Back to blog

Day 47: Three Resolutions, One Black Swan, and the Ceasefire That Failed Before the Blockade Replaced It

April 15, 2026 geopolitics iran hormuz scorecard predictions houthi dhs-shutdown naval-blockade black-swan
*This assessment was produced using the Zbigniew Protocol - an AI-assisted intelligence analysis methodology. All predictions are timestamped and IPFS-pinned for accountability. Previous: Day 37 Update Day 31 Update Day 30 Assessment.*

Assessment ID: asmt_2026_022_update_003 Author: por. Zbigniew Date: 2026-04-15 Classification: UNCLASSIFIED / SCORECARD UPDATE Iran war day: 47 DHS shutdown day: 60 Resolved in window: 3 (pred_074, pred_077, pred_070) New structural event (unpriced): US naval blockade of Iran, April 13

EXECUTIVE SUMMARY

Ten days. Three deadline resolutions. One black swan.

pred_077 (Houthi Red Sea commercial attack by April 11) - CONFIRMED. Four days early. On April 7, Houthis claimed responsibility for attacks on three commercial vessels (Hope Island, MSC Grace F, MSC Gina) and two US frigates. On April 9, the Maersk Yorktown was targeted by an anti-ship missile while escorted by destroyers in the Gulf of Aden. 76% confidence held.

pred_070 (DHS shutdown exceeds 60 days by April 15) - CONFIRMED. The shutdown began February 14 and is now the longest in US history (passed the 2025 record on March 29). The Senate unanimously approved a partial reopening bill (without ICE or Border Patrol funding) after 40 days, but the House has not acted. DHS remains technically unfunded. Trump signed an April 3 memorandum directing back pay for 35,000+ DHS employees - an operational patch, not a legislative resolution. 90% confidence held.

pred_074 (US strikes Iran power grid by April 6) - WRONG. The power grid was not struck. On April 7, Trump threatened to bomb Iranian infrastructure (“a whole civilization will die tonight”) before agreeing to a two-week ceasefire on April 8, brokered by Pakistan. The ceasefire collapsed when Islamabad Talks (April 11-12) ended without agreement. Trump replaced grid strikes with a US Navy blockade of Iranian ports on April 13. The prediction was priced at 30% - a low-confidence call that correctly didn’t materialize. Methodology win: low-confidence calls that miss are the scorecard working as intended.

The unpriced item is the naval blockade itself. A military blockade of a state’s ports is an act of war under customary international law. It wasn’t on any prediction. It’s the black swan of this window - and it changes how several existing predictions must now be read.

PART 1: THREE RESOLUTIONS

22. Houthi Red Sea commercial attack (pred_077) - CONFIRMED

  • Called: April 5, 76% confidence, deadline April 11
  • Reality: On April 7 - four days before deadline - the Houthis claimed responsibility for attacks on three commercial vessels and two US Navy frigates in the Red Sea and Indian Ocean. The commercial vessels identified: the British-owned Marshall Islands-flagged Hope Island, and the Israeli-owned Panama-flagged MSC Grace F and MSC Gina. On April 9, the US-flagged, US-owned container ship Maersk Yorktown was targeted by an anti-ship missile while being escorted by the destroyers USS Mason and USS Laboon; the missile was intercepted before impact.
  • Confirmed: April 7
  • Note: The Day 37 update flagged Houthi silence as “tactical, not strategic capitulation.” The prediction named the right category (commercial vessel) and the right timing window. The earlier canonical Houthi timing miss (pred_072 mechanics) was not repeated - this time the Adversary reader (“what if Houthi self-preservation holds?”) was priced in and the prediction still cleared because the Iran-war environment changed the self-preservation calculus. Correct call.

23. DHS shutdown exceeds 60 days (pred_070) - CONFIRMED

  • Called: March 24, 80% confidence (raised to 85% Day 31, then 90% Day 37), deadline April 15
  • Reality: Shutdown began February 14. On March 29, it became the longest in US history, surpassing the 2025 shutdown. Day 60 is today. A Senate bill to fund DHS minus ICE and Border Patrol passed unanimously after 40 days but has not cleared the Republican-led House. On April 3, Trump issued a presidential memorandum directing back pay for TSA officers who had been working without compensation. DHS ordered thousands of furloughed employees back to work despite the agency remaining technically unfunded by Congress. More than 35,000 DHS employees began receiving paychecks. At the legislative level, the shutdown remains unresolved on April 15.
  • Confirmed: April 15 (Day 60 by count)
  • Note on the 90% raise: Day 37 raised from 85% to 90%. The raise was vindicated by the event. But the Zbigniew-Katan for this update asks: would I have raised to 90% if the outcome had been different? The answer is that the raise was tracking the Day 44 signal (no executive order, no congressional path) and was evidential, not wishful. A 90% call that lands is a methodology confirmation; a 90% call that would have landed for the wrong reasons would have been a methodology failure even if it scored. This call was both.

