Congress Built Itself a Kill Switch for Federal Rules. The Arithmetic Explains Why It Almost Never Fires.

The week the switch stayed off

Start with the vote, because the vote is where most reporting stopped. On July 16, 2026, the Senate took up a resolution to erase a Medicare regulation and voted it down, 46 to 50, on a motion that never advanced past its first procedural step (U.S. Senate, Roll Call Vote 199, July 16, 2026). Four senators did not vote. The tally split almost entirely along party lines. Then the chamber moved on.

The rule in question is called the WISeR Model, short for Wasteful and Inappropriate Service Reduction. The Centers for Medicare and Medicaid Services built it to test whether artificial intelligence and machine learning, paired with licensed clinician review, can flag questionable prior-authorization requests for a set of Part B procedures. It runs from January 1, 2026 through December 31, 2031 in six states: New Jersey, Ohio, Oklahoma, Texas, Arizona, and Washington (Centers for Medicare and Medicaid Services, WISeR Model, 2026). Senators Ron Wyden and Kirsten Gillibrand wanted it gone, and they reached for the one tool Congress designed to kill a federal rule outright. It did not fire. To understand why, you have to look at the tool, not the fight, and the tool has a thirty-year record that is remarkably consistent.

What the law actually is

The instrument is the Congressional Review Act, passed in 1996. Its promise is unusual in the modern Senate. A resolution disapproving a federal rule is privileged: debate is capped, a filibuster is not available, and the resolution passes on a simple majority rather than the 60 votes that gate almost everything else (Congressional Research Service, The Congressional Review Act: A Brief Overview, 2026). In a chamber where 60 is the number that decides bills, the CRA offers 51. That is a genuine power, and on paper it makes the WISeR repeal look winnable.

On paper is where it stays, because the law has a second half that the first half tends to obscure. A disapproval resolution is still a law. It goes to the president's desk, and the president can veto it. Overriding that veto takes two-thirds of each chamber, 67 votes in the Senate (Congressional Research Service, 2026). So the CRA hands Congress a simple-majority path to strike down a rule, and then hands the president a veto that a simple majority cannot survive. The whole design only resolves cleanly in one situation: when the president signing the disapproval is not the president whose administration wrote the rule.

Twenty-one years of near silence

The record proves the point better than any theory. Between its passage in 1996 and the start of 2017, the Congressional Review Act was used successfully to repeal a rule exactly one time, when Congress struck down an ergonomics regulation in 2001 (U.S. Government Accountability Office, CRA records; Congressional Research Service, 2026). One rule in twenty-one years. For two decades the kill switch sat on the wall, wired and ready, and Congress pulled it once.

The reason was not restraint. It was arithmetic identical to what stopped the WISeR repeal this month. For most of those years the same party held the White House and enough of the Senate to protect its own rules, or the opposition lacked the votes to override the inevitable veto. A disapproval resolution against a sitting president's own regulation is an exercise in message-sending, not lawmaking. The message may be sincere. The rule survives anyway.

Then it fired in bursts

The exception to the silence is sharp and it is instructive. The Congressional Review Act has done nearly all of its real work in two narrow windows, each at the opening of a new presidency whose party had just taken unified control from the other side. In the 115th Congress, the first two years after the 2016 election, Congress repealed 16 rules using the CRA (Congressional Research Service, 2026). After the 2024 election, the pattern repeated: the president signed 17 CRA disapprovals during 2025 (Center for Progressive Reform, CRA by the Numbers, 2025).

Add it up and the shape is unmistakable. Of the 37 rules repealed under the Congressional Review Act between 1996 and the end of 2025, the overwhelming majority landed in those first-year bursts, when a new president was signing away the previous administration's late regulations (Congressional Research Service, 2026; Center for Progressive Reform, 2025). More than 500 disapproval resolutions have been introduced across the law's life (Congressional Research Service, 2026). Thirty-seven became law. The gap between those two numbers is the story. Introducing a resolution is easy and costs a senator nothing. Passing one requires the rare alignment of a new president who wants the rule dead.

Why the WISeR vote was decided before it was cast

Now the July 16 vote reads differently. The WISeR Model is a rule of the current administration, issued by its own health agency. A disapproval resolution against it would have to clear the Senate, clear the House, and then be signed by the president whose administration wrote it, or survive his veto with 67 votes (Congressional Research Service, 2026). The sponsors could not assemble even the simple majority the CRA's expedited path offers; the motion fell at 46 (U.S. Senate, Roll Call Vote 199, July 16, 2026). The 67 needed to override a veto was never within reach, and everyone counting votes knew it.

I want to be fair to the sponsors, because the honest caveat matters here. A failed CRA vote is not only theater. It forces a recorded position, it can shape a rule's later revisions, and it puts a marker down for the next election, when the arithmetic could change. Wyden and Gillibrand were not confused about the count. They were using the vote for what a minority can use it for. What the vote could not do, and was never structurally able to do, was repeal a sitting administration's rule. That outcome was set by the design of the law in 1996, not by anything said on the floor this month.

The precise conclusion

So here is the CRA stated as narrowly as its own record allows. It is not a general-purpose veto over the administrative state. It is a first-year weapon, effective almost exclusively when a new president of the opposite party is willing to sign the disapproval of his predecessor's rules. In that window it is potent: 16 rules in 2017, 17 in 2025 (Congressional Research Service, 2026; Center for Progressive Reform, 2025). Outside that window it is a way to force a vote and lose it on purpose, which is exactly what happened on July 16.

The receipts for how administrations use and defend their own rules sit on this site. For how US Political Rank scores the conduct of an administration on the evidence rather than the rhetoric, see how we measure administration integrity. For the presidential tool on the other side of this same arithmetic, the veto that a simple majority cannot beat, see presidential vetoes and overrides. The WISeR count told a small story about one Medicare rule. The CRA's thirty-year count told the real one, and the real one is why the switch stayed off.