The Man Paid to Disagree
Description
The first failure mode the prior series left unsolved was that experts lock onto an answer and cannot be overruled. The first concrete safeguard anyone built for it is a professional contrarian. After the 1973 Yom Kippur War — the worst intelligence failure in Israel's history, in which analysts had the data but had agreed together on the wrong story — the Agranat Commission's reforms produced a Devil's Advocate Unit inside military intelligence, known by the Aramaic phrase 'Ipcha Mistabra,' 'the opposite appears to be true.' Its only mandate is institutionalized dissent: examine the unlikely scenario, attack the Research Department's consensus, and force the question nobody wants asked, so the system never again suffers a 'failure to warn.' The clever part is structural: the unit's head is a colonel who reports directly to the head of Military Intelligence rather than to the department whose conclusions he is paid to tear apart — independence by design, so the critic cannot simply be silenced by the people he criticizes. But the safeguard has a fatal limit, and it showed on 7 October 2023: before the Hamas attack the head of the Devil's Advocate Unit reportedly tried repeatedly to warn that something was coming, and was overruled. The contrarian did his job perfectly; the decision-makers were simply free to listen and not act. The lesson: you can build the critic, protect the critic, give him a title and a direct line to the top — but a warning is not a veto. If those in power can nod, thank him, and proceed, the safeguard is a comfort, not a constraint. Dissent without teeth is permission to feel careful on the way to the same mistake.
- Devil's Advocate Unit — Wikipediahttps://en.wikipedia.org/wiki/Devil's_Advocate_Unit
- The devil's advocate in intelligence: the Israeli experience — Taylor & Francishttps://www.tandfonline.com/doi/full/10.1080/02684527.2018.1470062
- Head of IDF Devil's Advocate Unit tried repeatedly to warn — Times of Israelhttps://www.timesofisrael.com/liveblog_entry/head-of-idf-devils-advocate-unit-tried-repeatedly-in-september-to-warn-of-possible-hamas-attack/
Script
Cold open
What if the dangerous thing in a room isn't one bad analyst — it's a group of good ones who all agree? After a major intelligence failure, Israel put someone on the payroll whose full-time job was to disagree.
Frame
Across the last series, competent states kept failing the same way: the experts settle on an answer, and nothing moves them off it. This is the first concrete fix anyone built for that. Not a vote, not a court — an in-house contrarian with a mandate to argue the other side. It's a good idea. It also has a clear limit.
Where does this safeguard come from?
Where did it come from? Nineteen seventy-three. Israel was caught by surprise in the Yom Kippur War — its worst intelligence failure. And the problem wasn't missing information. The warning signs were there. The analysts had settled on one assessment — that war wasn't coming — and read the new evidence to fit it.
Why do smart experts fail together like that?
Why do capable people do that? Because agreement starts to feel like accuracy. When a room of experts converges, the consensus itself becomes the evidence, and the search for anything that contradicts it quietly stops. That's not really a knowledge problem. It's a structural one — which means structure can address it.
What did they actually build to fix it?
So what did they build? A small unit known by an Aramaic phrase — Ipcha Mistabra, 'the opposite appears to be true.' Its job is narrow: challenge the official assessment, take the scenario everyone's dismissing seriously, and put the dissent on paper. A devil's advocate, made permanent.
How do you keep a professional critic from being crushed by the people he criticizes?
The careful part is the wiring. To keep a professional critic from being overruled by the people he's criticizing, you place him outside their chain. The unit's head is a colonel who reports straight to the head of military intelligence — not to the research division he's there to challenge. Independence built into the structure, not left to goodwill.
So does mandated dissent actually stop the next disaster?
Does it work? Not on its own. On October seventh, twenty twenty-three, before the Hamas attack, the head of that unit reportedly warned more than once that something was coming — and was overruled. The dissent was produced. The decision-makers just weren't required to act on it.
Turn
So here's the limit. You can create the critic, protect the critic, and give him a direct line to the top — but a warning is not a veto. If the people in charge can hear it and proceed anyway, the safeguard doesn't constrain them; it mostly records that they were warned. Mandated dissent improves what a system knows. It doesn't change what the people in charge are willing to do about it.
Closer
So the first answer to experts you can't challenge is to build a challenger. It's worth doing — and it isn't sufficient by itself, because the final decision still sits with the people it was meant to check. Which is the problem the rest of this series keeps meeting. A guard only works if you can make him heard. So who makes the deciders listen?