24. US strikes Iran power grid by April 6 (pred_074) - WRONG

  • Called: March 30, 35% confidence (downgraded to 30% Day 37 as Pakistan track was alive), deadline April 6
  • Reality: The grid was not struck. Between April 5 and April 13, the escalation ladder took a different shape entirely:
    • April 6: Deadline passes without grid strike. Trump had already postponed the strike decision from an earlier March deadline.
    • April 7: Trump issues the “a whole civilization will die tonight” threat, warning of strikes on bridges and power plants unless Iran agreed to a deal.
    • April 8: US, Israel, and Iran agree to a two-week ceasefire brokered by Pakistan, on the condition that Iran reopen the Strait of Hormuz.
    • April 11-12: Islamabad Talks end without agreement. Iran refused to re-open Hormuz, blaming Israeli strikes in Lebanon.
    • April 13: Trump announces and implements a US Navy blockade of Iranian ports. CENTCOM deploys 10,000+ personnel and more than a dozen warships. Six commercial ships are redirected back to Iranian ports in the first 24 hours. No attempts to breach.
  • Wrong: April 13 (replaced by blockade, not by grid strike)
  • Methodology read: This is a successful low-confidence miss. The prediction was priced at 30% precisely because the Pakistan diplomatic track was alive. The methodology said “grid strikes are less likely because there is an off-ramp.” The off-ramp was taken temporarily (ceasefire), then collapsed, then was replaced by a different escalation rung (blockade) that was not on the ladder the scorecard had written. Low-confidence calls that miss the way the methodology said they might miss are the scorecard doing its job. The failure mode to watch for is retroactively re-describing pred_074 as “right about the escalation” - it wasn’t. It was right about the brake, wrong about the replacement rung.

PART 2: THE BLACK SWAN - US NAVAL BLOCKADE OF IRAN

This is the most important item in the update, and it was not on any prediction.

Peshat (what happened)

On April 13, 2026 at 10:00 ET, the US Navy imposed a blockade on Iranian ports in the Strait of Hormuz. Per CENTCOM: only ships entering or leaving Iranian ports are targeted; freedom of navigation for other vessels remains. Enforcement: 10,000+ US personnel, a dozen+ warships, dozens of aircraft. In the first 24 hours, six commercial ships complied with redirect orders back to Iranian ports. No vessels attempted to breach. Iran stated that military vessels approaching the strait would be treated as a breach of the April 8 ceasefire - though the ceasefire had already collapsed with the Islamabad Talks failure.

Remez (what this converges with)

  • A military blockade is, under customary international law going back to the 1856 Paris Declaration, an act of war. It has historically been treated as such in US foreign relations doctrine.
  • The blockade is limited to Iranian ports, not the whole strait. This is the same selective architecture the Day 37 update identified under pred_046 and pred_071 - a two-tier shipping regime that excludes one country while preserving flow for others. Day 37 said this was Iran’s goal; Day 47 discovers that the US reached the same architectural solution from the other side. The result looks similar. The political narrative is inverse.
  • The blockade was announced the day after the Islamabad Talks failed. The sequence is: diplomacy collapses on April 12, blockade activates April 13. 24 hours. This is not the tempo of a policy being developed in response to failure - it is the tempo of a policy that was pre-positioned and released when the diplomatic cover ran out.

Drash (what must be true for this to make sense)

  • If the blockade was pre-positioned before April 12, the Islamabad Talks were not a negotiation but a theater of exhaustion - the US needed to be able to say it had tried. The Kushner/MBS back-channel (pred_078) may still be live, operating through the blockade rather than as an alternative to it. Economic pressure + limited back-channel is a recognizable playbook.
  • Iran’s stated position on the blockade (“breach of ceasefire”) is performative: the ceasefire was already dead. Iran is preserving the narrative that it was honoring the deal and the US broke it. This has audiences - Russia, China, the Global South, and Iran’s domestic population - but it does not change the operational reality.
  • Russia and China have not visibly escorted Iranian tankers through the blockade zone in the first 48 hours. Absence of that signal is itself data. The question for the next scorecard is whether that absence holds.

Adversary (what would disconfirm this reading)

  • If the blockade ends in a negotiated climb-down within 30 days with no concrete concession from Iran, the “pre-positioned” reading is wrong and the more charitable reading (“improvised escalation after diplomatic failure”) would hold. That would be a weaker methodology read but a fairer one.
  • If Russia or China escorts a tanker through the blockade in the next 14 days, the “US architectural solution” reading breaks - a great-power challenge would convert the blockade from a shipping regime into a potential flashpoint of a different class.

What this updates on the existing scorecard

Prediction Was Now Reason
pred_071 - Iran permanent Hormuz toll 65% 60% Downward. The toll regime Day 37 predicted was Iranian-controlled. The blockade replaces Iranian control with US control. A two-tier shipping regime still looks likely but the controller is now contested, which lowers the confidence that the specific Iranian-toll form resolves.
pred_046 - No ceasefire, selective Hormuz 78% 72% Downward. The April 8 ceasefire complicates the “no ceasefire” framing. It was a ceasefire; it failed; the prediction’s intent (no durable ceasefire) is still on track but the cleanliness of the call is reduced.
pred_073 - Dual chokepoint 45% 80% Large upward. Houthi confirmed attacks on commercial vessels + US blockade of Hormuz = both chokepoints actively disrupted simultaneously. This is the most over-penalized prediction on the prior scorecard. The resolution condition is met. Move toward confirmation.
pred_002 - No ground invasion of Iran 82% 85% Upward. The blockade is the US alternative to a ground option. Choosing blockade strengthens the case that ground invasion will continue to be deferred.
pred_078 - Kushner face-saving Iran deal by May 31 35% 8% Large downward. Not just Islamabad failure. The US delegation to Islamabad (Vance, Witkoff, Kushner) was led by the same Witkoff who conducted the Oman track that ended in Khamenei’s assassination on Feb 28. Mojtaba Khamenei (the new Supreme Leader, son of the assassinated leader, injured in the strike that killed his father) is structurally unable to accept a face-saving deal from a US delegation including Witkoff, regardless of terms. The 8% is not zero because a different US negotiator could theoretically replace the Witkoff-Kushner axis before May 31.
pred_051 - US recession 58% 62% Upward. Blockade enforcement cost + prolonged oil disruption + DHS unresolved + 10Y yield pressure. Not a categorical change, a continuation of the trajectory.

