Everything that survived has someone else's name on it: ten years of Lyft's platform
Which of the components a company builds for its own platform are still alive ten years later, traced across Lyft's published artefacts from 2015 to 2026: what it donated, what it archived with a dated notice, what it left to drift, and the one variable that separated the two groups.
Reads eleven of Lyft's published platform components through git history, governance files and the package, container and module registries that date their distribution. A reader leaves able to predict which in-house component will survive, to price the deprecation calendar that arrives with a donated project, and to run the same audit over their own organisation's repositories in an afternoon.
Donation preserved the software and transferred the roadmap: every lyft.com address in Envoy's governance file now sits under Emeritus or Friends, the founder remains a senior maintainer with a personal address, and the DynamoDB filter Lyft said it used for all its DynamoDB traffic was demoted to contrib in 2022 by a maintainer at another company.
What you get out of it
- The import path predicted survival. Every component still shipping in 2026 is owned by someone other than Lyft, including the asset graph that joined a foundation in late 2024 and now lives under a cartography-cncf organisation; every component that kept a lyft/ path and a single operator is archived, dormant or drifting.
- Lyft-addressed commits to Envoy fall from 26 percent in 2016 to 3 percent in 2022 and zero across 2025 and 2026, a measurement that is a floor because contributors commit from personal addresses, but whose direction is unambiguous.
- Donating transfers the deprecation calendar: the rate limiter deleted two wire protocols on the project's schedule, the validator is now in maintenance mode behind its new owner's replacement, and the workflow engine's Lyft-era generation is maintained on a side branch after the April 2026 cutover.
- Retirements arrive in cohorts and arrive late. The registry was archived five years after the proxy replaced the need for it, and the secret store and the console were archived two months apart in 2025.
- Publication cadence, not the README, is the liveness signal: Confidant's container images stopped four years and two months before its archive notice, and its client shipped a month after it.
Scope
Why this, now. The 2025 archive wave closed Lyft's in-house console and secret store within two months of each other, and in 2026 the projects it gave away cut over to new generations under new owners, so the whole decade is finally legible as a single outcome.
What it does not cover. Lyft's internal architecture, traffic, fleet size and costs, and the reasons behind each retirement. This session's egress policy resolved only GitHub, PyPI, npm, the Go module proxy, pkg.go.dev, Maven Central and Docker Hub, so the guide contains no engineering-blog, talk, paper or financial evidence, and the two internal ticket identifiers in the corpus are the only clue to motive.
Other field guides
When one platform runs everything: eleven years of Monzo's architecture
A single-company dig, reconstructed from Monzo's own incident reports and engineering posts, its conference talks, and the archive banners and commit…
31 sources · 19 organisations · 3 postmortemsChanging what the services stand on: ten years of ByteDance, read from its own repositories
An organisation past a certain size cannot ask its services to change, so when the layer beneath them becomes the constraint it replaces that layer a…
24 sources · 7 organisations · 5 postmortemsHardening the wrong plane: ten years of Cloudflare's architecture
A single-company dig: Cloudflare, 2016 to 2026, reconstructed from eight of its own incident reports, its engineering accounts of Pingora, Quicksilve…
30 sources · 12 organisations · 9 postmortems