The discipline letter was drafted and sitting in a folder, waiting for one last read. I want to describe that read, because it is the moment this whole piece turns on — and I am compositing here, as I always will in this space, several matters into one so that none of them is nameable. The habit I have tried to build is to read a file one final time before anything issues, not as the person who helped build it but as the union representative who is about to receive it: slowly, at arm's length, hunting for the thing I do not want to find. This time I found it. The very first interview in the file — the one the entire investigation had grown from, the trunk every later branch depended on — had been conducted with no notice of the allegations and no representation offered. The respondent had been called into a room, asked to account for themselves, and had answered, alone.
Everything downstream was built on that conversation.
Last week I argued that surprise is not strength — that process is, and that the open, deliberate path is what makes a record hold when it is read at arm's length a year later. This week is the harder half of that argument, because sometimes what you find at arm's length is your own hole. The question stops being how do I build a file that holds and becomes what do I do when I discover, before anyone else has, that mine doesn't.
There were, as far as I could see, two roads. The first was to press on and hope: the interview was months old, the answers were useful, the union had not raised it, and the letter was ready. The second was to void the interview, strike it from the record, and redo that stage properly — notice given, representation offered, the questions asked again in the light. Inside the building, the first road felt safe and the second felt like handing the union an admission of failure. In my experience it is almost exactly the reverse.
Here is what pressing on actually purchases. A flawed first interview does not stay contained; it contaminates. The union representative will find it — they read files for a living, and an unrepresented, un-noticed interview is precisely what they are reading for — and once they do, every finding that touches it becomes arguable, and the argument is no longer about what the respondent did but about what management did. The research bears out what practitioners feel in the room: in their study of three decades of harassment arbitrations, Abraham and Voos (2021) found that weak procedure pushes arbitrators toward leniency even where the misconduct is proven. You do not get to keep both the full penalty and the flawed process. One of them will pay for the other.
So we voided it. We wrote a short memo saying what had happened, why the interview could not stand, and how the stage would be redone — and that memo, I would argue, became part of the twenty percent of the file that carried the weight. The slow-moving beast doctrine from last week turns out to have a corollary I had not fully articulated until this case: if the organization has no room to act covertly against an employee, it has no room to act covertly about its own mistakes either. Self-correction done quietly is just a second flaw. Self-correction done in the open is process.
But the restart did not fully cure the flaw, and it would not have been honest to pretend it did. The respondent had still sat through that first conversation alone; what was heard in it could not be unheard. So when it came time to land on a penalty, we priced our own error in — arriving, before any grievance was filed, at roughly the number we judged an arbitrator would substitute. And this is not guesswork, because substitution is exactly what the framework provides for. The three questions that have anchored Canadian just-cause analysis since Wm. Scott (1977) are, in plain terms: did the employee give cause for some discipline, was the penalty excessive in all the circumstances, and if so, what should be substituted. Brown and Beatty's treatment of proportionality runs the same way. If you do not ask the second and third questions of your own file, someone else will ask them of it later — and answer them without you.
Which is why I have come to believe procedure is necessary but not sufficient. A clean process wrapped around a disproportionate penalty loses the same way a fair penalty wrapped around a broken process does. Proportionality is its own axis, weighed separately, and the file has to survive both readings.
What did all of this buy? Less, in that one case, than the untouched penalty would have been — and more, everywhere else, than I can easily count. The union representative who watches management void its own interview and discount its own penalty does not forget it. In my experience that is the moment the relationship shifts from adversarial toward something closer to two professionals who disagree about outcomes but trust the process between them. And that trust compounds, because you are not managing one file with a union; you are managing a hundred files over many years, and the way you handle the file where you were wrong sets the exchange rate for all the others.
The practical habits, if you want them: read your own file, before anything issues, the way the person across the table will read it — hostile, at arm's length, hunting for the hole. When you find one, fix it in the open and document the fixing; the memo that says "we got this wrong, and here is how we corrected it" is stronger paper than the flaw it cures. And when a flaw cannot be fully cured, price it into the penalty yourself, before the grievance, rather than paying for it at a worse rate at arbitration.
Here is what I have not resolved, and I want to leave it as honestly as I can. Pre-mitigating was fair to the respondent, and honest about how these matters actually get decided. But there is a third person in this story: the complainant, who did nothing wrong, and who watched the penalty shrink to pay for an error that was ours, not theirs. The cost of a procedural failure has to land somewhere, and in this design it lands partly on the person the process was supposed to protect. I have not fully worked out where the fairness in that sits — I am not sure the frameworks have either. That is where a future issue is headed.
If you have ever had to void your own process — or watched someone press on and hope — I would like to hear how this lands for you. Reply and tell me what it cost, or what it bought.
Sources & Further Reading
Steven E. Abraham & Paula B. Voos, "Procedural and Distributive Justice in Sexual Harassment Arbitrations: Evolution of Decisions in the Union Context," Advances in Industrial and Labor Relations, Vol. 26 (Emerald, 2021). Emerald Insight
Wm. Scott & Company Ltd. and Canadian Food and Allied Workers Union, Local P-162, [1977] 1 Can. L.R.B.R. 1 (B.C.L.R.B., Weiler, Chair). The source of the three-question just-cause framework applied across Canadian labour arbitration. Full text (PDF)
Donald J. M. Brown, David M. Beatty, Adam J. Beatty & Christine E. Deacon, Canadian Labour Arbitration (Thomson Reuters, looseleaf). On proportionality, mitigating factors, and the substitution of penalties. Thomson Reuters