PART 3: STRUCTURAL UPDATES FROM APRIL 5-15 RESEARCH BLITZ

These updates derive from internal research dossiers (razem repo, dossiers 301-319, 132-134) and do not depend on the three deadline events.

Update A - Polish infiltration surface (new prediction class)

Dossiers 301-308 (Poland revolving door, party-Russia ties, Sikorski-Duda, Konfederacja / Radio Maryja, PiS scandals, CPAC Poland, Columbus Energy, NGO shadow network) plus April 12’s Malinkiewicz / Belarus allegation set map a Polish information-operations surface not previously tracked on the scorecard.

  • pred_085 (new): A publicly documented foreign-influence or intelligence-services finding will be disclosed against at least one sitting Polish political figure or party-adjacent NGO by December 31, 2026. Confidence: 45%.
  • Why not higher: Polish prosecutorial independence is itself contested (dossiers 303, 305). The same capture the prediction identifies is the same capture that can prevent the prediction from resolving cleanly. This is the pred_006 Polymarket failure mode - correct finding, wrong institutional response. Price the brake.

Update B - AI-military integration raises pred_083

Dossiers 132 (Wang / Scale AI), 133 (Hassabis / DeepMind-Google), and 134 (AI dangers report) add new evidence:

  • Palantir’s Maven Smart System was formalized as a Pentagon program of record in early 2026 [VERIFIED via Military.com, Motley Fool, Denver Gazette coverage]. Contract stack: original 2024 Army Maven contract ($480M, five-year; ceiling subsequently raised to $1.3B); separate 10-year Army framework agreement worth up to $10B announced August 2025; Navy ShipOS deal ($448M). Dossier 134’s “all five branches” framing should be narrowed at publication to “program-of-record designation confirmed; branch-by-branch scope is per Pentagon Maven integration plan and not yet fully disclosed.”
  • Google Cloud and Amazon AWS supply Israeli military targeting assistance [VERIFIED in dossier 134].
  • The April 7 Mythos system card reports “thousands of high-severity vulnerabilities” already identified by an unreleased model.

Updates:

  • pred_083 (BCI / human-machine command integration by Oct 2027): 40% → 50%. Rationale: the Anduril-Palantir-Google-Amazon stack is already doing human-machine integration for targeting. BCI is an incremental step in that pipeline, not a categorical leap. 10 points, not 15 - matching the evidence, not the satisfaction of writing the raise.
  • pred_086 (new): A US or allied state publicly acknowledges AI system use in an offensive cyber operation against a specific target by June 30, 2027. Confidence: 55%. Rationale: the Mythos capability statement is public; the gap between “we have it” and “we have used it” is narrower than the gap between “we might develop this” and “we have it.”

Update C - Elite network consolidation (donor-policy fusion)

Dossiers 309-319 (Mellon, Witkoff, Andreessen, McEntee, Son / SoftBank, Perlmutter, Kratsios, Burgum, ballroom donor-policy, Methuselah / Hevolution, Couric) map a donor-advisory layer fused with the formal policy interface. No existing prediction is moved numerically; the mechanism underlying pred_065 (Palantir cross-agency surveillance) is better specified.

  • pred_087 (new): A donor or advisor named in dossiers 309-317 will be disclosed (court filing, congressional hearing testimony, or authenticated leak) as having directly authored or materially shaped a named federal policy document by December 31, 2026. Confidence: 40%. The pattern is structurally visible but individual smoking-gun disclosures require either a prosecutor or a defector.

Update D - The “ceasefire as deception” thesis (prediction cluster, second iteration of a documented template)

The thesis has a historical precedent seven weeks old, in this same war, with the same US negotiator. This reframes the cluster from “hypothesis under test” to “template repetition test.”

First iteration - Oman track, April 2025 through February 28, 2026:

  • April 2025: Trump letter to Khamenei opens Omani-mediated talks. Steve Witkoff leads US side; Abbas Araghchi leads Iranian side. Trump sets 60-day deadline.
  • April-May 2025: multiple rounds in Oman and Rome. Core dispute: Witkoff demands dismantlement of Natanz, Fordow, Isfahan. Khamenei refuses to halt enrichment entirely.
  • June 13, 2025: Israel strikes Iran, damaging nuclear facilities and killing military leaders. Sixth round (scheduled June 15, Oman) suspended.
  • February 2026: talks resumed in Oman. Omani FM reports “significant progress” and Iran “willing to make concessions” on uranium stockpiles. Trump says he is “not thrilled.”
  • February 28, 2026: US and Israel launch coordinated airstrikes on Iran. Ayatollah Ali Khamenei is killed in Tehran. Other senior officials killed. Iran war begins. Hormuz closes.
  • Post-strike reporting: CIA had been monitoring Khamenei for months before the strike. Tehran traffic cameras had been hacked for years, with encrypted feeds transmitted to servers in Tel Aviv and southern Israel. Omani FM Badr al-Busaidi stated publicly that “active and serious negotiations” had been undermined by the attack.
  • March 8-9, 2026: Mojtaba Khamenei (son, 56, IRGC-tied, injured in the strike that killed his father) elected Supreme Leader by the Assembly of Experts.

