September 11

Twenty-five years on, the September 11 attacks are among the best-documented events in modern history — and still the subject of the most persistent evidentiary argument in American public life. This case exists because both halves of that sentence are true, and because almost everything written about the second half fails at the first: it argues about the record instead of reading it.

So this page begins by conceding what is not in dispute here. Nineteen al-Qaeda hijackers took four airliners; the identities, the travel, the financing, and the training are established by mutually independent evidence streams, and 1,202 exhibits were admitted and tested in an adversarial criminal trial with defence counsel and a judge^C019, nearly all of them published online. This site does not relitigate that, and no claim on this page grades anyone's guilt.

What it maps instead is narrower and more specific: which particular evidentiary disputes survive contact with the primary documents. The answer is that some collapse immediately, some are real and unresolved, and — the finding that surprised this case's own construction — several popular rebuttals fail too.

The building that is actually in dispute

World Trade Center 7, a 47-story tower no aircraft struck, collapsed completely after burning unfought for about seven hours^C001. The water main was severed; the fire department, per the 2002 federal study, took no manual firefighting action. That study's own conclusion is worth remembering before anyone claims the question was ever easy: it called its best hypothesis one with "only a low probability of occurrence" and asked for more work.

The later federal investigation supplied a specific answer. Heat expanded the floor beams until a girder walked off its seat at Column 79, floors cascaded, and the column buckled^C003 — a mechanism requiring no melted steel, with modeled column temperatures near 300 °C. And the report contains something else its critics rarely credit it for: the north face fell at gravitational acceleration for approximately 2.25 seconds^C002, stated in the government's own principal findings, added after public comment pressed the point.

Here is where this case parts company with both camps. The mechanism is graded mixed — not because a better explanation exists, but because nobody outside the agency can check this one. Roughly 68,500 modeling files were withheld under a finding that their release might jeopardize public safety, and a federal court upheld it^C005. The stated rationale is coherent: validated connection models really would be a general-purpose blueprint. The consequence is also real: the most consequential engineering claim in the case is one that only its author can evaluate. Unaudited is not the same as wrong, and this page will not let either side collapse that distinction. Compounding it permanently, no identified WTC 7 steel was preserved^C006 — the agency says so itself, explaining that unlike tower steel, WTC 7's carried no distinguishing markings and could not be identified once removed.

The main published challenge to the official mechanism^C004 is graded weakly supported, and the reasons are symmetrical to the ones above. It released all its data, which is more than the investigation it disputes has done. It was also funded by the organization campaigning for its conclusion, never peer-reviewed in a journal, driven by a single static temperature field rather than a modeled fire progression, and — the objection that does the most damage — critics note that its simultaneous column failure is an input to the simulation rather than a finding from it. Its own text is more careful than its reception: it says it identified no other scenario, not that none exists.

What remains, therefore, is the claim that WTC 7 was deliberately demolished^C010: weakly supported, resting on a genuine unaudited gap rather than on any affirmative artifact. No device, no residue finding on identified steel — there is none to test — and no blast signature on any recording^C009, against a published prediction that the smallest column-severing charge would have registered 130 to 140 decibels half a mile away. That prediction is falsifiable and unmet.

Read the acoustics claim precisely, though, because this is where verification changed our wording. Witnesses did report explosion-like sounds^C008 — and the agency does not deny it. Its FAQ answers a responder who heard two loud booms. What it denies is that any reported sound level matches a column-severing blast. "They said no one heard anything" is not what the record says.

The towers, and an honest gap

The towers are a different matter: that dispute is effectively closed, and closed by positive disproof rather than absence. Impact damage, widely dislodged fireproofing, and multi-floor fires initiated both collapses^C011 — with the report's own hedge, "likely would not have collapsed," restored here because both camps drop it. The claim that pre-planted explosives destroyed the towers^C013 fails on evidence rather than on plausibility: initiation coincided with the impact floors, which nobody could have designated in advance, and the seismic record contains no signal before either collapse.

But the conventional account has one unpaid debt, and this page states it plainly: no end-to-end model of the total collapse progression has ever been published^C012. The agency says propagation was "readily explained without the same complexity of modeling," supporting that with a capacity argument rather than a simulation. That is a defensible engineering judgment. It is not a demonstration, and the difference is exactly the kind this site exists to keep visible.

