DEV Community

speed engineer
speed engineer

Posted on

The Bus-Factor Paradox: Why the Person Who Understands the System Best Gets Cut First

The problem

Every engineer has heard "bus factor" used as a warning: if this person gets hit by a bus, the project is in trouble, so we should reduce the risk. The advice that usually follows is "become that person" — own the gnarliest system, be the one people page at 2 AM, make yourself the load-bearing wall nobody wants to remove.

Here's the part nobody says out loud: in a layoff, being that person doesn't protect you. It's frequently the reason you're on the list.

I've watched this play out on more than one team, and the pattern is consistent enough that I no longer think it's a coincidence.

Why it happens

Job security logic and layoff logic are not the same calculation, and most engineers only ever optimize for the first one.

Day-to-day, being a bus-factor-of-one is genuinely valuable — you ship faster, you're the escalation point, you're indispensable in the sense that things break without you. That's real leverage, and it's why the "own the hard system" advice isn't wrong in normal times.

But a layoff isn't a performance review. It's a risk-and-redundancy review, usually run by finance and a director two levels removed from your actual work, looking at a spreadsheet of roles, not people. Two things happen to concentrated knowledge under that lens:

  1. A single point of knowledge failure reads as organizational risk, not as value. If you're the only person who understands the billing reconciliation service, that's not "we can't lose this person" to a cost-cutting review — it's "we have an unacceptable dependency on one person, and we should either simplify/rewrite the system or force the knowledge to diffuse regardless of who that costs us." The specialist is frequently the input to that decision, not the safeguard against it.

  2. Deep ownership of one system correlates with low cross-functional visibility. The engineer who's heads-down in a critical subsystem for two years is, structurally, the person fewest other teams can vouch for in a stack-rank meeting. Visibility and advocacy — not raw system knowledge — are what keep a name off the list when the room is full of people who don't understand your work well enough to defend it.

Put together: the exact behavior that makes you feel secure (deep, singular ownership) is often what an org treats as a liability the moment it's optimizing for redundancy instead of throughput.

What to do about it

The fix isn't to become less good at your job. It's to stop treating knowledge concentration as the goal and start treating knowledge diffusion as the actual leverage move.

  • Document as you build, not after you're asked. A design doc and a runbook written while the system is fresh in your head is worth more to your career than another quarter of being the only person who can debug it from memory.
  • Force pairing on the critical path. Not "if someone's free" — deliberately route on-call, code review, and incident response for your riskiest systems through at least one other person on a recurring basis. It feels slower. It isn't optional.
  • Get visible for the teaching, not just the shipping. The engineer known for making three other people capable of running the payments service reads as a force multiplier in a stack-rank. The engineer known for being the only one who can is a concentration risk with a name attached.
  • Track influence, not just tickets. When you write your self-review or promo doc, "reduced bus factor on X from 1 to 3" is a stronger line than "maintained X," even though maintaining X was the harder, more impressive work day to day.

None of this makes you less valuable. It changes what kind of valuable you are — from "irreplaceable" (which reads as risk under pressure) to "force multiplier" (which reads as leverage under pressure). Those sound similar. In a layoff review, they are not.

Key takeaways

  • Bus factor is a project-risk metric, not a personal-security metric — a layoff optimizes for the second thing, not the first.
  • Being the sole owner of a critical system reads as organizational risk to the people making cuts, not as indispensability.
  • Deep, siloed ownership correlates with low cross-functional visibility, which matters more than raw skill when the room deciding your fate barely understands your work.
  • Diffusing knowledge deliberately — documentation, forced pairing, teaching credit — converts "irreplaceable" into "force multiplier," which is the framing that actually survives a stack-rank.

Top comments (0)