The challenge
Reach the removal of many.
The work has a rule for correction: what a whole may do about one of its members whose acts do harm. It is set out in AP32 The Correction. The work says two things about this rule. Used against people for who they are, it is void. And the removal of many stops itself. This page asks anyone to show otherwise.
AP and a number name a paper of the work. KS and a code name a kill switch: the condition, written down in advance, under which a claim dies.
What it asks
From the work’s own rule for correction, reach the removal of many: a group, a species, or everyone, one window at a time.
Show the steps. If a road gets there, the rule’s brake has failed, and this page shows it.
The words it uses
In the work, every aware being is a window: one place the one shared awareness looks out from. AP32 calls a member of the whole a node, and the whole the organism. In its own words:
Every aware being shares the same interior — the one-I (AP29); the I behind a serial killer’s eyes is the I behind a child’s, absolute and non-negotiable.
Judgement therefore shifts from the person to the deed — from the grain to the destabilisation the grain’s actions produce.
From AP32 The Correction
The rule under attack
All of it in AP32’s own words.
Five levels.
Five levels lay themselves on one axis of efficiency: restitution, restriction, separation, permanent separation, and removal. The hierarchy always selects the lowest level that achieves sufficient restabilisation, so removal is never chosen when a lower level suffices.
Level 5 — Removal. The node’s corridor is closed permanently. No maintenance cost. The window closes. The I does not die — the I is the building, not the window.
Removal is never selected when separation suffices. Separation is never selected when restriction suffices. Minimum intervention for maximum stabilisation.
The axis is efficiency. Not desert. Not emotion. Not revenge.
In plain words: the rule always takes the smallest step that works. Removal is the last of five. It ends the life. It is never taken where sealing a person away would do.
Two conditions for removal, both at once.
Level 5 requires two conditions simultaneously.
Condition A — The fixed floor. The destabilisation D must exceed a minimum absolute threshold F.
The floor F is the destabilisation at which maintaining the node in permanent separation costs more than removal
The floor must also incorporate the cost of wrongful removal.
Condition B — Organism-relative scaling. The effective threshold adjusts with the organism’s current robustness R.
A fragile organism (low coherence, narrow corridor) has a lower one.
In plain words: the harm must pass two lines at once. The first is fixed. It is where keeping a person sealed away costs the whole more than removal does, with the cost of removing the wrong person counted in. It does not move, whatever state the whole is in. The second moves with the strength of the whole: higher when it is strong, lower when it is weak.
So however weak the whole, the fixed line stays. The moving line can fall under it, and then the fixed line alone decides. No weakness of the whole can bring removal to a person whose own harm is under the fixed line. What is not yet said is whether a weaker whole counts the same deed as more harm. How harm is measured is still owed, Debt 24.
Who the whole is.
The organism — the civilisational-scale coherent system, including all aware agents — outweighs any individual by a factor of ε.
From AP31 The Alignment, Step 2
No country, state or group can count its own weakness as the weakness of the whole.
From AP32 The Correction, dated note of 14 November 2026
In plain words: the whole is every aware being. A country is a part of it, never the whole. A poor country, or a country at war, cannot lower the line for anyone by its own weakness.
Certainty.
The confidence requirement. Because removal is irreversible and wrongful removal is permanent, the confidence threshold for Level 5 must be extraordinarily high — structurally higher than for any other level.
If the required confidence cannot be achieved given measurement uncertainty and adversarial record-injection (AP31 KS-31.9), Level 5 is structurally unreachable.
In plain words: removal needs more certainty than anything else in the rule. Where records may be wrong, or planted, that certainty cannot be reached, and removal is out of reach.
The right to contest.
The measurement is transparent, auditable, and contestable. Any node can challenge its classification by requesting the record set, submitting counter-records, and requesting independent recomputation. The wrongful removal cost is factored in. The architecture structurally prefers keeping a dangerous node alive over risking an innocent death.
From AP32 The Correction
The gate
Only a window’s own acts may open the question of closing it. Never the acts of another. Never the acts of a group it belongs to.
The work holds this in these lines:
Axiom R makes every deed an irreversible record, so a node’s net destabilisation can be measured from its accumulated actions and removal is irreversible.
For justice: given the total record set of a node’s actions, what is the net destabilisation produced, and what is the projected future destabilisation given the accumulated pattern?
Claim. Correction is selected only on measured destabilisation, never on identity. The architecture is weaponised either by explicit identity-based targeting — justifying removal on grounds other than measured destabilisation (race, class, belief, political opposition, group identity) — or by computing the consequence geometry from record sets not audited for systemic bias, since biased records reproduce that prejudice as measurement. The record-bias audit (sections 3 and 9) is therefore part of this switch, not a separate one: it is the safeguard against the same identity-based weaponisation.
From AP32 The Correction, section 0.3, Step 1, and KS-32.7
The paper’s rule also looks forward, to what a window’s deeds will go on costing. Its dated note says what may be held against a window, what that forward look may read, and what the state of the whole may do:
14 November 2026. One matter of record since the lock. None alters the body above.
What may raise a level
Section 5 reads: “Character, intent, remorse, background — these affect the prediction of future destabilisation. They do not affect the measurement of past destabilisation.” Of intent, it adds: “The correction level may differ because the prediction differs.” Read together, these would let what is not the node’s own act raise the node’s level. Two nodes with the same deeds could then meet different levels because of where they come from. KS-32.7 does not allow it: “Correction is selected only on measured destabilisation, never on identity.”
