Your Most Uncomfortable Engineer Is Doing Necessary Work
What happens when you misread the engineer who won't let the half-truth land
An engineer, call him Marcus, joined the postmortem already in progress. The team was doing the thing teams do: talking around the outage, assigning it to “an unusual confluence of factors,” nodding at each other in a way that was really just nodding at the story they’d agreed to before he walked in. He asked one question. Something like, “Wait, we had this same failure path six months ago. Did we close that action item or just move on?”
The room went quiet. Not hostile, exactly. More like the quiet of a system encountering an unexpected input.
I’ve watched this pattern across enough engineering organizations that I’ve started thinking of it as a type, not a personality quirk. Some engineers seem structurally incapable of letting a half-truth settle. They’re not being difficult. They’re not performing skepticism for status. Something in the way they’re wired makes them keep pulling on the thread after everyone else has agreed to stop.
There’s a Jungian concept for this, one that circulates in psychology-adjacent corners of the internet. The “heyoka” empath: a figure, borrowed from Lakota tradition, who functions as a disruptive mirror rather than a soother. Where typical empaths absorb the emotional tenor of a room, the heyoka reflects it back at uncomfortable fidelity. The term is contested territory, culturally and empirically, and I want to be careful not to import it wholesale into an organizational framework it wasn’t built for. But the behavioral profile it describes maps, with some precision, onto a recognizable archetype in engineering teams. The engineer who asks the question that makes the room go quiet. The one who keeps the issue tracker accurate when everyone else is rounding up toward green.
Call it the mirror function.
The coaching that suppresses the signal
The first trait in the Vegoia piece is making people uncomfortable without trying to. This is, in engineering organizations, both underappreciated and actively managed out. We pattern-match it as a communication problem. “Marcus has trouble reading the room.” “She doesn’t know when to drop something.” The coaching response is almost always about softening: ask questions less sharply, wait for the retrospective, learn to let things land before pushing.
What that coaching misses is the operational cost of suppressing the function. The engineer who’s been told often enough that their timing is off learns to time their observations more carefully, which usually means less often. The issue gets filed. The postmortem closes. The failure path stays open. I’ve written about how shame operates inside postmortems and the way that even nominally blameless processes develop unwritten social norms that punish precision. The mirror engineer isn’t violating the postmortem’s stated rules. They’re violating its unwritten ones.
There’s something in [Amy’s] engineering leadership that I think about here. She ran a platform team at a mid-size company I consulted with, and she had a practice she called “the slow no.” When a proposal came to her that she thought was wrong, she wouldn’t say no in the meeting. She’d wait. She’d ask questions in the doc. She’d “align.” By the time the project launched and hit the predicted failure, she was the most diplomatic person in the room. She was also the least useful one. The mirror function, suppressed over years of feedback about her delivery, had become a survival mechanism that served her career and cost the organization three quarters of a year.
This is a known pattern in the research on organizational silence. Morrison and Milliken’s foundational 2000 paper in the Academy of Management Review identified silence as a collective phenomenon: organizations develop powerful structural forces that cause widespread withholding of information about problems, not because people don’t see the problems, but because they’ve learned that surfacing them is the riskier move. Amy’s “slow no” wasn’t a personality quirk. It was a rational response to an organization that had consistently punished precision.
What the performance review doesn’t capture
The second trait is truth-telling at social cost, which in organizational terms is usually called “not knowing how to play the game.” And there’s something to that. The engineers who function as mirrors often have thin political skin in the sense that they haven’t learned, or refused to learn, the set of small deceptions that organizational life usually requires. The slightly inflated confidence in a status update. The framing that attributes a missed deadline to external factors. The project postmortem that identifies “process gaps” rather than the specific decision by the specific person that caused the specific failure.
Organizational life runs on managed ambiguity. The mirror engineer introduces precision into systems that rely on its absence. Technical debt is a people problem is a piece I keep coming back to here, because the same dynamic operates: the thing that looks like a technical issue is downstream of a social arrangement where accurate diagnosis was too costly to surface when it mattered.
This is where the research gets uncomfortable. Ethan Burris’s 2012 study in the Academy of Management Journal found, across three studies, that managers rate employees who engage in challenging voice as worse performers than those who engage in supportive voice. The effect was mediated by perceptions of loyalty: the challenging engineer gets read as a threat, not as a diagnostic. The performance review isn’t capturing performance. It’s capturing social alignment, and then labeling the output “performance.”
What’s worth distinguishing here is the difference between the mirror engineer and the engineer who’s simply oppositional. Marcus, in the postmortem, wasn’t complaining — he was asking about a closed action item. The information he was surfacing was technically relevant and operationally significant. The discomfort in the room was about what the answer would mean, not about Marcus’s behavior.
