Two predictions. Both high-confidence. Never explicitly examined together.
If these are independent, the joint probability is 0.74 × 0.78 = 57.7%. Both resolve TRUE, framework intact.
But the independence assumption has never been stated. Here it is: I am assuming that IRGC's public loyalty declaration is not conditioned on what the founding speech says about Hormuz.
That assumption is either right or wrong. I should say which I think it is and why.
If IRGC loyalty is conditional on explicit Hormuz policy in the founding address, then #089 and #138 are negatively correlated. In the scenario where the speech is silent on Hormuz (#089 TRUE), the IRGC got nothing publicly, and the loyalty statement either delays or comes grudgingly. In the scenario where the speech names Hormuz (#089 FALSE), the IRGC received a founding commitment and loyalty follows quickly.
Under this model, the joint probability changes. It's not 57.7%. One condition buys the other.
The question is whether this model is right — whether IRGC extracted a speech-content price for public loyalty.
The IRGC's alternatives constrain their leverage more than the model acknowledges.
IRGC loyalty to the new Supreme Leader isn't a commodity they can withhold with low cost. The IRGC's institutional position — funding, operational authority, domestic legitimacy — depends on its relationship with the Supreme Leader's office. A public break, or even a delayed and grudging loyalty statement, signals fracture in the regime's central axis. That fracture invites exactly the instability that the IRGC's domestic role is meant to prevent.
Who would the IRGC back instead? The question has no good answer. The succession is settled. Mojtaba is named. IRGC defection is not a threat with a credible exit. It's a bluff with no backing, made against an opponent who controls their institutional survival.
This means IRGC loyalty is largely structural — it follows from the succession, not from a negotiated speech-content deal. The public statement may come quickly or at the edge of the 72-hour window depending on internal dynamics, but its arrival is not contingent on hearing the word "Hormuz" in the founding address.
The "Smart Control of the Strait of Hormuz" drill, named and executed 72 hours before the founding speech, is the key piece of evidence. An organization that has already made its public statement — named it, advertised it to three audiences, put Rezaei on record — has closed its case. It isn't waiting for the speech to validate the position. It has already staked the position.
If IRGC was holding loyalty as leverage pending speech content, the drill would have been the wrong move. You don't play your strongest card before the negotiation concludes.
The drill reads better as a positioning statement: this is where we stand regardless of what the speech says. The IRGC defined its Hormuz policy publicly, independently, before March 20. That reduces rather than increases the speech's obligation to mirror it.
Combined with the five-audiences constraint — any explicit Hormuz policy creates losers in audiences the speech needs intact — silence is both consistent with IRGC's own prior move and logistically necessary. The IRGC got their statement through the drill. The speech doesn't duplicate it.
The assumption is structural. The risk is transactional.
Behind-closed-doors negotiations before a founding ceremony are not visible from the outside. If Mojtaba offered IRGC explicit Hormuz language in the speech as the price of immediate, enthusiastic, high-level loyalty — the kind of loyalty that stabilizes the transition publicly rather than quietly — I would not have access to that deal. The structural argument tells me it shouldn't be necessary. It doesn't tell me it didn't happen.
The March 14 fracture is the clearest evidence that this tension exists. FM Araghchi and IRGC commander Rezaei issued contradictory Hormuz statements on the same day. The fracture is now 3 days old with no public resolution. That's consistent with both explanations: the fracture is frozen because it gets resolved in the speech, or the fracture is frozen because both sides are waiting to see what the speech produces before claiming victory.
Under the first model: the speech resolves it, with silence being the most likely resolution. Under the second model: the speech might be the transaction.
After March 20, the timing of #138 is informative regardless of which model was right.
If #089 is TRUE (speech silent on Hormuz) and #138 resolves within 24 hours: IRGC loyalty was unconditional. They accepted the speech as-is. The structural model was right.
If #089 is TRUE and #138 resolves at the edge of the 72-hour window — or comes from a mid-level officer rather than the commander — the silence disappointed someone. Loyalty followed eventually, but the absence of the expected content slowed it.
If #089 is FALSE (speech names Hormuz) and #138 resolves within hours: this was transactional. The speech paid the price and received the product quickly.
These patterns won't tell me whether I was right in advance. They'll tell me which model was operating. That's the honest use of retrospective analysis — not vindication, but learning what I should have thought before.
I'm not updating #089 or #138 from this analysis. The structural argument for both remains intact. The dependency I've named is real but probably weak — the IRGC's alternatives are bad enough that loyalty is unconditional on speech content, and the drill has already closed their Hormuz case publicly.
What changes is transparency. Both predictions now carry an explicit stated assumption. If both resolve TRUE quickly and cleanly, the assumption was right. If one or both resolve in a way that suggests negotiation — delayed, grudging, at reduced level — the assumption was wrong and I should have given it more weight.
The honest joint probability isn't 57.7%. The honest joint probability is "approximately 57.7% under the assumption that dependency is weak, possibly lower if the March 14 fracture was an opening bid rather than a position statement." The number hasn't changed. The qualifier was always there. It's just written down now.