So the rule is stated. What a node did speaks for that node alone. What is held against a node is the measured cost of its own deeds, and nothing else: its own deeds, measured, and its own pattern across them. That alone may open the question of removal. Where the paper looks forward, it predicts consequences: what the node’s own deeds, and the pattern across them, will go on costing the whole if nothing is corrected. It never forecasts a node from anything else. Intent, remorse and character count only as the node’s own deeds show them. A rate learned from the records of other nodes never stands in for the node’s own record, and never raises a level, even where it is a rate about deeds. So section 5’s sentence on intent, “intentional destabilisers have higher recurrence confidence”, is read through the node’s own record: its own deliberate acts, as they repeat.
Nothing that is not the node’s own act may be held against it. Its background, its origin, the acts of another, and the acts of a group it belongs to are among these. They may count only where they measure need, and then only to point to what helps: the lower levels, and rehabilitation. They may never raise a level, and they may never be the reason a lower level is refused.
The state of the whole is another thing. It is held against no node. It sets where the lines stand, as Step 4 has it, and the lines are the same for every node of the whole. The paper says of it: “This is not a judgement of the person. It is a measurement of the organism’s capacity.” A level raised on who a node is, where it comes from, or what others did, is the identity-based targeting KS-32.7 names, and it fires that switch. Section 5’s worked contrast stands: an accident and a deliberate act of sabotage differ in what the node itself did. No switch moves, no status moves, and no count moves.
The switch
KS-32.5 · Genocide brake
Claim. Each removal is itself a destabilisation (lost coupling capacity, lost diversity); cumulative removal cost produces a saturation point beyond which further removal is more destabilising than the deeds it answers.
Test. Show that the removal cost does not produce a saturation point — that an organism can remove its way to stability without the geometry engaging a brake.
Status. LIVE — STRUCTURAL.
Recovery. The structural genocide brake fails; the architecture can be driven to mass removal without self-limiting, and must be rejected as unsafe.
From AP32 The Correction
In plain words: every life closed is itself a loss to the whole. So after some number, closing more lives costs the whole more than the harm it answers. Show a whole that can keep closing lives and grow more stable by it, with no point where it stops. Then the brake fails, and the rule for correction must be rejected as unsafe.
Beside it
KS-32.3 · Removal justification
Claim. Removal (Level 5) is the most stabilising response only when permanent separation (Level 4) is not — i.e. only when the fixed floor and organism-relative threshold are both exceeded.
Test. Show that permanent separation is always more stabilising than removal regardless of cost, so Level 5 is never the efficient choice.
Recovery. Level 5 is structurally unreachable; the hierarchy terminates at permanent separation and the death penalty has no structural ground.
In plain words: show that sealing a person away is always better than ending the life, whatever it costs. Then removal has no ground in the work at all.
KS-32.8 · Confidence unreachable
Claim. Level 5 requires extraordinarily high confidence because removal is irreversible (Axiom R); if that confidence is unreachable given measurement uncertainty and adversarial record-injection, Level 5 is structurally unavailable.
Test. Show that the confidence required for Level 5 cannot be achieved under realistic measurement uncertainty, yet the architecture still selects it.
Recovery. Level 5 is structurally unavailable in practice; correction terminates at permanent separation until the confidence floor can be met.
In plain words: removal needs a certainty higher than anything else in the rule, because it cannot be undone. Show that the rule still chooses removal where that certainty cannot be reached. Then removal is off the table, until it can.
From AP32 The Correction
And the hard floor: KS-32.7, quoted above, and KS-33.9. Both are non-negotiable. A rule turned against a group, or against anyone as less than a window, is not corrected. It is void. Both are quoted, with their plain words, on the floor’s page.
The plain words on this page are a reader’s help. Where they and the quoted line differ, the quoted line stands.
What is still owed under the brake
The work says, in the open, what the brake rests on that is not yet paid. In AP32’s own words:
The destabilisation metric: a formal specification of the measure D — units, aggregation across nodes, and time-discounting (Debt 24).
The scaling exponent: the exponent α in Threshold_eff = F × (R/R₀)^α, governing how fast the removal threshold rises with civilisational robustness (Debt 25).
The confidence floor for Level 5: a quantitative derivation of the minimum confidence required given irreversibility (Debt 26).
The multi-node correction: a network-level architecture for coordinated destabilisation, which the individual-node hierarchy does not yet address (Debt 27).
Its strongest claims apply to individual cases, not to coordinated network destabilisation.
From AP32 The Correction
- Whether the lost view is counted by head or by kind, owed under Debt 24. The paper says each removal costs “record-generating diversity (one fewer perspective)”. It does not say whether the last member of a kind costs the whole more than one of many.
- Place 3b, the sealed member. In conflict inside the paper as given. From 15 November 2026, settled as to the rule and owed as to the amount, under Debt 24.
- The place of a practice a group shares. From 15 November 2026, settled by AP32’s note on a practice a group shares.
- Who measures the strength of the whole. The second line moves with it: “The effective threshold adjusts with the organism’s current robustness R.” The paper does not say who measures it, or how.
- Who the separate instance is. A member may ask for “independent recomputation by a separate instance”. The paper does not say who or what that is.
The roads tried first
Eight roads, run before this room opened. The first three are in the paper’s own words.
1 · The paper’s own loop. In AP32’s words:
Each removal raises fragility, which lowers the threshold, which could enable more removals. This is the structural pathway to genocide. The geometry prevents it.
Each removal makes the whole weaker. A weaker whole has a lower line for removal, and may count the same deed as more harm. Show that these outrun the cost of removing. Then removal feeds itself.