The dust chemistry deserves its own note, because it is the clearest case of both sides being wrong at once. The claim that the red/gray chips are unreacted nanothermite^C014 is contradicted: the independent analysis found no elemental aluminum, which the claim requires. That same report, however, explicitly found the chips do not match the towers' column primer, with none of 177 candidate products matching — so the standard debunking ("just paint") is also unsupported, and what the material actually is remains unidentified in the published literature. Both analyses sit outside normal peer review; this case applies that objection to both, not to one.

Anomalies that are real, and inferences that are not

The pre-attack put-option volumes were genuinely anomalous^C015 — 12:1 and 6:1 against roughly even expectations — and two peer-reviewed analyses found the pattern statistically consistent with informed trading. That is the observation. Whether it reflected advance knowledge^C016 is the inference, and it is graded weakly supported for an unusual reason: the Commission's explanation is specific and plausible — 95 percent of the airline puts bought by one institutional investor "with no conceivable ties to al Qaeda," much of the rest traced to a newsletter recommendation — and it is impossible to audit. The footnote names neither the investor nor the newsletter, and cites interviews and an internal memorandum that have never been published.

The same discipline sorts the foreknowledge file. The warnings were real, numerous, and unspecific^C018; the August 6 brief warned of "hijackings or other types of attacks" — not, as it is usually quoted, of a hijacking. The congressional inquiry's central finding is routinely quoted in half; both halves appear here, because the first says no intelligence identified the time, place, and nature of the attacks, and the second says the community held information clearly relevant when considered collectively. That officials had specific advance knowledge and let it happen^C019 is contradicted: at every node, the documented failure mode is information not being connected, cataloged in detail by the government's own inspectors general, who found no intentional misconduct where complicity would require it.

Whether a military data-mining program identified the plot's tactical leader a year early^C021 is the one foreknowledge question that stays open, and it cannot be closed: the underlying data was destroyed. The inspector general concluded four witnesses who recalled such a chart were mistaken and identified a real 1999 chart naming 53 terrorists — including a Brooklyn cell, but not Atta. Against that sits the department's decision to bar five witnesses from testifying at a Senate hearing, which the committee chairman said might be obstruction.

Where officials have not earned deference

This case's calibration anchor is not a theory but a proven finding. The agency's assurances that the air near Ground Zero was safe were made without sufficient data, and the White House Council on Environmental Quality convinced it to add reassuring statements and delete cautionary ones^C024 — established by the agency's own inspector general. Health-relevant public statements were centrally edited. That is why generalized official reassurance carries no presumptive weight on this page. It is also why the inference some draw from it fails: the deception was caught and published by the government's own oversight, and its object was air quality, not authorship.

The air-defense record runs the same way. Officials gave accounts the Commission's staff found contradicted by recorded evidence^C020, including a fighter scramble launched after an aircraft that had already crashed — a response "not recounted in a single public timeline or statement issued by FAA or DOD." Documented official misstatement, on the record. And the same tapes are the strongest evidence against the shootdown theories, because fighters chasing a phantom were intercepting nothing.

The one file still moving

Two men connected to Saudi institutions assisted two hijackers in California^C022: documented, in a congressional section withheld until 2016 — which states in its own text that the inquiry made no final determinations about the reliability of the information it reports. Both sentences travel together here; in public argument they almost never do.

Whether a Saudi state institution knowingly supported the plot^C023 is unresolved, and it is the only dispute in this case with an independent forcing mechanism attached. The Commission found no evidence of institutional or senior-official funding — while expressly leaving open, in the same sentence's parenthetical, diversion through charities with significant Saudi government sponsorship. In August 2025 a judge found the civil claims legally sufficient to proceed to trial, deciding jurisdiction rather than merits. The Kingdom appealed the following month and proceedings are stayed, so nothing is currently proceeding anywhere.

What would settle it

Almost all of it is cheap. A blinded re-analysis of one structural connection. Release of the withheld model files, or a bounded re-derivation. A custody-controlled dust re-test with proper controls, which would answer a question both camps have got wrong. The sulfur-source study the FEMA metallurgists asked for in 2002^C007 and nobody performed. Declassification of the trade-tracing files. The data-mining destruction certificates. Not one of these needs a new investigation or a change in anyone's mind — they need somebody to do them, which in twenty-five years no institution on either side has been sufficiently interested to arrange.

That is the honest summary. The conventional account is substantially correct and, at two specific points, not independently checkable. The challenge to it is mostly refuted and, at one specific point, pressing on a real gap. Anyone who tells you the record is closed has not read it; anyone who tells you it is a house of cards has not read it either.