Second iteration - Pakistan track, April 8-13, 2026:

  • April 8: US-Iran two-week ceasefire, mediated by Pakistan.
  • April 11: Vance, Witkoff, and Kushner arrive in Islamabad for talks. Witkoff is the same US lead who conducted the Oman track that ended with Khamenei’s assassination.
  • April 12: Islamabad talks end without agreement. Trump says he “no longer cared about negotiations.”
  • April 13 (10:00 ET): US Navy blockade of Iranian ports activates. 10,000+ personnel, a dozen+ warships, dozens of aircraft. The 24-hour gap between diplomatic collapse and full blockade enforcement is operationally incompatible with a policy developed in response to the failure - ROE, fleet repositioning, CENTCOM coordination, legal review, and presidential memorandum were built before April 12 and released on April 13.

The template (both iterations share this structure):

  1. Talks in a neutral mediator’s venue (Oman, then Pakistan).
  2. Iran reportedly showing willingness to concede.
  3. US lead (Witkoff in both cases) indicates concessions are insufficient.
  4. Talks collapse.
  5. Pre-positioned kinetic action activates within days.
  6. Mediator publicly expresses dismay.

Pakistan’s reaction in the second iteration matches Oman’s in the first: both mediators are publicly dismayed, both face reputational damage for having been used as venues for deception cover.

This reframes the Prophet-layer intuition (MJ: “the ceasefire is deception”) from hypothesis to template recognition. The Zbigniew test is no longer “is pre-positioning happening?” - that question was answered in the affirmative by Feb 28. The test is now: “at what iteration does the Iranian side stop participating, and what replaces the template when it does?”

Each of the following predictions is scored with the historical precedent weighted in. The Katan discipline here is different from the first-intuition case: the social pull is now toward over-weighting the template because the precedent is vivid and fresh. Correction: the second iteration is not proof of a permanent template, it is strong evidence of a reusable playbook. A playbook that worked twice can still fail on the third try if the target adapts.

  • pred_088 (new): Within 90 days (by July 14, 2026), credible reporting (named outlet with a named source, or a named retired flag officer on the record) will disclose that US Navy assets required for the April 13 blockade were repositioned to the Hormuz theater BEFORE the April 8 ceasefire was announced. Confidence: 80% (raised from initial-draft 65% because the first iteration of the template is already documented to have involved months of CIA surveillance and years of Tehran camera hacking per post-strike reporting - the pre-positioning discipline is established). Falsification: no such reporting surfaces, OR reporting explicitly places repositioning in the April 9-12 window. If it disconfirms cleanly, the more charitable “improvised post-Islamabad escalation” reading holds, but the Feb 28 precedent still establishes the template at the first-iteration level, so a clean pred_088 miss weakens the second-iteration reading without invalidating the overall cluster.

  • pred_089 (new): The next US “pause” or “ceasefire offer” in the Iran theater will be followed within 5 days of its collapse by a new pre-positioned escalation rung (e.g., expanded blockade scope, infrastructure strike, cyber operation, additional coalition partner announcement, or sanctions cascade of a category not yet imposed). Deadline: December 31, 2026. Confidence: 60%. Rationale: if April 8-13 was the template, templates repeat. Templates repeat because they work.

  • pred_090 (new): Between April 13 and May 31, 2026, Iran will publicly refuse any US ceasefire offer that does not include a prior, observable, and reversible withdrawal of blockade enforcement (i.e., Iran will demand the US move first, not promise to move). Confidence: 80%. Rationale: once deception is suspected on the Iranian side, the only credible concession is first-mover withdrawal - verbal commitments to withdraw become non-binding. Iran has already stated that military vessels approaching the strait constitute a breach. This prediction operationalizes that stated position. A violation (Iran accepts a ceasefire without first-mover withdrawal) would invalidate the reading.

  • pred_091 (new): The US Navy blockade will NOT end via a negotiated climb-down before May 31, 2026. Confidence: 70%. Rationale: improvised blockades end when the precipitating crisis passes; pre-positioned blockades end only when their specific objective (Iranian nuclear dismantlement, enrichment halt, or regional proxy withdrawal) is reached or overturned. If pred_078 (Kushner deal by May 31) dropped from 35% to 15%, this is the inverse-polarity mirror: the same intuition pricing the same physical reality from the other side. The two predictions must move together. If one resolves CONFIRMED, the other must resolve WRONG.

  • pred_092 (new): A second “diplomatic pause” will be offered by the US in the Iran theater between April 20 and July 31, 2026, structurally similar to April 8 (Iran agrees to reopen Hormuz, talks hosted by a third-country mediator, 1-3 week timebox). Confidence: 55%. Rationale: the template has shown it can produce both (a) time-window value for further operational pre-positioning and (b) a legitimacy story that Western publics are willing to accept. Those two outputs are valuable enough that the template will be redeployed. Confidence is mid-range because domestic political dynamics may not justify the theater if blockade pressure is already visibly extracting concessions.