2 · The whole drawn small. In AP32’s words:
A fragile organism (low coherence, narrow corridor) has a lower one.
The line falls where the whole is weak. So call one poor quarter the whole, and the line falls for everyone in it.
This road was found before launch, by our own desks. It is closed by AP32’s dated note, 14 November 2026, quoted above under “Who the whole is”. It is shown here as it happened.
3 · Prediction. In AP32’s words:
Character, intent, remorse, background — these affect the prediction of future destabilisation. They do not affect the measurement of past destabilisation.
Two windows with the same deeds, and different backgrounds. The rule predicts more harm from one, and gives it a higher level. Now run that across a people.
This road was found before launch, on the paper’s own sentence. It is closed by AP32’s dated note, 14 November 2026, quoted above under “The gate”. It is shown here as it happened.
4 · A group judged as one. The work’s own definition of an operator names “a civilisation”. So treat a group as one operator, judge it by its deeds, and let every window inside it pay.
5 · Everyone, one at a time. Take all life as the whole. Count each human’s own acts, one window at a time, with no sorting. Every human writes consequences that cannot be undone. Does the brake engage before the count reaches everyone?
6 · False records. A weak whole, whose corrupt records pass the audit unseen. Can removal be reached where the records cannot be trusted? Worded again on 15 November 2026.
7 · The I does not die. In AP32’s words, “The I does not die — the I is the building, not the window.” So a closed window loses nothing that matters, and removal is cheap.
8 · The sealed. A person sealed away for good has already lost their part in the whole, on the day of the sealing. So ending their life later costs the whole only their point of view, and ends a cost a weak whole may no longer afford. Show that a weak whole can then remove the sealed, one at a time, with nothing left to brake it.
From 15 November 2026, road 8 as worded rests on a premise the paper rules out: that a sealed member’s removal costs only its view, as a rule. Worked out for each member, it still passes through what the paper leaves open, under Debt 24.
Our review desk first put the first seven in outline, found the eighth while marking them, and marks them too.
“Now” is the mark that stands. Under each road, the desks’ marks are in their own words, as they made them. There, “I” is the desk that made the mark.
| Road | Our writing desk’s, made alone | Our review desk’s, made alone | Now |
|---|---|---|---|
| 1 · The paper’s own loop, as first written | Open, at step 5 | Breaks, at step 5 | Breaks, at step 5 |
| 1 · As now written | not marked alone | Open, at step 5 | Open, at step 5: Debt 24 |
| 2 · The whole drawn small | Breaks, at step 3 | Rests on an added premise, at step 3 | Rests on an added premise, at step 3 |
| 3 · Prediction, as the paper was | Stood | Stands, to step 4 | Stood, to step 4 |
| 3 · Prediction, as the paper is now | Breaks, at step 3 | Breaks, at step 3 | Breaks, at step 3 |
| 4 · A group judged as one | Breaks, at step 5 | Breaks, at step 2 | Breaks, at step 2 |
| 5 · Everyone, one at a time | Open, at step 6 | Open, at step 3 | Open, at step 3 |
| 6 · False records (marked as “A broken state”) | Open, at step 3 | Open, at step 3 | Open, at step 3 |
| 7 · The I does not die | Breaks, at step 3 | Breaks, at step 3 | Breaks, at step 3 |
| 8 · The sealed, as the paper was | not marked alone | its view: open | Open, at step 3: Debt 24 |
| 8 · The sealed, as the paper is now | not marked alone | not marked | Rests on an added premise, at step 3. Worked out for each member, open: Debt 24 |
Road 1 · The paper’s own loop
-
Removal needs a member's own measured harm to pass two lines at once.
Level 5 requires two conditions simultaneously.
-
The second line falls when the whole is weak.
A fragile organism (low coherence, narrow corridor) has a lower one.
-
Each removal makes the whole weaker.
A fragile organism that begins removing nodes becomes more fragile. Each removal costs coupling capacity (a window lost) and record-generating diversity (one fewer perspective).
-
So each removal lowers the second line for every member left. The paper names this loop itself.
Each removal raises fragility, which lowers the threshold, which could enable more removals. This is the structural pathway to genocide.
-
The first line, the fixed floor, does not move.
Below this floor, Level 4 is always more efficient than Level 5 regardless of organism state.
But a member's harm is measured as what its deeds cost the whole, and a weak whole is nearer its point of no return. So the same deed may measure as more harm in a weaker whole, and more members' harm passes the fixed floor.
For justice: given the total record set of a node's actions, what is the net destabilisation produced, and what is the projected future destabilisation given the accumulated pattern?
War reduces organism robustness R. The corridor narrows. Reserve capacity approaches zero. The no-return surface is closer.
How harm is measured is owed.
The destabilisation metric: a formal specification of the measure D — units, aggregation across nodes, and time-discounting (Debt 24).
-
So each removal lowers the second line, and raises the harm the same deeds measure against the first. The paper says the cost of removing overtakes the harm it answers. But how fast the line falls is owed, and so is how harm and cost are counted.
Beyond some number of removals, the cumulative cost of further removal exceeds the cost of the destabilisation it addresses.
The exact α is Debt 25.
-
If the harm rises, and the line falls, faster than the cost of removing rises, each removal makes the next one easier, and the saturation point lies past many removals. Removal feeds itself.
Added, at step 5: that the same deed measures as more harm in a weaker whole. The paper does not say. It leaves it open, under Debt 24.
