We spend hours optimizing API payloads and database queries to shave off milliseconds. Yet we accept days of latency in our human communication protocols without batting an eye.
When a frontend dev waits two days for a backend dev to clarify a JSON schema, that's latency. When a spec is misunderstood and a feature has to be rewritten, that's a packet loss.
The Async Trap
Remote work is great, but "async first" often devolves into "async only." We rely heavily on chat tools and PR comments. Text is low-bandwidth. It lacks tone, urgency, and nuance.
We've seen teams where two engineers argue in PR comments for three days about an implementation detail that could have been resolved in a 5-minute huddle.
Optimizing the Human API
1. Define the Protocol
Establish clear expectations for communication channels.
- Chat Tools: For non-urgent questions and status updates.
- PR Comments: For code-specific feedback, not architectural debates.
- Video/Huddle: For unblocking. If a chat thread goes back and forth more than 3 times, escalate to a call immediately.
2. Shorten the Feedback Loops
The longer the distance between doing work and getting feedback, the higher the cost of correction. This applies to code (CI/CD) and requirements.
Encourage "Draft PRs" or "Design Spikes." Share a messy prototype or a rough interface definition before writing the implementation logic. Get the feedback when the cost of change is near zero.
3. Observability for Collaboration
Just as you monitor server health, monitor collaboration health. Velocinator tracks Review Turnaround Time. If it's creeping up, it means your team's internal communication bus is congested.
Is one person the bottleneck for all reviews? Are PRs getting posted at 5 PM on Fridays? These are communication bugs. Fix them with the same rigor you fix production outages.