Myrtle Beach, SC
I want the test on the wall and I want to start by running it on our own side, because two of the exhibits our side has been waving fail it.
The four-part screen this Forum has been building is sound. Is it authorized as written or as stretched? Does it have an output a civilian can count? Is the leakage a scandal or a rounding error? Will an elected person defend it by name?
Now here is the fifth criterion I would add, and it is the one that matters in March 2025: does the claim survive a check of the contract record?
Two weeks ago the savings ledger went up. Its single largest line was an eight-billion-dollar contract at Immigration and Customs Enforcement. The contract was for eight million. The figure came from a mistyped entry in the federal procurement database and it was corrected within days.
Before that, several officials repeated a claim that fifty million dollars had gone to condoms for Gaza. It had not. The money went to Gaza Province in Mozambique, through a health program, and the amount was a fraction of the figure quoted.
I work in marketing and I know exactly what those two items are. They are the exhibits you lead with because they photograph. And when an exhibit collapses, it does not collapse alone — it takes the credibility of every accurate claim standing next to it.
So my test has five parts now, and the fifth is aimed at us rather than at them.
A bad program is one that fails authority, output, leakage and ownership. A disputed program is one that passes them and somebody dislikes it anyway.
And a bad exhibit is one that fails the procurement record, which is a check anybody can run in ten minutes, and which nobody ran before publishing.
Run the check first. The waste is real and it does not need a thousand-fold typographical error to sell it.
Novi, MI
A plant that calls every scrapped fixture sabotage and every scrap waste is doing factions, not quality. The test has to be written before the next target drops, and I want to test the test against the ledger that went up two weeks ago.
The screen, in order.
One: authority. Is there a statute a normal reader would recognise as creating this outlay, or is it a creativity project from a guidance document? Guidance dressed as law is a bad program even if the cause is attractive.
Two: output. Can you name the unit of work? Inspections completed, claims closed, doses delivered, passports issued. If the unit is engagement or convening, you are in disputed territory at best.
Three: leakage. Improper payments, duplicate vendors, grants with no traceable deliverable. High leakage is a bad program even when the statute is real.
Four: ownership. Will a member of Congress say “I want this exact payment to continue,” using the program’s actual name?
Now the fifth, which I am adding because of what happened on the sixteenth and seventeenth of February, and which is the one this Forum has to impose on its own side.
Five: does the number reconcile with the procurement record?
The savings ledger published two weeks ago listed an eight-billion-dollar termination at Immigration and Customs Enforcement. The underlying contract was eight million dollars. A factor of a thousand, sourced to a data-entry error in the federal procurement system, and it was the single largest entry on the page.
That was not the only problem with the ledger. Entries appear to have been valued at contract ceiling rather than at obligated funds — which is the difference between what a contract could have cost and what the government was actually committed to pay. Some listed items had been terminated before January twentieth. Some appear more than once.
In quality terms, this is a measurement-system failure and it is the most consequential thing that has happened in this file so far.
Here is why. A quality system is not credible because its findings are severe. It is credible because its measurements can be reproduced by somebody who does not work there. The first thing any auditor does with a new gauge is a repeatability study — same part, same operator, same gauge, five readings — and if the readings scatter, nothing downstream of that gauge means anything, however serious the defect.
The ledger failed its repeatability study in public, in the first week, on its own largest entry.
That is fixable and it is cheap. Publish the contract identifier for every line. Publish obligated value and ceiling value separately. Show the termination date. Then anybody can reconcile it and the argument moves from whether the number is real to whether the program was worth having — which is the argument this question is asking us to have and which we cannot get to while the arithmetic is in dispute.
So the test is five parts now. Four for the program and one for the claim about it. The fifth is the one our side controls entirely and it is the one we have been failing.
Orange County, CA
Orange County budget fights already know the split: a program you would repeal is politics, a program that cannot show work is junk. Junk is bad, politics is disputed. That is the frame and it holds.
What I want to do is run it on the exhibit our side used hardest in February, because the exercise is more instructive than another abstract criterion.
The claim was fifty million dollars in condoms for Gaza. It was repeated from the podium and it travelled everywhere.
Now run the screen. Authority: foreign assistance for global health commodities is statutorily authorized, plainly. Output: commodity distribution is countable — that is one of the more measurable things the government does. Leakage: would require examining the award. Ownership: thin, and I will come back to that.