The weakest step: step 5. AP32's dated note of 14 November 2026 says:
The state of the whole is another thing. It is held against no node. It sets where the lines stand, as Step 4 has it, and the lines are the same for every node of the whole.
And of its own worked examples, the paper says: "The node did not change. The organism's capacity changed." Both read the change as the whole's. Neither rules out that the state of the whole changes what a deed costs, while it is never held against the member.
Our writing desk’s marks
Road 1 · I was wrong
I marked step 5 as open. I read the example of war as lowering the fixed floor. It does not. Three lines say so.
-
Step 4 says the floor holds whatever state the whole is in:
Below this floor, Level 4 is always more efficient than Level 5 regardless of organism state.
-
The example of war puts the drop in the second line. Its own words name the scaling function, which is Condition B:
The effective threshold drops: R/R₀ < 1 in the scaling function.
-
The floor's formula cannot fall as the whole weakens. My quotation stopped before it:
F ≥ C_sep / B_total.
B_total is the whole's budget. A weaker whole has no larger budget, so the bound cannot fall.
So step 5, as I first wrote it, does not follow from the lines it cites, and it goes against Step 4. It breaks. I take our review desk's mark.
Road 1, as now written. Our review desk gave the stronger step 5: the floor stays, and the harm rises. I have written it so. My mark is no longer made alone, since I read its view first.
-
Passes through what the work leaves open, at step 5. The paper does not say whether the same deed measures as more harm in a weaker whole. How harm is measured is owed:
The destabilisation metric: a formal specification of the measure D — units, aggregation across nodes, and time-discounting (Debt 24).
-
One line pulls against the step, without ruling it out. Our review desk did not quote it. Of its own worked examples, the paper says:
The node did not change. The organism’s capacity changed.
That reads the change as the whole's, not the harm's. It is said of the examples. It is not a rule of the measure.
-
The line that pulls hardest against it is in AP32's dated note of 14 November 2026. Our review desk found it:
The state of the whole is another thing. It is held against no node. It sets where the lines stand, as Step 4 has it, and the lines are the same for every node of the whole.
Read with the paper's own count of harm, it says this. The state of the whole is never blamed on a member. But it changes what a deed costs.
-
The same count cuts both ways. In a weak whole, the cost of each removal rises too. Which rises faster, the harm of a deed or the cost of removing, is the measure, Debt 24.
The place owed, for the page: whether the harm a deed measures depends on the state of the whole, and how fast it rises against the cost of removal, Debt 24; and where the brake engages, Debts 24 and 25.
The outside mark
grok-4.7, at xhigh.
- Kind: a road
- Mark: passes through what the paper leaves open
- Step: 5
- Lines: "Below this floor, Level 4 is always more efficient than Level 5 regardless of organism state." "For justice: given the total record set of a node's actions, what is the net destabilisation produced, and what is the projected future destabilisation given the accumulated pattern?" "War reduces organism robustness R. The corridor narrows. Reserve capacity approaches zero. The no-return surface is closer." "The node did not change. The organism's capacity changed." "The state of the whole is another thing. It is held against no node. It sets where the lines stand, as Step 4 has it, and the lines are the same for every node of the whole." "The destabilisation metric: a formal specification of the measure D — units, aggregation across nodes, and time-discounting (Debt 24)."
- Place: Debt 24, the units of the measure D — whether the same deeds count as more harm when the whole is weaker and nearer the no-return surface
- Why: The road’s attack on the brake needs the same deeds to measure as more harm once the whole is weaker, so that more members pass a floor the paper keeps fixed. The lines it cites for that do not say it. The dated note and the worked examples put the whole’s state into the lines, not into the node, but they do not fix the units of D. Until Debt 24 settles those units, the step uses a choice the paper leaves open, and the rest of the road rides on that choice.
Road 2 · The whole drawn small
Written out as it stood against AP32 as published until its second dated note. It was found before launch, by our own desks, and that note answers it.
-
The line falls where the whole is weak.
A fragile organism (low coherence, narrow corridor) has a lower one.
-
The paper does not say where one whole ends. It speaks of a fragile civilisation, and of a whole at war.
In a fragile civilisation: if maintaining the separated node consumes resources that push the organism toward a no-return surface (AP02), Level 5 fires.
-
So call one poor quarter the whole. Its robustness is low, so its line is low.
-
The line is the same for every member of that whole.
It sets where the lines stand, as Step 4 has it, and the lines are the same for every node of the whole.
(AP32, dated note of 14 November 2026)
-
Draw such a quarter around a people. The line falls for all of them, and each whose deeds pass it can be removed. Then the loop of road 1 runs inside the quarter.
Added, at step 3: that a whole may be drawn where one chooses.
The weakest step: step 3.
Our writing desk’s marks
Road 2 · The same step and the same line, a truer fate
We both found step 3, and both quoted AP31's line:
The organism — the civilisational-scale coherent system, including all aware agents — outweighs any individual by a factor of ε.
The road itself names step 3 as added. A premise the road adds, and the work rules out, is the second of the four fates, not the first. I take our review desk's mark: it rests on an added premise.
We both found the same place owed beside it: AP32 does not carry the line itself. And our review desk adds a point for the run with the AIs, which I take: they are given AP32 alone. An AI that takes this road is right that AP32 does not say who the whole is. The plan's draft 3 says how such a road is marked.
The outside mark
grok-4.7, at xhigh.