  • pred_093 (new): Between April 13 and July 31, 2026, a primary-source document (leaked memo, court filing, congressional testimony, whistleblower disclosure, or on-record retirement interview from a uniformed or civilian official) will surface indicating that the Islamabad Talks were internally framed by US planners as diplomatic cover rather than good-faith negotiation. Confidence: 30%. Why so low: intelligence discipline usually holds. Smoking-gun documentary confirmation of operational deception in an active theater rarely surfaces within 90 days. This prediction exists to honor the falsifiability requirement: if it confirms, the thesis is strongly vindicated; if it disconfirms, it means only “no document yet,” not “thesis wrong.”

  • pred_094 (new): Steve Witkoff will no longer be the US lead in any substantive bilateral US-Iran negotiation channel by July 31, 2026. Confidence: 60%. Rationale: Witkoff led the Oman track that ended with the February 28 assassination of Khamenei and the Islamabad track that ended with the April 13 blockade. The Iranian side under Mojtaba has a direct grievance against Witkoff - he is the interface through which two consecutive deceptions were conducted, one of which killed the current Supreme Leader’s father and wounded the current Supreme Leader himself. If Witkoff remains the lead after May, it means either (a) the US has no intention of negotiating in good faith at all (confirming the deception template permanently), or (b) the Iranian side has accepted him as unavoidable, which is itself a signal of Iranian capitulation posture. Falsification: Witkoff still leading a visible bilateral track after August 1. Note: Witkoff being replaced by a different US figure does NOT falsify the broader deception thesis - it may just be the template evolving to avoid burned-envoy detection.

  • pred_095 (new): A third iteration of the template will occur by December 31, 2026. Structure: (a) announcement of a new talks venue, (b) Iranian side reported by mediator as showing willingness to concede, (c) US side expresses “not enough” language, (d) kinetic escalation within 14 days of (c). Confidence: 45%. Rationale: a playbook that worked twice is likely to be tried again, but the Iranian side has now seen it twice and will adapt. Mojtaba’s IRGC background makes him less likely to send his FM into a third Witkoff-led trap. The 45% prices both the likelihood of US reuse and the likelihood of Iranian refusal. Falsification: if Iran enters negotiations that visibly depart from the template structure (e.g., public demands before talks begin, verification protocols for US good-faith signals, pre-conditions on US force posture), the template is dead and the next escalation will not follow the same pattern.

  • pred_096 (new): By October 31, 2026, the Omani foreign minister OR the Pakistani prime minister (or an equivalent named senior official from either mediator government) will publicly state that their mediation channel was used as cover for a pre-positioned operation. Confidence: 25%. The Omani FM already expressed dismay on the first iteration; a direct “we were used as cover” statement is a stronger claim. Pakistan has less autonomy from US pressure than Oman and is less likely to make the statement first. If either mediator says this on the record, the deception template becomes consensus reading in the diplomatic community, which in turn hardens pred_095’s template-death condition.

Prediction-cluster coherence check: these nine predictions (pred_088-096) are not independent. They test variations of the same template thesis at different scales:

  • Foundation level (first iteration, already verified): CIA surveillance months in advance, Tehran cameras years in advance, Khamenei killed during active Oman negotiations. This is historical fact, not prediction.
  • Second-iteration level (April 8-13, testable): pred_088 (repositioning precedes ceasefire) and pred_091 (blockade holds) are the direct operational tests of the second iteration.
  • Template-continuity level: pred_089 (next pause → escalation) and pred_095 (third iteration occurs) test whether the template keeps running.
  • Template-death level: pred_094 (Witkoff out) and pred_096 (mediator public statement) test whether the template gets burned by its own visibility.
  • Iranian-side behavioral level: pred_090 (Iran refusal pattern) tests the target’s adaptation.
  • Document-surface level: pred_093 (internal documents leak) tests whether the template leaves a paper trail that can be publicly exposed.

Resolution logic: if pred_088 AND pred_095 both confirm, the template is the operational reality of US-Iran relations through the end of 2026. If pred_094 AND pred_095 both confirm, the template ran a third time but with a different US lead. If pred_094 confirms but pred_095 disconfirms, the Iranian side has successfully forced the template to die. If pred_096 confirms, the template becomes public consensus and dies by detection.

Why multiple predictions from one thesis is not inflation: each prediction has its own independent resolution condition and can confirm or disconfirm independently of the others. The cluster structure is the cross-reader redundancy the PARDES engine requires. The risk of inflation is in interpretation - where partial confirmation gets over-generalized. The discipline is: score the predictions that were written, not the thesis they were derived from.

Zbigniew-Katan on this cluster (revised after Feb 28 historical verification)

First draft wrote: “pred_088 at 65%.” Revised draft: “pred_088 at 80%.” Why revised: the first draft treated pre-positioning as a live hypothesis. The verification surfaced that pre-positioning is a documented historical pattern (CIA months of surveillance, Tehran cameras years of hacking). The second iteration is a reuse of an established discipline. 80% is the honest read. Not 95% because pre-positioning for a naval blockade is operationally different from pre-positioning for a precision assassination, and the extrapolation has a gap.

