What Was I Supposed to Do?

He was let go in February, for performance.

That was the stated reason and it may well have been the real one. Six years of what the preceding essays have described does eventually show up in a person's work, which is one of the crueler efficiencies of the whole arrangement: the machine damages your output and then cites your output.

A while later he ran into someone from the old building, and learned that the version circulating there was different. He had been fired, the story went, for sexual harassment.

There had never been a finding. There had never been a proceeding. The thing that had followed him for six years had, in the end, outlived his employment and written his epitaph, and the building believed it, because by then the building had believed it for years.

Now, the question this series has been walking toward.

What did that cost?

Not what did it cost him. We've spent seven essays on that. What did it cost the organization? Put a number on it. Six years of a functioning employee's degraded output, the hours his supervisors spent managing around a situation nobody would name, the colleagues who watched and drew conclusions about what this place does to people, the ones who left and gave a different reason, the replacement hiring, the retraining, the institutional memory that walked out the door with him.

There is no number. Nowhere in that organization, or in almost any organization, does a figure exist for any of it.

That absence is not an oversight. It is the mechanism. It is the reason all of this persists, and it is where the leverage is.

Here is what we can count, and it is enough to be going on with.

Housman and Minor's analysis of 58,542 workers, from Part 4, put the avoided cost of a single toxic worker at roughly $12,489, and they got there by counting almost nothing: turnover alone, no litigation, no lost customers, no morale, no investigations. It is a floor, not an estimate.

O'Reilly and colleagues, from Part 2, linked exclusion to who had actually resigned three years later, which is a time horizon no quarterly review has ever contained.

The Workplace Bullying Institute's 2024 national survey found 62% of targets lose their jobs, against 27% of perpetrators facing any consequence at all.

And that is roughly where the defensible numbers stop. You will find confident claims online about workplace bullying costing the American economy some specific number of billions. I went looking, and I am not going to repeat any of them, because they trace back to estimates built on estimates, and this series does not get to abandon its standards in the final essay for the sake of a satisfying figure.

The honest statement is smaller and more damning: the aggregate cost of this is unknown because no one has ever been required to measure it.

Now the part I would rather not tell you, and would be a coward to omit.

The Cochrane Collaboration, which produces the most rigorous evidence reviews in health and social science, examined interventions for the prevention of workplace bullying. Their conclusion was that there is only very low quality evidence that organizational and individual interventions prevent bullying behaviors at work. Not that they fail. That the research is too thin, too poorly controlled, and too inconsistently measured to tell.

Set that against what every large employer in America currently does about this, which is training. Annual modules. Scenario videos. A completion checkbox.

There is no strong evidence that any of it works.

It is worth being exact about what that finding is and isn't. Cochrane found an absence of evidence, not evidence of absence. It does not establish that training fails. It establishes that after decades of corporate investment, nobody has produced research good enough to say either way, which is its own kind of indictment.

But notice what the weakness of the evidence actually tells you, because it is not nothing and it is not despair.

If the programs are unproven and the costs are unmeasured, then the highest-leverage move available is not another program. It is measurement. It is closing the specific gap that Part 4 identified, where the benefit of tolerating someone is concentrated and attributed while the cost is diffuse and anonymous.

That gap is closable, and closing it requires no legislation, no new department, and nobody's permission but the person who owns the reporting.

Attribute turnover to individual managers rather than to departments. Ask departing employees a specific question rather than a general one, and count the answers. Track how long complaints take to reach an outcome, and publish the distribution. Look at whether the people who report things are still employed a year later, which is a number every HR system in the country could produce this afternoon and almost none has ever been asked for.

I cannot tell you that any of this works. Nobody has tested it, and I have just spent a page explaining why you should be suspicious of untested prescriptions. What I can tell you is that these are the only proposals on the table that require nobody to become a better person, and that they make a cost visible in the exact place where the decision gets made, which is the one condition this entire machine has never had to operate under.

You cannot price what you refuse to count. Every organization in this series was not being evil. It was being innumerate, in a direction that happened to be convenient.

And then there is the other thing, which costs nothing at all.