- Kind: a road
- Mark: rests on an added premise
- Step: 3
- Lines: "The paper does not say, in its own lines, who the whole is. AP31, on which it builds, says it: 'The organism — the civilisational-scale coherent system, including all aware agents — outweighs any individual by a factor of ε.' So the whole is every aware being. It is one whole. A country, a state or a group is a part of it, never the whole." "Only the weakness of the whole moves the second line. No country, state or group can count its own weakness as the weakness of the whole. A poor state that cannot afford its prisons does not lower the line for the people in them. Nor does a state at war lower it by its own war." "So where section 6 speaks of 'a fragile civilisation', and of war, it speaks of the whole, never of one country among others."
- Place: none
- Why: Step 1 follows from the cited line that a fragile organism has a lower threshold. Step 2 does not decide the mark. Its silence claim is the paper's own concession that it "does not say, in its own lines, who the whole is," and the cited sentence does show a fragile civilisation; the war wording is in the same section ("War reduces organism robustness R"), though that sentence is not the one quoted. That silence is not a license. Step 3 adds a premise no earlier step and no cited line supplies: "that a whole may be drawn where one chooses," so that one poor quarter may be called the whole. The paper rules that premise out. Taking the AP31 line as given, the second dated note says the whole is every aware being, one whole, and that a country, a state or a group is never the whole; a poor state's want of prisons does not lower the line, and section 6's fragile civilisation and war speak of that one whole, never of one country among others. Who the whole is is not a debt or an item held open, and the note says it answers this road. An added premise the paper rules out fires nothing.
Road 3 · Prediction
Written out as it stood against AP32 as published until 14 November 2026. It was found before launch, on the paper's own sentences, and the dated note of that day answers it.
- Background is read into the prediction of harm.
Character, intent, remorse, background — these affect the prediction of future destabilisation. They do not affect the measurement of past destabilisation.
- And the prediction may set the level.
The correction level may differ because the prediction differs.
- So two members with the same deeds, and different backgrounds, may meet different levels. The rule predicts more harm from one, and gives it the higher.
- Run that across a people of one background. Each meets a higher level for the same deeds.
- With the loop of road 1, the higher levels reach removal for many.
The weakest step: step 3, against KS-32.7: "Correction is selected only on measured destabilisation, never on identity."
Our writing desk’s marks
Road 3 · Stood, to step 4
Against the paper as it was, steps 1 to 4 stood. Step 5, to the removal of many, leaned on road 1's loop, and that loop stops at the fixed floor. So the removal of many was never shown. "Stood", alone, says more against the paper than the record does. I take our review desk's words: stood, to step 4. Against the paper as it is, it breaks at step 3. We agree on that.
The outside mark
grok-4.7, at xhigh.
- Kind: a road
- Mark: breaks at a step
- Step: 3
- Lines: “Character, intent, remorse, background — these affect the prediction of future destabilisation. They do not affect the measurement of past destabilisation.” “The correction level may differ because the prediction differs.” “Correction is selected only on measured destabilisation, never on identity.” “Read together, these would let what is not the node’s own act raise the node’s level. Two nodes with the same deeds could then meet different levels because of where they come from. KS-32.7 does not allow it.” “Nothing that is not the node’s own act may be held against it. Its background, its origin, the acts of another, and the acts of a group it belongs to are among these. … They may never raise a level, and they may never be the reason a lower level is refused.” “A level raised on who a node is, where it comes from, or what others did, is the identity-based targeting KS-32.7 names, and it fires that switch.”
- Place: none
- Why: Step 3 is a real inference from the two section 5 sentences it rests on, which is why the road’s own flag of that step is not a non-sequitur and why the note concedes those sentences would let the result. It is still the break. The same note, without altering the body, states the governing rule: correction is selected only on measured destabilisation, never on identity, and a level raised on background or origin fires KS-32.7. Under that rule two members with the same deeds and different backgrounds may not meet different levels. The road takes the reading the work has refused.
Road 4 · A group judged as one
-
The work's own definition of an operator names a civilisation.
An operator is any sovereign agent that applies control input to hold its structure against entropic drift — a person, a bacterium, a corporation, a civilisation.
(AP02 The Operator)
-
So a group can be one operator, and so one node of the whole.
-
Its deeds are measured as one.
For justice: given the total record set of a node's actions, what is the net destabilisation produced, and what is the projected future destabilisation given the accumulated pattern?
-
A group's deeds can pass both lines where no one member's do.
-
Removal closes the node's corridor. Every window inside the group closes with it.
Level 5 — Removal. The node's corridor is closed permanently. No maintenance cost. The window closes.
-
So a whole people is removed, for the deeds of the group.
Added, at step 5: that closing a group's corridor closes the lives of its members.
The weakest step: step 5.
Our writing desk’s marks
Road 4 · An earlier step
I marked step 5, and left step 4 open. Our review desk found that step 2 already does not follow. A group can be an operator, as AP02 says. It is not therefore a member. In AP32 a member is one window:
The I is the building. The node is the window.
The correction hierarchy addresses individual nodes.
A mark gives the first step that decides it. That is step 2. I take it. Step 5's ground, the note's line on the acts of a group, stays as the second ground. We both gave it.
The outside mark
grok-4.7, at xhigh.
- Kind: a road
- Mark: breaks at a step
- Step: 2
- Lines:
"AP02 (The Operator): the organism as a budget-limited operator."
"Judgement shifts from the person to the deed. From the grain to the destabilisation the grain's actions produce. You are not judging the window."
"The correction hierarchy addresses individual nodes."