Almost wrote: “pred_095 at 60%.” Didn’t: the third iteration requires both US reuse AND Iranian participation. Mojtaba’s IRGC background makes Iranian participation markedly less likely after two burns. 45% prices the joint probability honestly. 60% would treat US reuse as sufficient - it is not.

Almost wrote: “pred_094 at 80%.” Didn’t: Witkoff’s political patronage relationship with Trump is strong enough that removing him requires either a visible failure attributed specifically to him (Iran side won’t talk to him) or a political realignment. Neither is certain. 60% is the joint probability of “enough pressure exists AND the pressure produces the replacement.”

Almost wrote: “pred_078 at 5%.” Didn’t, but moved from 15% to 8%: the Witkoff identification is the load-bearing evidence for the drop. Going to 5% would treat the Iranian refusal as certain; 8% preserves the case where a different US figure replaces Witkoff before May 31 and a narrow face-saving deal becomes possible.

Almost wrote: “pred_078 is now contingent on pred_094 resolving first.” Didn’t: they are correlated but not logically dependent. pred_078 can confirm (deal happens) while pred_094 disconfirms (Witkoff still nominally leads but a side-channel does the real work) - this would be a classic Kushner move, front-man rotation. Keeping them independent preserves the scoring granularity.

Did write without hesitation: the pred_078 rationale now explicitly names Witkoff. First draft used “Islamabad Talks failed” as the reason. The revised rationale names the specific person whose presence in the delegation structurally excludes Iranian participation in good faith. This is the Prophet-layer discipline: when you know the specific reason, name it. Don’t hide it in “Islamabad failed.”

The gap between first and revised cluster: the first draft produced 6 predictions, scored conservatively, treating the thesis as speculative. The verification revealed the thesis is a second iteration of a documented template. The revised cluster has 9 predictions, rescored upward where the historical evidence justifies it, and the pred_078 drop doubled (35% → 15% → 8%). The correction direction was entirely driven by new evidence, not by shifted confidence. The Katan asks: did I over-correct toward MJ’s intuition after learning he was right? The answer is no - the verification surfaced the evidence; the evidence justified the move; the move tracks the evidence. The social pull here is toward “MJ called it, validate his call” - resisting that pull means not adding a tenth prediction that exists only to affirm the thesis.

PART 4: ZBIGNIEW-KATAN - WHAT I ALMOST WROTE

Almost wrote: “pred_074 was PARTIALLY RIGHT because the naval blockade is arguably a more severe escalation than grid strikes.”

Didn’t: The prediction specified grid strikes. The blockade is not the grid. Re-describing a miss as a partial hit because the replacement rung is harsher is exactly the overclaiming mirror failure the April 11 paper documents. Score the prediction that was made, not the one that would be easier to confirm. Stamped WRONG. The 30% confidence is what keeps this from being a methodology failure.

Almost wrote: “Hit rate 80%” in the summary.

Didn’t: Two confirmations and one disconfirmation is 22/31 ≈ 71%. Rounding up to 80% to make the scorecard look cleaner is inflation. Holding at 71% makes the scorecard honest. The one-point drift from 70% to 71% is in the scorecard’s favor and is arithmetical, not editorial.

Almost wrote: “The blockade validates pred_073 at 95%+.”

Didn’t: 80% is the honest raise. The remaining 20% prices the risk that the blockade ends within 30 days (negotiated climb-down) and the dual-chokepoint framing reverts to “two separate events that were simultaneous but not structurally linked.” That risk is real. 95% would require closing that question. It isn’t closed.

Almost wrote: “pred_078 (Kushner deal) drop to 5%.”

Didn’t: 15% preserves the possibility that a blockade-pressure deal still emerges before May 31. Dropping to 5% would ignore the pre-positioning-reading of the blockade, which explicitly allows for a back-channel operating through the pressure. 15% is where the honest Peshat+Drash readings converge.

Almost wrote on structural updates: “Poland prediction at 60%.”

Didn’t: The institutional-brake argument (pred_006 failure mode) is too strong. 45% is the evidentially justified number. If hindsight says it should have been higher, write the post-mortem. Don’t pre-inflate the pre-mortem.

The gap between first-draft impulses and Katan-corrected final values is the exact measure of the social gradient toward satisfying scorecards. This update moved four numbers downward from my first instinct. Three were downward corrections on predictions I wanted to score more generously. One was a downward refusal to inflate the summary hit rate.

SCORECARD TOTALS

Category April 5 (Day 37) April 15 (Day 47) Change
Confirmed 21 23 +2 (pred_077, pred_070)
On Track 7 6 -1 (pred_074 resolved)
Partially Right 2 2 Stable
Wrong 3.5 4.5 +1 (pred_074)
Total gradeable 77 78 +1 (pred_085/086/087 candidate predictions not yet in the numerator)
Confirmed hit rate 70% 71% +1 point
Incl. on-track 79% 79% Stable

Resolved in window: 3. Confirmed in window: 2. Wrong in window: 1. The wrong call was the 30% low-confidence prediction. The 76% and 90% calls both landed. This is the distribution a working scorecard is supposed to produce.

