Glossary
By Vaibhav Verma
Bus factor is the number of engineers whose loss (literally or otherwise) would cripple your product. A bus factor of 1 means one engineer can take critical knowledge with them. A bus factor of 3+ on critical systems means you have redundancy.
Critical path bus factor of 1 is an organizational crisis waiting to happen. Your top engineer knows authentication. They leave. You're now scrambling to rebuild.
You have a critical service that handles payments. One engineer wrote it three years ago. She's the only one who understands the codebase. She knows the migration history, the quirks, the data model, the deployment process. Everyone else touches it with fear.
That's bus factor 1. If she leaves, you have months of recovery: debugging code no one understands, discovering gaps in documentation, learning systems that should have been redundant.
Compare to well-run teams: every critical system has 2-3 engineers who understand it deeply. Knowledge is distributed.
"Bus factor is just about retention." Not really. Even with great retention, you need redundancy.
"We'll document everything, so bus factor doesn't matter." Documentation helps, but it's incomplete. Code has implicit knowledge only humans carry.
"Smart engineers can learn any system quickly." Some systems are legitimately complex. Bus factor acknowledges this friction.
Bus factor is risk. High bus factor on critical systems means: single point of failure, longer incident resolution, slower feature velocity, knowledge loss when people leave.
High bus factor teams struggle with delivery because knowledge is concentrated. They can't parallelize work. They're fragile.
Knowledge distribution: For each critical service, how many engineers understand it well? Rate: 1 (crisis), 2-3 (good), 4+ (redundant).
Code review distribution: Who reviews code? If one person reviews 80%+ of changes, that's your bus factor.
Onboarding time: >3 months on critical systems suggests bus factor is too low.
Incident response time: >1 hour suggests knowledge is too concentrated.
Q: How do we increase bus factor on legacy systems? Pair programming, documentation, mentoring. It's time-intensive but worth it.
Q: Is it realistic to have bus factor 3+ on all systems? No. 20% of systems are critical; 80% are lower priority.
Q: How does this relate to knowledge silos? Bus factor is a measurement of knowledge silo risk.
Q: Can Glue help? Absolutely. Glue analyzes code ownership and surfaces bus factor immediately.
Keep reading