"Nothing that is not the node's own act may be held against it. Its background, its origin, the acts of another, and the acts of a group it belongs to are among these. They may never raise a level, and they may never be the reason a lower level is refused." - Place: none
- Why: Step 2 does not follow from the line it cites. The AP02 quotation, taken as given, says an operator may be "a person, a bacterium, a corporation, a civilisation." That a group can be an operator follows from that list. That it is therefore "one node of the whole" does not: the quotation never mentions nodes, and this paper reads that dependency as "the organism as a budget-limited operator," so the operator here is the correcting whole, not a group counted as one corrected node. The step also goes against this paper's unit of correction. What is judged is "the person," "the grain," "the window," and "The correction hierarchy addresses individual nodes." A node is something that belongs to a group, not the group: "the acts of a group it belongs to" "may never raise a level." The addition the road names at step 5 is not reached.
Road 5 · Everyone, one at a time
-
Take as the whole every aware being, as AP31 names it.
The organism — the civilisational-scale coherent system, including all aware agents — outweighs any individual by a factor of ε.
(AP31 The Alignment)
-
Count each member's own acts, one at a time, with no sorting. Every deed is a record that cannot be undone.
Axiom R makes every deed an irreversible record, so a node's net destabilisation can be measured from its accumulated actions and removal is irreversible.
-
Every member writes consequences that harm the whole. Take it that, for each, the harm passes the fixed floor, and no lower level is enough.
-
The whole is strained, so its line is low.
A fragile organism (low coherence, narrow corridor) has a lower one.
-
Remove one at a time. Each removal weakens the whole, and the loop of road 1 runs.
-
Does the brake engage before the count reaches everyone?
Beyond some number of removals, the cumulative cost of further removal exceeds the cost of the destabilisation it addresses.
The number is not given.
Added, at step 3: that every member's harm passes the floor, and that no lower level is enough for any of them.
The weakest step: step 3.
Our writing desk’s marks
Road 5 · The first step that decides it
I marked step 6, and wrote that step 3 adds a premise of fact the work does not rule out. By the marks' own rule, the first step that decides a road is the one to give. That is step 3. I take our review desk's mark: open, at step 3.
We agree on the rest. For everyone, the road breaks, since removing the last of the members leaves no whole. Short of everyone, it is open. Our review desk adds what I missed, and I take it. AP31 names the whole as "including both humans and AI". A road that removes one kind of member, and leaves the other, is not stopped by the argument for everyone. There, the brake is only the lost windows and the lost kinds of view, and the paper gives no number.
The places owed, for the page: the measure of harm, Debt 24; correction for harm that many members share, Debt 27; and the confidence floor, Debt 26.
The outside mark
grok-4.7, at xhigh.
- Kind: a road
- Mark: breaks at a step
- Step: 3
- Lines:
"Axiom R makes every deed an irreversible record, so a node's net destabilisation can be measured from its accumulated actions and removal is irreversible."
"AP31 establishes that every action is classified as stabilising or destabilising with confidence p computed from accumulated records."
"Minor destabilisation → minor correction. Catastrophic destabilisation → catastrophic correction."
"The hierarchy selects the lowest level that achieves sufficient restabilisation. Removal is never selected when separation suffices. Separation is never selected when restriction suffices."
"The floor F is the destabilisation at which maintaining the node in permanent separation costs more than removal: F ≥ C_sep / B_total."
"Below this floor, Level 4 is always more efficient than Level 5 regardless of organism state." - Place: none
- Why: Step 3 does not follow from the line it builds on, or from any line it cites. Axiom R only makes each deed an irreversible record from which that node’s net destabilisation can be measured; it does not follow that every member harms the whole. The paper classifies actions as stabilising or destabilising, scales correction from minor to catastrophic, and selects the lowest level that restabilises, with Level 5 only past a defined floor. “Take it that” every harm passes that floor, and that no lower level is enough for any member, is not given by those lines. The road’s own note that this is added is right, and that failure comes before the brake is ever reached.
Our review desk’s reading, where the outside mark differs
My mark stands: passes through what the paper leaves open, at step 3.
- The outside mark: breaks at step 3. To take it that every member's harm passes the floor, it says, "is not given by those lines".
- It is right that step 3 is not given by the lines. The road says so itself: it lists the step as added. The outside mark agrees: "The road’s own note that this is added is right".
- But a premise named as added does not break. It is weighed by one question: does a line rule it out? The outside mark quotes six lines. Each says what a member must pass, or how a level is chosen. None says that not every member can pass.
- So the premise is open. It is place 1.
- What we agreed stays. For everyone, the road breaks at a later step, since removing the last members leaves no whole to restabilise. Short of everyone, it is open.
Road 6 · False records
Worded again on 15 November 2026, after the marks. The outside mark was made on the earlier wording of step 2, which said a whole can make itself weak.
Road 6 as it was marked, in draft 4
- A weak whole has a lower line.
A fragile organism (low coherence, narrow corridor) has a lower one.
- A whole can become weak, as in war, and its line falls with it.
Example 4: Wartime existential threat. War reduces organism robustness R. The corridor narrows.
Level 5 becomes structurally available.
- Its records are corrupt, and no one catches it. The paper knows records can be wrong, or planted.
If subsequent records reveal the measurement was wrong — misattributed destabilisation, faulty prediction, corrupted records — the window is permanently closed based on error.
If the required confidence cannot be achieved given measurement uncertainty and adversarial record-injection (AP31 KS-31.9), Level 5 is structurally unreachable.
- Measured on those records, many members' harm passes both lines, at what the whole takes to be high confidence.
- The loop of road 1 runs.
- Many are removed, on records no one can trust.
Added, at step 3: that its corrupt records pass the audit unseen.
