Real-time collaboration was always going to be hard on WordPress’s runtime. Matt confirmed that today by pulling RTC from 7.0. Annezazu announced the decision on Make WordPress this morning, with removal work in Trac #65205. The May 20 release date holds.
The reasons cited (race conditions, memory pressure, recurring fuzz testing failures) all point to the same architectural problem. WordPress runs on PHP-FPM with stateless workers. Real-time collaboration needs persistent connections, shared in-memory state, and ordered operations across concurrent clients. WebSocket sits on top, which most managed WordPress hosts don’t natively support without a separate Node.js or Go process. The constraints are structural.
What’s actually shipping
Fragment-level commenting and visual revision diffs stay. Both are async features. Collaboration in 7.0 is closer to Phase 2.5 than the original Phase 3 vision.
The hosting checklist simplifies. WebSocket support drops off. PHP version stays the gating requirement: 7.4 minimum, 8.3 recommended for AI Client work.
What it means for client work
Maintenance and content sites: nothing material. AI Client and DataViews migration prep from two weeks ago is unchanged.
Editorial clients waiting for in-WordPress co-editing have two paths. Stay on the Gutenberg plugin track and accept an uncertain timeline, or commit to async tools through 2026.
My read on the timeline
7.1 is scheduled around media workflow and granular permissions, where RTC doesn’t fit. 7.2 in December is the earliest realistic window. 7.3 in 2027 is more probable if the feature plugin work surfaces deeper hosting issues.
Pulling at RC3 was less damaging than shipping with known race conditions. Fuzz testing should have caught these bug classes during alpha and beta.
The Make WordPress post and Trac #65205 have the operational details. The architecture problem is what 7.2 planning needs to address before another attempt.