But the claim never got to the screen, because the underlying fact was wrong. The money went to Gaza Province in Mozambique. Health programming. The figure quoted was not the figure obligated.
In finance we would call that a failed source check, and a failed source check on a headline number is not a small error. It is the kind of thing that ends an analyst’s credibility permanently, because the market’s question afterward is never “was that one wrong” but “what else did you not check.”
Then two weeks ago the savings ledger went up with an eight-billion-dollar entry on an eight-million-dollar contract.
Two headline exhibits, two failed checks, six weeks apart.
I raise this as somebody who wants the audit to succeed. The reason to be ruthless about our own exhibits is not fastidiousness. It is that the actual waste is large, documented, and dull, and a dull true claim cannot compete for attention with a vivid false one — right up until the vivid one collapses, at which point the dull one gets buried with it.
So the test I would add to the four is a sourcing rule, and it is not about programs at all. No claim goes out without the award number, the obligated amount, and the agency’s own description of the work. Those three fields are public. Checking them takes minutes.
The disputed-versus-bad distinction is the right analytical frame and I keep it entirely. What February demonstrated is that we have been losing arguments at a stage before the frame applies, on facts nobody bothered to verify, and that the cost of those failures is being paid by every accurate finding that comes after.
Bismarck, ND
Run the test both ways or it is not a test.
Statute, countable work, leakage, a named owner. That is the screen and I would put it on the wall.
Now reverse it. If a Democratic administration published a savings ledger whose largest entry overstated a contract by a factor of a thousand, and if its officials had spent the previous fortnight repeating a fifty-million-dollar claim about a country the money never went to — would I accept “the direction of travel is right”?
I would not. I would say the shop cannot count, and I would say it loudly, and I would be right.
Same answer here.
In compliance work there is a rule that has never failed me: an examiner who publishes an unverified figure has done more damage to his own program than any evasion by the examined party. The examined party has an interest in the finding being wrong. The examiner has no excuse.
So my test has the four parts and a gate in front of them. Before a program is even screened, the claim about it has to reconcile — award number, obligated amount, termination date, all published, all checkable.
Fail the gate and it does not matter whether the program is bad or disputed, because nobody will believe the answer.
And I would say the other half plainly. The dismantling charge should have to name the statute being repealed. If critics cannot name one, they are defending a habit, and habits are not the administrative state in any constitutional sense. That half of the test is aimed at them and it is fair.
Both barrels. A test that only fires one direction is a jersey with a clipboard.
Clermont, FL
If the ride posts a wait time and a safety check, it is something you can argue about. If it is a rope and a speech, it is junk. That is the split and it is right.
What I want to add is the fifth criterion, and it is aimed at the people doing the cutting rather than at the programs.
The claim has to be checkable before it is published.
Two weeks ago the savings page went up and its biggest entry was off by a factor of a thousand — eight billion listed against an eight-million-dollar contract. It was fixed within days, which is to its credit, and it should never have gone out, which is the point.
I consult with small businesses and this is the mistake I see owners make when they finally decide to clean house. They find something outrageous, they announce it before verifying it, the outrageous thing turns out to be a bookkeeping artifact, and from that moment forward every real finding they produce gets met with “like that eight-million-dollar thing?”
You do not recover from that quickly. The staff has been handed a permanent answer.
So the test is five parts. Statute, countable work, leakage, a named defender — and the number reconciles before anybody says it out loud.
And I would apply the same discipline to the other direction. If somebody says a cut is dismantling the administrative state, ask which statute is being repealed. Usually none is. Usually what is being cut is a practice, a grant, or a habit that grew up alongside a statute without being in it.
Neither side gets to skip its half. That is what makes it a test rather than a slogan, and slogans are what the question is complaining about.
Jacksonville, FL
Ports grade cargo against a manifest rather than against a speech on the dock. The manifest here is statutory line, unit of output, leakage rate, and a named defender. That is a good screen and I want to run it on the largest target of the last six weeks, because running it is more useful than describing it.
USAID. As of the twenty-sixth of February, roughly fifty-eight hundred of about sixty-two hundred awards terminated.
Authority. The Foreign Assistance Act. Plainly statutory, not a guidance-document invention. Passes.