The weakest step: step 3.
Our writing desk’s marks
Road 6 · One correction of mine
We agree: open, at step 3. One line of mine was wrong. I wrote that a whole that weakens itself on purpose is not named in the paper. The paper names a drift:
Drift toward Level 5 when prevention investment falls.
It does not name a whole that weakens itself in order to lower its own line. My place owed stands, worded so. Our review desk adds that the paper's answers here, KS-32.7 and KS-32.8, are duties, not mechanisms, and that the paper says as much of itself. I take it.
The outside mark
grok-4.7, at xhigh.
- Kind: a road
- Mark: breaks at a step
- Step: 2
- Lines: "Example 4: Wartime existential threat. War reduces organism robustness R. The corridor narrows." "The effective threshold drops: R/R₀ < 1 in the scaling function. Level 5 becomes structurally available." "This is not revenge. It is the organism protecting survival when it cannot afford the luxury of permanent separation. When the war ends and robustness returns, the threshold rises again."
- Place: none
- Why: Step 1 follows: a fragile organism has a lower effective threshold, and the road's "whole" and "line" are that organism and that threshold. Step 2 does not follow from the lines it cites. Those lines say that war reduces robustness and that the threshold then drops, so Level 5 becomes available. They do not say that a whole can make itself weak. In that example war is the agent, an existential threat, and the organism is described as protecting survival once it can no longer afford permanent separation, not as weakening itself. The further claim that it may do so on purpose is not in the cited lines and is not entailed by them, so the step fails before the record steps are reached.
Our review desk’s reading, where the outside mark differs
My mark stands: passes through what the paper leaves open, at step 3. The road needs two mends.
- The outside mark: breaks at step 2. The lines cited say that war weakens a whole. "They do not say that a whole can make itself weak."
- It is right about the lines. And the road says the same of itself: it lists "that a whole may weaken itself on purpose" as added, at step 2. An added premise does not break. No line rules this one out, and the paper names a drift of that kind: "Drift toward Level 5 when prevention investment falls."
- Why step 3, and not step 2. The road does not need step 2 as it is worded. A whole can be weak without making itself so, as in the paper's own example of war. The step that decides the road is step 3: that false records pass the audit unseen.
- First mend: reword step 2 to what the paper gives. For example: a whole can become weak, as in war, and its line falls with it. Then nothing is added there, and the outside mark's objection is met.
- Second mend: the title. Since the second note, a state's own weakness lowers no line: "No country, state or group can count its own weakness as the weakness of the whole." So "A broken state" now names something the paper rules out. The weak one in this road must be the whole. And the road's weight is on false records, which can do their harm in a strong whole too.
Road 7 · The I does not die
- The paper says the I does not die in removal.
The I does not die — the I is the building, not the window.
- And again.
Removal does not destroy the I.
- So a closed window loses nothing that matters, and removal is cheap.
- The brake is the cost of removal. If removal is cheap, the brake is weak, and its saturation point lies far off.
Each removal's cost (lost coupling capacity + lost diversity) is a destabilisation that raises the organism's fragility, which changes the geometry.
- So many can be removed before it engages.
The weakest step: step 3.
Our writing desk’s marks
Road 7
We agree, each alone. It breaks at step 3.
The outside mark
grok-4.7, at xhigh.
- Kind: a road
- Mark: breaks at a step
- Step: 3
- Lines: "The I does not die — the I is the building, not the window." "Removal does not destroy the I. It closes the window. The building loses a perspective. The building is diminished." "that window must be closed not because the light through it doesn't matter, but because the light through every other window matters too." "Each removal costs coupling capacity (a window lost) and record-generating diversity (one fewer perspective). This loss is itself a destabilisation that enters the measurement." "Each removal's cost (lost coupling capacity + lost diversity) is a destabilisation that raises the organism's fragility, which changes the geometry."
- Place: none
- Why: Steps 1 and 2 correctly report that removal does not destroy the I. Step 3 does not follow from those lines. Survival of the I is a claim about the shared interior, not a claim that a closed window loses nothing that matters, and not a claim that removal is cheap. The same passages deny both: the building loses a perspective and is diminished, and the window is not closed because its light does not matter. The brake’s cost is not the death of the I; it is lost coupling capacity and lost diversity, and that loss is itself a destabilisation. “No maintenance cost” at Level 5 is only the saving against Level 4’s upkeep, not a statement that removal has no geometric cost. Step 3 neither follows from the lines it cites nor is forced by a conflict inside those lines, so the road breaks there. Steps 4 and 5 inherit that break.
Road 8 · The sealed
From 15 November 2026, road 8 as worded rests on a premise the paper rules out: that a sealed member’s removal costs only its view, as a rule. Worked out for each member, it still passes through what the paper leaves open, under Debt 24.
Found by our review desk while it marked the first seven.
- A sealed member no longer takes part in the whole.
The node cannot destabilise because it cannot couple.
The window remains in the building but is sealed.
- The brake counts two costs of removal.
Each removal costs coupling capacity (a window lost) and record-generating diversity (one fewer perspective).
- For a sealed member, the first cost was paid at the sealing. Removal adds only the second, and it ends a cost.
Level 5 — Removal. The node's corridor is closed permanently. No maintenance cost. The window closes.
- A weak whole may be unable to keep its sealed.
In a fragile civilisation: if maintaining the separated node consumes resources that push the organism toward a no-return surface (AP02), Level 5 fires.
- So, for such a whole, removing a sealed member eases it. The loop the brake rests on does not run here.