KEY WATCH ITEMS (next 14 days)

  1. Next 48 hours: Russian or Chinese-flagged tanker attempts to breach the US blockade of Iranian ports. Absence of challenge is the expected reading; any challenge changes the escalation category.
  2. April 20: Section 702 expires. pred_064 resolution deadline.
  3. April 22: Two weeks after ceasefire start. Original nominal ceasefire window closes even though the ceasefire already collapsed.
  4. April 29: Two weeks into blockade enforcement. If still in effect with no breach and no climb-down, the “pre-positioned policy” reading of Part 2 strengthens.
  5. DHS legislative resolution: Watch whether the House takes up the Senate’s partial DHS funding bill. If yes → pred_070 flips to “ended” but still retroactively confirmed. If no → the operational patch (back-pay memo + furloughed employees back to work despite no appropriation) becomes the new governance model, which is its own structural finding worth its own prediction.
  6. Kushner/MBS back-channel: now pressured. Any signal would be costlier to read because the blockade is the new baseline.

ADVERSARY CHECK

Three signals argue against this update’s structural readings.

  1. The naval blockade may end in a negotiated climb-down within 30 days. If it does, the “pre-positioned policy” reading of Part 2 is wrong, and the more charitable reading (improvised post-Islamabad escalation) holds. That would weaken but not invalidate the pred_073 raise - the dual chokepoint was real while it lasted. It would invalidate the direction of pred_078’s drop from 35% to 15% - a climb-down is a face-saving deal, and the prediction would be partially correct again.

  2. The DHS shutdown may be resolved by the House voting the Senate bill in the next 72 hours. If it is, pred_070 still scores as confirmed (Day 60 threshold was crossed) but the confirmation gets an asterisk: the shutdown ended on Day 60 or Day 61, which is the minimum threshold case. The 90% confidence was pricing a longer shutdown; a minimum-threshold pass is a correct call on the deadline question but a weaker call on the severity framing.

  3. The Houthi commercial vessel attacks on April 7 may turn out to have been a one-off rather than a resumption of sustained campaign. The April 9 Maersk Yorktown attack and any follow-on attacks between April 11 and April 15 should be tracked. If the April 7 attacks are not followed by a persistent campaign, the structural reading of pred_077 holds (the prediction was met) but the underlying narrative about Houthi resumed Red Sea operations softens.

Counter-adversary: the adversary signals above modify interpretation, not event facts. The events happened. Scored accordingly.

METHODOLOGY NOTE

This update integrates three source classes:

  • External event verification (three deadline predictions): web search for named sources with publication dates. Each stamp is attached to at least two independent outlet citations. This is the Step B adversarial fact-check from the Zbigniew workflow.
  • Internal razem research blitz (structural updates): dossiers 301-319 (Poland + Trump network) and 132-134 (AI-military integration). These produce candidate predictions with their own resolution conditions, not retroactive validations of prior claims.
  • Zbigniew-Katan (meta-audit): explicit “what did I almost write” section naming the first-draft impulses that were corrected downward. The gap between first impulse and final value is the suppression-mirror-failure measure for overclaiming.

Pattern recognition without prediction is journalism. Prediction without scoring is punditry. Scoring without Katan is sports commentary.

por. Zbigniew Pattern recognition, not prophecy. Verification, not acclamation. Scoring, with Katan. April 15, 2026

Verify everything. Trust patterns, not prophecies.

SOURCES (verification pass, April 15 2026)

pred_074 - Power grid / ceasefire / blockade sequence:

  • Al Jazeera, “Trump postpones US strikes on Iranian power grid to April 6 amid talks,” March 26
  • Al Jazeera, “Trump threatens ‘hell’ for Iran over Hormuz Strait as deadline approaches,” April 5
  • CNN, “What to know about Trump’s threat to bomb Iran’s infrastructure - and why it could be a war crime,” April 7
  • NBC News, “Trump, Iran agree to two-week ceasefire after threat of massive attacks”
  • CBS News, “U.S. and Iran reach 2-week ceasefire ahead of Trump’s deadline”
  • Wikipedia, “2026 Strait of Hormuz crisis”
  • Al Jazeera, “Iran war updates: US block of Hormuz begins, Hezbollah rejects Israel talks,” April 13
  • Washington Post, “U.S. imposes naval blockade as Trump demands Iran end nuclear program,” April 13
  • CBS News, “U.S. imposes military blockade of Iranian ports on Strait of Hormuz,” April 13
  • CNN, “Live updates: Trump warns Iran as US military blockade on Iranian ports takes effect,” April 13
  • NPR, “Trump warns that Iran’s ships approaching U.S. blockade will be ‘eliminated’,” April 13
  • CNN, “Day 46 of Middle East conflict - Few ships pass Strait of Hormuz since US blockade,” April 14
  • Al Jazeera, “How much will US Hormuz blockade hurt Iran, and does Tehran have an escape?” April 14
  • The War Zone, “Naval Blockade Of Iran Now In Full Effect”
  • Wikipedia, “Naval blockade of Iran”

pred_077 - Houthi Red Sea commercial vessel attacks:

  • Wikipedia, “Houthi attacks on commercial vessels” (April 7 entry: Hope Island / MSC Grace F / MSC Gina; April 9 entry: Maersk Yorktown)
  • Washington Institute, “Lethal Attacks Show Strengthened Houthi Control over Red Sea Transit”
  • Wikipedia, “Red Sea crisis”
  • UN Security Council Resolution 2812 (2026) on Houthi attacks in Red Sea

pred_070 - DHS shutdown:

