Knowability: frame over against window, and the two foundations
The model does not stand or fall as a whole; it has two foundations, and they must be kept separate.
The first is the structural foundation: the whole/peak distinction, the two cohorts, the thlipsis/orgē key, the sixth-seal position. This is strong and redundant (§22), and it is largely independent of the precise date. This is the frame — the form of the end times as the model reads it.
The second is the chronological foundation: the master clock, the week 2027-2034, the midpoint 2031. This is chain-like and externally-dependent. This is the window — the placement of the frame in time.
The frame is defended with a firm hand; the window held with an open hand. There is convergent support for the window — several lines point to the same period (the six thousand years, the generation since 1948, and, as confirmation not as foundation, the solar eclipses and the year-for-day line of the Dome of the Rock — the abomination on the Temple Mount, completed in 72 AH (691/692), plus 1,335 years to Tishri 2025/26 or 2026/27, the threshold of the window, as type of the 1,335 days still to come, §20) — but that convergence must be honestly weighed: some lines are not fully independent, and the eclipse- and Dome-lines carry the least weight — the latter hangs on a pliable year-for-day reckoning (anchor, counting). The net picture: the window has convergent support that raises confidence that it is approximately right; the precise year (2034 over against a near year) rests on the chain, above all on the Masoretic epoch.
Here the model touches the paradox of Matthew 24:36. The solution is not to evade the text but to honour it on its own level: the season has been made knowable — Jesus Himself commands to read the fig tree (Mt 24:32-33), Paul says the day does not overtake the watchful as a thief (1 Th 5:4), Daniel is told that the words are unsealed in the time of the end (Dan 12:4). But season is no specific year. The season rests on the knowability that Scripture itself promises; the year rests on the chronology, and bears its risk. Whoever holds this distinction can present the model with a good conscience: firm about the form, modest about the date.
Falsifiability: what would refute the model
An honest position says what it would die of. The solar eclipse of 2027 is expressly disclaimed as non-founding and non-trigger — a marker that the model cannot fail, it also cannot confirm. But its earlier touchstones the model fixes in advance, in operational, deposited form, and after 2034 it becomes sharply falsifiable: if that autumn passes without the visible return, the chronological keystone falls. There the model sticks out its neck, and that speaks for its intellectual honesty.
The protocol, deposited. This model is testable and its breaking points are fixed beforehand in a separately deposited falsification protocol — Pre-Registered Falsification Protocol for the Pre-Abomination Model (Zenodo, deposited alongside this paper) — which is the binding text wherever any formulation elsewhere, including in this paper, is looser. Its architecture, in brief. The covenant gate: the covenant with the many (Dan 9:27a) is expected to come into force at Trumpets, 1 Tishri 5788, 2 October 2027 — Daniel's dated act is the making strong; preparation and signature may precede it, even by years. If no publicly identifiable covenant has come into force by the end of Tishri 5788 (31 October 2027), the primary configuration is dead, and the cascade is strict: the 1,335 days never begin, every downstream prediction is void with the thesis, and no calendar call will be scored over a fallen model. No event will be retroactively reinterpreted into covenant status; near-misses are misses. A publicly identifiable covenant is a treaty or declaration that meets all three requirements: it is publicly signed or declared in force between 1 and 30 Tishri 5788; Israel is a party, or is bound into it by a large majority of the surrounding states; and it regulates peace, security or worship in Jerusalem, or all three. A resumed daily sacrifice in the months before the covenant may count as a hit; the expectation remains that it is bound up with the covenant. The midpoint condition: if no abomination stands within the Hebrew day 14 Adar 5791 — sunset 8 March to sunset 9 March 2031, Jerusalem — the thesis has fallen, whatever any calendar authority declares in any year; the halting of the offerings is that day's necessary consequence, for the abomination and the loyalty offering cannot coexist on the Mount. If no sacrificial service is running on the Temple Mount at Purim 2031, the midpoint condition is not met. The calendar layer, subordinate, within the week (the springs of 2028 through 2034): the clock intercalates in 2028, 2031 and 2034, not in 2029, 2030, 2032 and 2033; the conventional calendar intercalates in 2030 and 2033. The spring of 2026 fell before the week and does not count as a test; it is named here all the same: the clock intercalated, no observing group did (Karaite: first month from 20 March 2026). With this come referees named in advance in a hierarchy the model itself predicts (a reconstituted temple authority determining the months by observation; failing that, the Karaite determination; failing that, the in-land aviv reports), two kill-conditions (a declared intercalation for spring 2033 fells the model; a declared non-intercalation for spring 2034 fells the terminus dating; if 2028 and 2030 do not give what the clock expects, those are misses for the calendar layer, not a kill), a degradation rule and a fall rule stated in advance, and scoring rules that remain operative in the model's own expected world: kill-conditions fire only on rulings, corroborations are claimed only on rulings, and the model earns nothing from silence. And what no historical outcome can falsify: the arithmetic itself — the day-structure and its corridor property, verified on independently written engines and invariant across all fifty-seven aviv-plausible intercalation patterns — with the shield running one way only: the surviving mechanism earns a fallen application no credit. (The operational form of this protocol — deposited in advance where the author cannot amend it, tolerances that will not be widened, the public record as referee — is owed to R. J. Allinson, whose falsification discipline is exemplary in this genre.)