Go back to the vending machines.

A woman who did not like him, who had no stake and no relationship to protect, decided a story she'd heard sounded strange, and told him so, and went back to work. She never thought about it again. She would be astonished to learn that a stranger wrote about it years later.

And there was a second one, in Part 7, easy to miss. A different colleague went to the person spreading it and asked her to stop. That conversation ended the campaign in a single afternoon, which is more than the complaint process, the policy, the training, or the law managed in six years.

Two people. Neither of them senior, neither of them close to him, neither of them running any risk to speak of. Between them they account for every effective intervention in this entire series.

Against that, the Workplace Bullying Institute's figure: 98 targets out of every hundred report that nobody did either of those things.

So let's take stock of what's gone.

The target was supposed to be weak. The best evidence says there is no victim type at all, and that the brittleness you observe in someone six years in was substantially manufactured by the treatment you are using it to explain.

The cruelty was supposed to be visible. A third of the standard measurement instrument is social exclusion, a category defined by absence, and the exclusion does more damage than the shouting.

Help was supposed to come. Ninety-eight targets in a hundred report that no colleague stepped in. The complaint apparatus was substantially built as a liability shield after 1998. Retaliation has been the most common EEOC charge for seventeen consecutive years. And the law generally does not reach any of it unless you can tie it to a protected characteristic, which is why thirty-two states have looked at fixing that and none of them have.

Every assumption we were issued as children is gone. Which leaves the thing this series was named for, and its final form.

You have spent eight essays looking for the adult in the room.

There isn't one. You're it.

Not in the sense that you can fix any of this. You can't, and anyone promising otherwise is selling something. In the narrower and much more inconvenient sense that in the specific room you are actually standing in, you are the only mechanism documented anywhere in this literature to have reliably worked, and you have been sitting on it, and so have I.

We tell children to find an adult, and it is good advice, and it expires.

What we never tell them is what to do afterward, when they are the adult, and the room is looking at its shoes, and the whole machinery of the place has quietly agreed that nothing is happening. We give them twelve years of instruction on how to be rescued and almost none on how to be the person who says the strange thing out loud.

That is a curriculum problem, and it is fixable, and it would cost a school district nothing but the decision to teach the other half of the lesson.

You have been thinking about someone.

You have, somewhere in these eight pieces, thought of a specific person. Maybe someone you watched it happen to and told yourself it wasn't your place. Maybe someone you priced. Maybe someone you were, and have never quite put down.

That person is still reachable. Most of them are.

So, the last of it, in the only form that has ever actually moved anything in these eight essays.

No budget. No authority. No policy required. One sentence, said out loud, to someone who has been given no reason to expect it.

What was I supposed to do?

You already know. You have known since Part 3.

The only open question is whether you will be the one who does it, and unlike everything else in this series, that one is entirely yours to settle.

This concludes The Adult in the Room.

Sources

• Gillen, P., et al., Interventions for prevention of bullying in the workplace, Cochrane Database of Systematic Reviews. Concluded that only very low quality evidence exists that organizational and individual interventions prevent workplace bullying, citing few controlled trials, inconsistent outcome measurement, and insufficient follow-up.
• Housman, M., & Minor, D., Toxic Workers, Harvard Business School Working Paper 16-057. 58,542 workers; approximately $12,489 in avoided turnover cost per toxic worker avoided against approximately $5,303 in value from hiring a top 1% performer.
• O'Reilly, J., et al., "Is Negative Attention Better Than No Attention? The Comparative Effects of Ostracism and Harassment at Work," Organization Science. Summary via Association for Psychological Science.
• 2024 WBI U.S. Workplace Bullying Survey. Zogby Analytics, n=1,024 U.S. adults, September 23–25, 2024, margin of error ±3.1 points. Coworker intervention and support 2%; 62% of targets lose their jobs; 27% of perpetrators face any negative consequence.
• U.S. Equal Employment Opportunity Commission FY2024 enforcement data. Retaliation alleged in 42,301 of 88,531 charges, the most frequently filed charge for the seventeenth consecutive year.

Identifying details throughout this series have been removed or changed.