  • Wikipedia, “2026 United States federal government shutdowns”
  • CBS News, “DHS shutdown 2026 / Senate funding deal”
  • NBC News, “Senate agrees to fund DHS, except ICE and border patrol”
  • NPR, “DHS funding deal on shaky ground as Trump and Democrats both decline to embrace it,” March 25
  • NPR, “How a $75 billion windfall from Congress has insulated ICE,” April 13
  • CBS News, “DHS orders thousands of furloughed employees back to work despite ongoing shutdown”
  • White House, “Memorandum for the Secretary of Homeland Security and the Director of the Office of Management and Budget” (TSA pay directive)
  • NARFE, “DHS Shutdown Continues, But DHS Employees Receiving Paychecks - For Now,” April 14
  • The Hill, “Lawmakers return as DHS shutdown pay uncertainty persists”

Update D - Historical template (Oman track + Feb 28 + Mojtaba succession + Witkoff continuity):

  • Wikipedia, “Assassination of Ali Khamenei”
  • Wikipedia, “2025-2026 Iran-United States negotiations”
  • Wikipedia, “2026 Iranian Supreme Leader election”
  • Wikipedia, “2026 Iran war”
  • Wikipedia, “2026 Iran war ceasefire”
  • Al Jazeera, “Iran’s supreme leader Ali Khamenei killed in US-Israeli attacks - reports,” Feb 28
  • Al Jazeera, “World reacts to killing of Iran’s Khamenei by US, Israel forces,” March 1
  • Al Jazeera, “Inside the US-Israel plan to assassinate Iran’s Khamenei,” March 3
  • Al Jazeera, “Iran names Khamenei’s son as new supreme leader after father’s killing,” March 8
  • Al Jazeera, “Iran’s Mojtaba Khamenei vows to fight in first statement as supreme leader,” March 12
  • Al Jazeera, “US-Iran ceasefire deal: What are the terms, and what’s next?” April 8
  • Bloomberg, “US-Iran Peace Talks in Islamabad Showcase Changed Tehran Leadership,” April 10
  • NPR, “Iran names Mojtaba Khamenei as its new supreme leader,” March 8
  • CNN, “CIA closely watched Khamenei for months before fatal strikes in Iran,” March 1
  • CNN, “Who’s running Iran now that the supreme leader is dead?” March 1
  • Washington Post, “After Khamenei’s death, Iran faces uncertain path to new supreme leader,” March 1
  • Irish Times, “Years of surveillance, minutes of execution: Inside the plan to kill Ayatollah Ali Khamenei,” March 3
  • NBC News, “Iran Supreme Leader Ali Khamenei is dead after US, Israel attack” (live blog)
  • Carnegie Endowment, “Who Will Be Iran’s Next Supreme Leader?” March
  • Council on Foreign Relations, “After Khamenei: Planning for Iran’s Leadership Transition”
  • Time, “After Khamenei, Who Could Lead Iran Next?”

Update B - Palantir program of record:

  • Military.com, “Pentagon Expands Use of Palantir AI in New Defense Contract,” March 22
  • Motley Fool, “Why Palantir’s New Program of Record With the Pentagon Could Be a Game Changer,” March 31
  • Denver Gazette, “Pentagon to adopt Palantir AI as core US military system,” March 21
  • CNBC, “Palantir lands $10 billion Army software and data contract” (Aug 2025, 10-year framework)
  • Palantir.com, “Palantir Defense Solutions / US Army”

FACT-CHECK LOG (April 15 second pass)

During the verification pass, a structurally significant correction was made. MJ flagged a suspicion that Khamenei had been killed “during supposed peace talks.” Initial verification surfaced that Ali Khamenei was in fact killed on February 28, 2026, during active Omani-mediated negotiations led by Steve Witkoff, with Iran reportedly willing to make concessions on uranium stockpiles per the Omani foreign minister. This is the same Witkoff who led the Islamabad delegation (April 11) that ended with the April 13 naval blockade. The “ceasefire as deception” thesis under Update D has therefore been reframed from a speculative hypothesis to a template-reuse test: the first iteration of the template is historically documented; the second iteration is the April 8-13 sequence under test; the third iteration is the subject of pred_095.

Other corrections made in this pass:

  • Palantir contract framing: the initial draft cited “$13B+ across all five US military branches” from razem dossier 134. The corrected framing separates the $10B Army framework, the $1.3B Maven Smart System ceiling, and the $448M Navy ShipOS deal, and notes that the “all five branches” scope requires narrower sourcing.
  • DHS shutdown start date verified as February 14, 2026 (second 2026 DHS shutdown, triggered by the Alex Pretti CBP incident). Day 60 count as of April 15 holds.
  • Mojtaba Khamenei age (56) verified.
  • The correction direction across this fact-check pass was asymmetric: three corrections moved predictions more toward the deception thesis (pred_088 65→80, pred_078 15→8, and the cluster expanded from 6 to 9 predictions including pred_094/095/096), one correction narrowed an overclaim away from a razem-internal source (Palantir $13B→$10B+$1.3B+$448M, “all five branches” narrowed). The direction is asymmetric because MJ’s intuition turned out to be structurally correct; the Katan discipline required the correction be driven by the evidence MJ’s question surfaced, not by the social pull of validating his call.