Gate
What is expected
When the model falls
The covenant gate
the covenant with the many (Dan 9:27a) is to come into force at Trumpets, 1 Tishri 5788, 2 October 2027
no publicly identifiable covenant in force by the end of Tishri 5788 (31 October 2027): the primary configuration is dead, and every downstream prediction is void. A publicly identifiable covenant is a treaty or declaration that meets all three requirements: it is publicly signed or declared in force between 1 and 30 Tishri 5788; Israel is a party, or is bound into it by a large majority of the surrounding states; and it regulates peace, security or worship in Jerusalem, or all three.
The midpoint condition
the abomination within the Hebrew day 14 Adar 5791 (sunset 8 March to sunset 9 March 2031, Jerusalem)
no abomination on that day, or no sacrificial service running on the Temple Mount at Purim 2031: the thesis has fallen, whatever any calendar authority declares
The calendar layer, subordinate
within the week (the springs of 2028 through 2034): the clock intercalates in 2028, 2031 and 2034, not in 2029, 2030, 2032 and 2033; the conventional calendar in 2030 and 2033; the spring of 2026 (before the week) is named, not scored
a declared intercalation for spring 2033 fells the model; a declared non-intercalation for spring 2034 fells the terminus dating; if 2028 and 2030 do not give what the clock expects, those are misses for the calendar layer, not a kill
The arithmetic
the day-structure and its corridor property, invariant across all 57 aviv-plausible intercalation patterns
falsifiable by no historical outcome; the surviving mechanism earns a fallen application no credit
But — and this is the architectural core — if the chronological keystone falls, the structure does not fall. The whole/peak distinction, the two cohorts, the thlipsis/orgē distinction, the ek-reading: these stand independent of the year. A failed 2034 would prove that the dating failed, not that the Bride shares the wrath or that the great tribulation is one whole with its peak. The fan survives what the chain does not survive. That is precisely why the two foundations must remain separate: it is possible that the frame is true and the window shifted — and in that case most of what this paper defends remains standing, while only the placement in time is revised.
Knowability: frame over against window, and the two foundations
The model does not stand or fall as a whole; it has two foundations, and they must be kept separate.
The first is the structural foundation: the whole/peak distinction, the two cohorts, the thlipsis/orgē key, the sixth-seal position. This is strong and redundant (§22), and it is largely independent of the precise date. This is the frame — the form of the end times as the model reads it.
The second is the chronological foundation: the master clock, the week 2027-2034, the midpoint 2031. This is chain-like and externally-dependent. This is the window — the placement of the frame in time.
The frame is defended with a firm hand; the window held with an open hand. There is convergent support for the window — several lines point to the same period (the six thousand years, the generation since 1948, and, as confirmation not as foundation, the solar eclipses and the year-for-day line of the Dome of the Rock — the abomination on the Temple Mount, completed in 72 AH (691/692), plus 1,335 years to Tishri 2025/26 or 2026/27, the threshold of the window, as type of the 1,335 days still to come, §20) — but that convergence must be honestly weighed: some lines are not fully independent, and the eclipse- and Dome-lines carry the least weight — the latter hangs on a pliable year-for-day reckoning (anchor, counting). The net picture: the window has convergent support that raises confidence that it is approximately right; the precise year (2034 over against a near year) rests on the chain, above all on the Masoretic epoch.
