ScorecardPublished monthly

Four numbers that say whether the loop is real.

Published on this page every month from the first month, even when the numbers are small. Sample values are shown in brackets until the first month closes.


[4] days

Request to decision, median

[6] days

Decision to ship, median

[23]

Changes shipped this month

[71]%

Started as a member’s problem

Sample values until the first month closes. Month: [month year]. How the scorecard works.


What each number proves

Number How it is measured Why it matters
Median time from request to decision Business days from the member’s post to the steward’s public status, median across every request decided in the month Proves someone is reading
Median time from decision to ship Business days from Planned to the flag turning on for everyone, median across every change shipped in the month Proves the AI team is real
Changes shipped this month Every changelog entry with the flag on for all members, removals included Proves cadence
Share of changes that came from members Shipped entries whose thread was opened by a member, divided by all shipped entries Proves the community steers

By product

The same four numbers for each product, so a steward’s cadence is visible on its own.

Product Request to decision Decision to ship Shipped this month From members
LevelUp [n] days [n] days [n] [n]%
Velocity [n] days [n] days [n] [n]%
Publish [n] days [n] days [n] [n]%
Producer [n] days [n] days [n] [n]%

Supporting numbers, published quarterly

The stewards’ numbers. They say whether the cadence above is being bought with quality.


What we will do with bad numbers

If the four headline numbers are good, the concept is true and the homepage can say so. If they are bad, the concept is a slogan, and the numbers will say that first. Either way they are published here, unedited, with the month they cover. Sample values are shown in brackets until the first month closes.


Want to move the numbers?

Post a problem. Every request becomes a row on this page.