import CountmeAnalyticsCharts from "@site/src/components/analytics/CountmeAnalyticsCharts";
import ContributorMetricsCharts from "@site/src/components/analytics/ContributorMetricsCharts";
Contributor Metrics#
Methodology, Privacy & Opt-Out#
Counting Methodology#
Bluefin and Bluefin Utah use countme.projectbluefin.io. Classic is
ublue-os/bluefin, counted by ublue-os/countme.
All Bluefins adds Bluefin, Bluefin Utah, and Classic on each date. Classic uses the
recorded Bluefin badge history from ublue-os/countme; its last published weekly
reading is used until the next publication. Bluefin and Utah use their own daily
counts. Missing daily counts leave gaps. The badge values are rounded, so the
combined chart is an estimate, not a deduplicated device inventory.
First-party Countme Service & Privacy Guarantees#
Project Bluefin Dakota and Utah images send a daily active-system ping from the
projectbluefin-countme systemd timer to https://countme.projectbluefin.io/v1/ping.
Other images do not send it.
- What is sent: a single string,
<image-name>/<image-flavor>:<stream>, where the stream
is the booted image's tag (stable,testing, orunknown). Nothing else: no install age,
architecture, machine ID, username, or persistent identifier. - How often: at most once per day per system, and never from a live ISO, installer, or
container. - What is stored: one counter per UTC day and image. No per-system rows and no IP addresses.
A day's count is the number of systems active that day, not distinct devices over time.
Opt-Out Instructions#
To stop Project Bluefin host countme reporting now and across reboots, mask its systemd timer:
sudo systemctl mask --now projectbluefin-countme.timer
A mask of the previous bluefin-countme.timer or bluefin-countme.service units, and the
/etc/projectbluefin/countme/disabled opt-out file, still stop reporting.
To disable countme reporting for a printer container, create a .countme_disabled file in its state directory (for ChairLift, in ${XDG_DATA_HOME:-$HOME/.local/share}/chairlift/printers/<family>/<instance>/.countme_disabled on the host, which maps to /var/lib/<family>-printer-app/.countme_disabled inside the container):
touch "${XDG_DATA_HOME:-$HOME/.local/share}/chairlift/printers/<family>/<instance>/.countme_disabled"
To disable upstream Fedora countme:
To disable Homebrew analytics:
brew analytics off
See the Homebrew analytics documentation for additional details.
Contributor Metrics Methodology#
These charts measure participation, not pull request throughput. A contributor is
one identifiable GitHub account with a recorded public contribution in a repository
currently owned by projectbluefin. Each account is counted once across the organization.
- Ways to contribute: code and documentation changes, pull requests, issues,
comments, submitted reviews, and GitHub Discussions, including replies. Changes
do not have to be merged to recognize their authors. - Repository coverage: all public organization repositories, including archived
repositories and forks. Commit authors are read from non-merge commits on non-fork
default branches; imported fork commit history is excluded. Other participation
in forks is included. Recorded history may predate a repository joining the org. - Community growth: the cumulative number of accounts by their first recorded
contribution date. This is surviving public history, not a claim to recover deleted
contributions or activity in private repositories. - Recent participation: distinct accounts active in the last 90 days. Monthly
counts use UTC calendar months; the current month is incomplete. The same person
may help in multiple ways, so the contribution-type counts must not be added together. - Identity limits: known bots and bot account types are excluded. Accounts that
cannot be linked to a public GitHub identity are not counted. Public GitHub data
cannot establish whether an ordinary account used AI assistance, or whether two
accounts belong to the same person. - Freshness and gaps: the scheduled collector follows every required page of
the original public GitHub APIs. A failed collection retains the last successful
snapshot and its timestamp, with a visible reason rather than a partial total.
Sources: GitHub's repository API,
GraphQL repository reference,
and pagination documentation.
The precise local definition is implemented in scripts/fetch-contributor-metrics.js.
Community forum discussions at ublue-os/bluefin
are outside this organization and are not mixed into its totals.