The decision had already been made. Not announced, not written down, but made, in the quiet way decisions actually get made, somewhere in the corridor between the pre-read and the meeting. What the room needed now was not a decision. It was a case. So it did the responsible thing, the thing that signals seriousness in every organization on earth. It commissioned the analysis.
And the analysis came back supportive, the way the tide comes back in. The team that ran it knew, without a single person instructing them, which answer was welcome in the building. They pulled the metrics that pointed the right way and left the ones that did not in the appendix nobody opens. They interviewed the stakeholders who were already enthusiastic. They found the two comparable companies that had tried this and prospered, and did not go looking for the twenty that had tried it and quietly bled out. The deck ran to forty pages. It was thorough. It was rigorous. It was also a verdict wearing the costume of an inquiry, and the room mistook the thoroughness for proof, when the thoroughness was the tell.
The search that only runs one way.
There is a difference between asking whether something is true and asking for the reasons it is true. From the inside, the two feel identical. Both involve data, meetings, late nights, a shared drive filling up with evidence. But one of them can end in no, and the other one cannot. When a room has privately decided and then goes looking, it is not testing the idea. It is assembling its defense, and like any competent defense, it calls only the witnesses who help.
Notice what the organization never did. It never said, before the first spreadsheet opened, what it would have to find in order to not proceed. No threshold. No disqualifying number. No finding that would be permitted to kill the thing. A real test names its own failure condition in advance and agrees to be bound by it. This one did not, which means there was no number the analysis could have returned that would have changed the outcome, which means the analysis was never an input to the decision. It was an output of it.
And the people doing the work knew. Nobody tells an analyst which conclusion to reach. They would deny the whole premise, honestly, if you asked them. But an analyst learns fast that the convenient finding and the inconvenient finding are not received the same way. The supporting slide is waved through. The contradicting slide is interrogated, sent back for more rigor, asked to clear a bar the supporting slide never had to clear. The burden of proof sits entirely on the evidence that threatens the plan. Do that for a few cycles and you no longer need to instruct anyone. The analysts instruct themselves, and call it reading the room.
So the contradictory evidence was never suppressed, because it was never allowed to form. You cannot delete the finding you did not fund the search to produce. And the most dangerous sentence in the entire deck was the one nobody actually wrote but everybody heard: we looked, and there was nothing against it. They did not find nothing against it. They did not look. The silence where the counterevidence should have been got read as the absence of counterevidence, which is the most expensive misreading an organization is capable of.
The objection writes itself. Sometimes the decision is correct, and the evidence supports it because it is genuinely supportable. True. But you cannot tell the difference from inside a confirmation search, and neither can the room, because the one procedure that would separate a sound yes from a flattered one, a real hunt for the strongest case against, is the exact procedure the room skipped. A right answer reached by a rigged process is not vindicated by turning out right. It was lucky. And luck is not a capability you can schedule for the next decision, which will arrive with the same blind spot and a thicker deck.
The confidence that should have fallen, rose.
Here is the part that should be uncomfortable. The entire point of gathering evidence is that it might change your mind. An honest inquiry can send you back to the room with less conviction than you walked in with. That is not a malfunction. That is the function. A confirmation search has no such setting. It can raise your confidence, and it can leave your confidence where it was, but it has no mechanism by which it can lower it, because it was never pointed at the thing that would. So the room walked out more certain than it walked in, and experienced that increase in certainty as learning, when nothing had been learned. The deck did not inform the decision. It upholstered it.
You did not gather evidence for the decision. You gathered permission for it, and then filed the permission under evidence.
And the thoroughness, the forty pages, the six weeks, the external validation, did not make this better. It made it worse. Every additional page of supportive material makes the conclusion heavier to move, not likelier to be right. The volume becomes the alibi. When it goes wrong later, and the reference class said it probably would, the deck is what gets pointed to. We did the analysis. We were rigorous. We could not have known. The rigor was real. It was just aimed in a circle.
What testing a decision would actually cost.
Rebuilding this is not a matter of respecting evidence more. Every organization already believes it respects evidence, and the one in this story respected it enormously, forty pages worth. It is about pointing the search at the thing you do not want to find. Before the analysis begins, somebody names the disconfirmer: the specific finding that would stop this, written down, agreed to, binding. Somebody is assigned to build the strongest possible case against, and is judged on how uncomfortable they manage to make the room, not on how safely they land. The question that opens the review is not what supports this. It is what would have to be true for this to be a mistake, and did anyone actually go and check.
This is the judgment no governance process catches, because the governance process received a beautiful deck and the deck said yes. It is only visible while the decision is being made, never in the postmortem, where the evidence already looks overwhelming because the overwhelming was manufactured. That is the layer SSUNDAR works in. In a simulation, under a clock, with new information arriving faster than anyone can curate it, you can watch whether a leadership team goes hunting for the signal that would stop it or only for the signal that lets it keep going. You can watch a contradicting data point land on the table and see, in real time, whether the room treats it as the most valuable thing in the exercise or as noise to be explained away before lunch. Most rooms explain it away. They have been practicing for years.
The analysis was never going to tell you no. You commissioned it after you had already decided, and it understood, better than you did, exactly what it was for.