Here the model touches the paradox of Matthew 24:36. The solution is not to evade the text but to honour it on its own level: the season has been made knowable — Jesus Himself commands to read the fig tree (Mt 24:32-33), Paul says the day does not overtake the watchful as a thief (1 Th 5:4), Daniel is told that the words are unsealed in the time of the end (Dan 12:4). But season is no specific year. The season rests on the knowability that Scripture itself promises; the year rests on the chronology, and bears its risk. Whoever holds this distinction can present the model with a good conscience: firm about the form, modest about the date.
Falsifiability: what would refute the model
An honest position says what it would die of. The solar eclipse of 2027 is expressly disclaimed as non-founding and non-trigger — a marker that the model cannot fail, it also cannot confirm. But its earlier touchstones the model fixes in advance, in operational, deposited form, and after 2034 it becomes sharply falsifiable: if that autumn passes without the visible return, the chronological keystone falls. There the model sticks out its neck, and that speaks for its intellectual honesty.
The protocol, deposited. This model is testable and its breaking points are fixed beforehand in a separately deposited falsification protocol — Pre-Registered Falsification Protocol for the Pre-Abomination Model (Zenodo, deposited alongside this paper) — which is the binding text wherever any formulation elsewhere, including in this paper, is looser. Its architecture, in brief. The covenant gate: the covenant with the many (Dan 9:27a) is expected to come into force at Trumpets, 1 Tishri 5788, 2 October 2027 — Daniel's dated act is the making strong; preparation and signature may precede it, even by years. If no publicly identifiable covenant has come into force by the end of Tishri 5788 (31 October 2027), the primary configuration is dead, and the cascade is strict: the 1,335 days never begin, every downstream prediction is void with the thesis, and no calendar call will be scored over a fallen model. No event will be retroactively reinterpreted into covenant status; near-misses are misses. A publicly identifiable covenant is a treaty or declaration that meets all three requirements: it is publicly signed or declared in force between 1 and 30 Tishri 5788; Israel is a party, or is bound into it by a large majority of the surrounding states; and it regulates peace, security or worship in Jerusalem, or all three. A resumed daily sacrifice in the months before the covenant may count as a hit; the expectation remains that it is bound up with the covenant. The midpoint condition: if no abomination stands within the Hebrew day 14 Adar 5791 — sunset 8 March to sunset 9 March 2031, Jerusalem — the thesis has fallen, whatever any calendar authority declares in any year; the halting of the offerings is that day's necessary consequence, for the abomination and the loyalty offering cannot coexist on the Mount. If no sacrificial service is running on the Temple Mount at Purim 2031, the midpoint condition is not met. The calendar layer, subordinate, within the week (the springs of 2028 through 2034): the clock intercalates in 2028, 2031 and 2034, not in 2029, 2030, 2032 and 2033; the conventional calendar intercalates in 2030 and 2033. The spring of 2026 fell before the week and does not count as a test; it is named here all the same: the clock intercalated, no observing group did (Karaite: first month from 20 March 2026). With this come referees named in advance in a hierarchy the model itself predicts (a reconstituted temple authority determining the months by observation; failing that, the Karaite determination; failing that, the in-land aviv reports), two kill-conditions (a declared intercalation for spring 2033 fells the model; a declared non-intercalation for spring 2034 fells the terminus dating; if 2028 and 2030 do not give what the clock expects, those are misses for the calendar layer, not a kill), a degradation rule and a fall rule stated in advance, and scoring rules that remain operative in the model's own expected world: kill-conditions fire only on rulings, corroborations are claimed only on rulings, and the model earns nothing from silence. And what no historical outcome can falsify: the arithmetic itself — the day-structure and its corridor property, verified on independently written engines and invariant across all fifty-seven aviv-plausible intercalation patterns — with the shield running one way only: the surviving mechanism earns a fallen application no credit. (The operational form of this protocol — deposited in advance where the author cannot amend it, tolerances that will not be widened, the public record as referee — is owed to R. J. Allinson, whose falsification discipline is exemplary in this genre.)
But — and this is the architectural core — if the chronological keystone falls, the structure does not fall. The whole/peak distinction, the two cohorts, the thlipsis/orgē distinction, the ek-reading: these stand independent of the year. A failed 2034 would prove that the dating failed, not that the Bride shares the wrath or that the great tribulation is one whole with its peak. The fan survives what the chain does not survive. That is precisely why the two foundations must remain separate: it is possible that the frame is true and the window shifted — and in that case most of what this paper defends remains standing, while only the placement in time is revised.