cumulative, they are never spent, and they have no supply.
KRED is currency. It has a fixed maximum supply, a double-entry ledger, and rules about who may create it.
They never convert into each other, in either direction, at any rate. That is not a product decision that could be revisited next quarter. A conversion rate would turn reputation into a second money supply with none of the first one's controls, and every supply limit would become bypassable by minting points.
Contribution Points
| Action | Points |
|---|---|
| Merged pull request | 10 to 50 |
| Code review | 10 to 60 |
| Issue resolved | 5 to 20 |
Ranges, not flat values. A one line typo fix and a week of work are both merged pull requests, and paying them the same teaches the wrong thing.
Points do not decrease when you spend KRED, carry debt, or participate in the economy at all. They are adjusted only when the underlying contribution turns out not to have happened: a revert, confirmed fraud, or an actor reclassified as a bot.
KRED: what creates it
| Action | KRED | Source |
|---|---|---|
Pull request merged to main / master |
5 to 35 | Issued from the Central Bank reserve |
Scaled by a quality score. Passing checks, a description someone can read, resolved review threads, a linked issue and a sane diff size all feed it.
That is the short list on purpose. Shipping is the productive act, so shipping is what creates value.
KRED: what moves it
| Action | KRED | Source |
|---|---|---|
| Code review | 5 to 40 | Paid by the pull request author |
A review moves existing KRED. It does not create any. The author pays for help improving their work, the reviewer receives it less a 2% protocol fee, and the fee returns to the Central Bank reserve.
Review value 12.00 K
Author pays -12.00 K
Reviewer gets +11.76 K
Protocol +0.24 K
Why a transfer and not an award? Because if reviews created KRED, two accounts could review each other and print unlimited supply. No quality bar survives that, because the attacker controls both sides of it. The fixed supply would describe nothing.
What is worth nothing
| Action | KRED |
|---|---|
| Reviewing your own pull request | 0 |
| Reviews from bots and AI agents | 0 |
| Reviews submitted after the merge | 0 |
Every GitHub identity is classified as human, bot, AI agent or unknown, and only eligible humans participate economically. Dependabot and Renovate appear in your history at zero, which is correct rather than rude.
Unknown fails closed. An identity Kreds cannot classify does not get paid.
What happens when you cannot afford a review
You still get reviewed, and the reviewer still earns. What changes is where the money comes from.
1. Your settled balance
2. Your team's Review Fund
3. The Kreds review credit facility
4. A receivable the reviewer holds until it is funded
What never happens is a negative balance. A KRED balance is always zero or more. What you owe is tracked beside it as debt, and future earnings clear that debt before anything else becomes available to you.
So a heavy shipper with a review deficit reads as:
Balance 0 K
Debt 204 K
Net position -204 K
Not as minus 204 KRED, which would be currency that never came from anywhere.
Specialising is allowed
Kreds does not require you to review as much code as you submit. A maintainer who ships constantly and reviews rarely is doing a real job, not gaming anything.
The obligation to balance sits with the project, not with each individual. A persistent deficit gets financed by productive work, a treasury, a sponsor or limited credit. It does not get enforced as tit for tat, which would punish exactly the people carrying the most responsibility.
Where the work happened matters
An activity can earn Contribution Points without being eligible to create KRED. The bar for money is deliberately higher than the bar for recognition, and when the two standards disagree, the economy defers and reputation proceeds.
Opening a pull request in your own private repository and merging it yourself is not evidence Kreds can verify, so it creates no KRED. Making that repository public ten seconds ago does not change it either.
Two things can establish eligibility. A meaningful review from an independent, trusted human is one: a second party with something to lose is itself evidence. An established public repository with real history is the other, which is why solo maintainers of widely used projects participate normally.
Repository trust is a score over many signals that moves gradually. It is never a threshold on one number, because any single metric that unlocks issuance becomes a market.
Awards are immutable
Every award is a row: what happened, who it went to, the source GitHub event, the rules version in force, and the amount granted at that moment.
Changing a value affects future awards only. Raise the merge ceiling today and last month's merges keep what they were paid. A leaderboard that rewrites itself when someone edits a setting is not a leaderboard, it is a rumour.
Nothing is ever counted twice
GitHub redelivers webhooks on failure, on timeout, and whenever someone presses Redeliver. A redelivery must never pay twice.
Every award is keyed on the event type, the GitHub node id and the recipient, with a unique index behind it. A replayed event collides with the row already there and is dropped.
Note what the key is not: the delivery id. GitHub issues a new one on every redelivery, so keying on it would pay for the same merge again every time.
Two leaderboards
| Board | Ranks by | Answers |
|---|---|---|
| Economy | KRED net position | Who is economically liquid? |
| Contribution | Contribution Points | Who is doing the work? |
They are allowed to disagree, and when they do it is information. Somebody leading contribution and trailing the economy is shipping a lot and funding little of the review it took to get there.
Tuning it for your team
Values are configurable. The ratios are the part worth thinking about.
A team that under reviews. Raise the review ceiling. Reviewing becomes the efficient way to stay liquid.
A team drowning in tiny pull requests. Leave the values alone and set a challenge instead. Rewarding responsiveness does not distort the base scoring.
A team with wide seniority gaps. Scope the board by squad. A leaderboard with the same winner every week stops motivating everybody else.
A team gaming it. That is a signal, not a betrayal. Somebody just told you the incentives are wrong. Move the numbers.
This page renders content/economy/kreds-rules.md from the repository. Same file, whether you read it here or after cloning. Suggest an edit.