Code With Me latency limits under concurrent editing on high-latency connections

Goal

Determine the maximum sustainable concurrent editing throughput for Code With Me sessions when participants are connected over high-latency network links, specifically measuring when cursor synchronization and edit transformation delays become perceptible.

Constraints

Code With Me relies on operational transformation coordinated through the host IDE's UI thread, with cursor updates using a lower-priority WebSocket channel than edit operations. JetBrains documents optimization for LAN scenarios but specifies no SLA or adaptive mechanisms for WAN latency. The host processes all operations sequentially, meaning latency scales with participant count and edit frequency, while external file system events add further contention.

Relevant uncertainty

The exact latency threshold at which concurrent edits from multiple remote participants cause noticeable degradation in real-time collaboration experience remains undocumented, particularly regarding the relationship between network RTT, participant count, and transformation queue depth.

Questions

• At what combined network RTT and participant count does Code With Me's operational transformation queue exceed 200ms processing time?
• Does cursor position broadcast frequency degrade measurably when edit operations exceed 10 per second per participant?