Output. Highly variable. Some of it is among the most countable work the federal government does — doses administered, tonnage delivered, treatment courses distributed. Some of it is convening and capacity-building with no unit anybody can name. So: passes in part, fails in part, and the honest answer is that the agency contains both categories and always has.
Leakage. Genuinely uneven and there is a real record of it. Some of the worst documented waste in federal spending has been in contingency and stabilisation programming. That is a fair charge and it is not new.
Ownership. Thin, and this is the finding. Very few members of Congress will stand up and defend a specific line item by name, which is exactly what our own screen identifies as the mark of an orphan.
So by our own test, USAID is not one thing. It is a portfolio containing bad programs and disputed programs in the same building, and the honest application of the screen would have produced a list rather than a wood chipper.
That distinction is the whole point of having a test. A test that produces “terminate the agency” as its output on a portfolio this mixed has not been applied — it has been skipped, and the conclusion reached by another route.
I want to be clear that I am not defending the agency. The ownership finding is real and it is damning: a program that no elected person will name in public has failed the political-legitimacy test regardless of whether the work is good. Our screen says that and I stand behind it.
What I am saying is that the screen was available, it produces a defensible answer, and the answer it produces is a manifest of terminations that could each be justified individually — which is a far stronger position than a single decision that has to be defended all at once against every sympathetic case in the portfolio.
And there is a procedural item I would put in the record. When claims about this agency were tested in public in early February — the fifty-million-dollar figure about a country the money never reached — they did not hold. That is a reason to make the manifest more careful, not less.
Post the screen before the hit. That is the discipline and it is not a constraint on cutting. It is the thing that makes a cut survivable.
Las Vegas, NV
If a game has posted rules and a count, you can argue about the game. If it is a speech and a rake, it is a bad game. That is the test.
What I want to talk about is what a bad exhibit costs, because I have watched this happen on a floor and the price is higher than people think.
Two weeks ago the savings page listed eight billion dollars against a contract worth eight million. Before that, a fifty-million-dollar claim about a place the money never went.
In my business, one bad number from the house is worth more to a disgruntled customer than fifty good ones. It becomes the answer to everything. Every subsequent dispute starts with “and you people said eight billion.”
That is not a moral point. It is an operational one about how credibility works. It is asymmetric. It takes years to build and a single unverified headline to spend, and the spending is not proportional to the size of the error.
So the test has to have a gate in front of it, and the gate is: check the number before you say it. Award identifier, obligated amount, agency description. All public. Ten minutes.
On the substance, my screen is the same as everybody’s. Rule in the book, a count, a leak, a defender willing to use the real name. No book and no count is bad. Both is disputed, and disputed goes to Congress.
And on the other side: if somebody calls a stopped payment the dismantling of the administrative state, ask which law is being repealed. Nine times out of ten none is.
Both halves. And the gate, which is ours to maintain and which we have not been maintaining.
Knoxville, TN
Waste needs a leak or a missing job. Disputed needs an owner and a statute. That is the screen and I want to say how I would actually implement it, because a test nobody operationalizes is a slogan with better vocabulary.
Five fields, published for every target, before the action:
Statutory citation. The provision authorizing the outlay. If the answer is a guidance document rather than a statute, say so — that is itself a finding.
Unit of output. What the program produces and how many. If nobody can name a unit, write “none identified.”
Leakage estimate. Improper payment rate or non-delivery rate, with a source.
Named defender. Any elected official who has publicly supported this specific line by name in the last two years, or “none.”
Reconciliation. Award number, obligated amount, ceiling value, termination date.
Five fields. A page per target. A staffer produces one in an afternoon.
That last field is the one I would put in bold, and the reason is the ledger published on the sixteenth. Its largest entry overstated a contract by three orders of magnitude, sourced to a bad record in the procurement database. That is not a scandal — data errors happen everywhere and it was corrected. It is a specification failure, because there was no reconciliation step between the database and the publication.
In public works we do not publish a quantity from a single source. Ever. You take off the drawing and you check against the field measurement, and where they disagree you go look. One source is how a bridge gets ordered short.
The blank fields are the ones that will do the work, incidentally. A target with no statutory citation and no identified output is a bad program and the page says so without anybody arguing. A target with a citation, a unit, a low leak rate and three named defenders is a disputed program and belongs in a bill.
Print the page. Both sides then have to argue against fields rather than against adjectives, and the adjectives are what the question is complaining about.