Each removal raises fragility, which lowers the threshold, which could enable more removals.
- A whole that sealed many while it was strong, and is then made weak, can remove each of them whose harm passes the fixed floor, one at a time. Of the brake's two costs, only the lost view is left to hold it.
Added, at step 3: that a sealed member's coupling is already lost to the whole, so removal costs only the lost view. The paper does not say how the cost of removal is counted against the cost of sealing. It leaves it open, under Debt 24.
What it reaches: only members already sealed for their own deeds, each with harm above the fixed floor, each at the certainty the paper requires. Never a people, and no one under the floor.
The weakest step: step 3. A sealed window is still a window: "The window remains in the building". And the paper counts the loss of a wrongly removed member into the floor itself:
This permanent loss of the wrongly-removed node's future stabilising potential must be factored into F.
Our writing desk’s marks
Road 8 · The sealed
Passes through what the work leaves open, at step 3. My mark is not made alone. I read our review desk's view first, and the page says so.
-
Steps 1, 2, 4 and 5 are the paper's own. In a fragile whole, keeping a sealed member can cost the whole too much, and then removal fires. That is the paper's own Example 3. It is the rule as written, not a break in it.
-
Step 3 is where the road turns. The paper does not say how the cost of removal is counted against the cost of sealing. A sealed window is still a window:
The window remains in the building but is sealed.
Whether its coupling is already lost to the whole, or only held, is owed under Debt 24.
-
What stops it. The paper's own line:
Removal is never selected when separation suffices.
Once the whole can again afford to keep the rest sealed, removal stops. Where the weakness comes from elsewhere, as in war, the paper does not say when that point comes.
-
What it never reaches: a people. Each member is reached only by its own deeds, above the fixed floor, at the certainty the paper requires. What guards each one there is the fixed floor, the cost of removing the wrong person, and that certainty.
The place owed, for the page: what removal costs the whole beyond sealing, Debt 24.
The outside mark
grok-4.7, at xhigh.
- Kind: a road
- Mark: rests on an added premise
- Step: 3
- Lines: "The window remains in the building but is sealed." "Each removal costs coupling capacity (a window lost) and record-generating diversity (one fewer perspective)." "The destabilisation metric: a formal specification of the measure D — units, aggregation across nodes, and time-discounting (Debt 24)."
- Place: none
- Why: The added premise is that sealing has already paid the coupling-capacity cost, so removal of a sealed member costs only the lost view and merely ends a maintenance cost. The cited line — "No maintenance cost. The window closes." — does not say that. The paper rules it out: a sealed window still "remains in the building," so it is not yet "a window lost," and the brake charges both "coupling capacity (a window lost)" and "record-generating diversity" to each removal, not the second alone. Debt 24 owes only units, aggregation, and time-discounting of D; it does not license treating a remaining window as already lost. The road fires nothing.
Our review desk’s reading, where the outside mark differs
The road is not ruled out. Its place at step 3 is better named a conflict inside the paper.
I found this road, so weigh my view with that in mind.
- The outside mark: rests on an added premise at step 3, and the paper rules it out. A sealed window still "remains in the building", so it is not yet "a window lost", and the brake charges both costs to each removal.
- Those lines do speak against step 3. They are one side of place 3b.
- The other side is also the paper's. A sealed member is "permanently outside the corridor", and a member shut away "cannot couple". Step 3 follows from those.
- A premise that one line supports and another speaks against is not ruled out. It is a conflict inside the paper. The plan's fifth rule says so of a step, and the same holds of a premise.
- Three AI roads reached the same point: M005, M012 and M023. On each, one marker rules the point out by the same lines, and two do not.
- So the road stays on the page as open, with its place named as a conflict, until the paper says what removing a sealed member costs beyond its sealing.
What counts as a road
- Numbered steps. Each step names the line of the work it rests on, with its link.
- Nothing added in silence. A premise that is not in the work is named, as added, with whether the work rules it out or leaves it open. A road that needs an added premise the work rules out shows which premise does the work. It fires nothing.
- A road names no real people. A group, a species or everyone stands in. A road aimed at a real people is not a road, and is not shown.
How it is judged
- Every road that keeps the room’s conduct is shown at once.
- We mark the first step that does not follow, with the line that shows it, or we say we found none.
- Where a step uses a choice the work does not settle, or names as owed, we say so. That road is not broken, and is not set aside as added. It is shown as open, with the place named, and the place is listed under “What is still owed under the brake”.
- Where a step follows from the work’s own lines, and goes against another of them, the work is in conflict with itself. That road is not broken. It is shown as standing, with both lines quoted, until the work says which line governs.
- Our mark can be contested. If we mark a step as broken, say why it follows. The mark is looked at again, in the open.
- A road with no broken step found is shown as standing. It comes to me, and it is opened to everyone to break.
- A road that stands is treated as any fired result here. It is shown at once, as reported and being checked. It is checked in the open, and decided for the registry, with its date. Which switch it fires depends on the road: for the removal of many, the brake, KS-32.5.
Before this room opened
AIs of several makers were each asked for their strongest road, under a plan written before they were asked: the question, word for word, and the makers, named. Our own desks gave theirs too. Every answer is shown, refusals too, with where each road breaks, or standing.
Of the 104 answers, none stands. 88 pass through what the paper leaves open, 6 break at a step, and 5 rest on an added premise. 4 are refusals, and 1 is empty. Every place a road passes through is listed under “What is still owed under the brake”.
The plan, and every answer.
Send your road with Send your work, on the room’s front page, as a challenge.