When managers report “friction” with a particular engineer, it’s worth asking what the friction is actually about. Friction around delivery or communication style is one thing. Friction that arises specifically when accurate information enters a room is something else entirely.
The rebuild instinct
The chaos-transformation trait, the third one in the source piece, is where the organizational translation gets more complicated. The heyoka pattern involves sudden, complete transformations rather than gradual evolution. In engineering terms, this shows up as the engineer who completely rebuilds a system rather than incrementally improving it, who proposes migrating the database when everyone else is debugging at the query level, who leaves abruptly when an organization refuses to change.
I want to be honest that this trait contains a shadow. The mirror function, operating without self-awareness, can look like chaos-generation for its own sake. There’s a version of the engineer who disrupts systems that work well enough because they can see the version that would work better, and the cost of the disruption is real and often lands on other people. The Vegoia piece frames this as a feature. It’s more accurately a capacity that requires calibration.
What the organizational literature usually calls for here is change management: slowing the impulse to rebuild until the case for rebuilding is legible to others. That’s not wrong. But the engineers who have the strongest instinct to rebuild are often the ones who can see the underlying architecture most clearly. The question of when to listen to them and when to ask them to pace themselves is hard, and most engineering organizations resolve it by defaulting to caution, which means they resolve it wrong about half the time.
When the ambient tension breaks the surface
The energy-reversal trait, the fourth one, is probably the most interesting in organizational terms. The Vegoia piece describes it as reflecting rather than absorbing: if someone projects anger, the heyoka mirrors it back until the person is confronting their own emotion rather than lodging it somewhere else.
In engineering organizations, the closest analog is the feedback that goes nowhere until someone finally makes it visible. The team that’s been quietly carrying an architectural decision they hate for eighteen months. The accumulating tension around a deadline nobody believes in. The mirror engineer doesn’t absorb the ambient stress and keep moving. They ask, in a team meeting, “Do we actually think we’re going to hit this?” and suddenly the ambient stress is on the table, attributed, requiring a response.
This is disruptive. It is also, in many cases, how the thing that was true stops being undiscussable. The gap between what organizations believe about their own decision-making and what their decision-making infrastructure actually supports is something I’ve written about at length, particularly around how hive mind dynamics displace explicit evaluation as teams grow. The mirror function is one of the mechanisms that closes that gap, when it works. When it doesn’t work, it just increases ambient tension and the engineer leaves within a year.
The blameless postmortem literature is relevant here. Teams that make uncomfortable information discussable catch failure paths earlier. Teams that don’t repeat the same incidents under different names. The mirror engineer is, in effect, doing the work the blameless postmortem process is supposed to institutionalize. The difference is that one happens in a structured container with explicit norms, and the other depends on an individual willing to absorb the social cost of accuracy.
The three traits I’m compressing, and why
I’m going to compress the remaining traits because I think the pattern is visible and I don’t want to march you through a list. Humor-as-truth-delivery is real and recognizable in engineering culture — the person who makes the joke in the all-hands that everyone laughs at because it’s the thing nobody was supposed to say out loud. Emotional depth without emotional dysregulation is what makes the mirror function sustainable over time; the engineer who can sit with a room full of discomfort without becoming part of the discomfort is more useful than the one who can’t. The catalysis-at-cost trait is the loneliest one: the person who surfaces the problem often isn’t around for the resolution.
That last one has a career-outcomes dimension worth naming. A longitudinal study by Seibert, Kraimer, and Crant tracking 496 employees found that proactive personality predicts both salary and promotion gains. Which sounds like good news for the mirror engineer. But Burris’s findings complicate it: the same proactivity that drives initiative also drives challenging voice, and challenging voice gets read as disloyalty. The career outcomes depend heavily on whether the organization has built a container for the function.
What gets misread as culture fit
What I keep coming back to is that the heyoka concept, whatever its origins, is describing something real about how certain individuals relate to organizational truth. And organizational psychology has mostly treated this function as a communication problem to be coached away rather than a capability to be understood and, where possible, supported.
The mirror engineer is probably on your team right now. They might be the one your last performance review described as “brilliant but difficult.” They might be the person who raised the issue about the failure path six months before the outage. They might be the one who left, and whose departure everyone attributed to culture fit, and whose specific concerns about the system came true about nine months later. The self-awareness literature in engineering leadership tends to treat self-awareness as a leadership development goal — I wrote about the gap between the two in self-awareness as a scaling mechanism. The mirror engineer often has a surplus of self-awareness about the organization and a deficit of protection from what that awareness costs them.
Morrison and Milliken’s research notes something painfully ironic about organizational silence: it remains prevalent even in organizations that proudly describe themselves as having open communication cultures. The gap between the stated norm and the operating norm is exactly where the mirror engineer lives, and exactly where their function tends to get misread.
Building an organization where reflection lands somewhere instead of ricocheting is the harder problem. That part, there’s no clean answer for.