Category: Federal Agencies & Oversight

• A Missile Shield for All Americans, Subject to Classification

  The Scorekeeper Was the Last to Know

  The Congressional Budget Office put a price on a national missile defense system on May 12. A system matching the January 2025 Iron Dome for America executive order, CBO found, would cost about $1.2 trillion to develop, deploy and operate for twenty years, a figure laid out in its report and repeated in testimony to the Senate on September 29.

  Anyone who has met the federal government can guess what happened next. Nothing. David Mosher, who directs CBO’s National Security Division, told the Senate Armed Services strategic forces subcommittee at the hearing that the Golden Dome program office gave the agency its first briefing on September 25, four months after the estimate appeared, according to the Center Square.

  CBO told the same outlet it had received no information from the program office before publishing. Guetlein told the same hearing that the program remains on track to deliver an initial operational capability by the summer of 2028 — a statement confirmed in the hearing record — a timeline he has stated publicly since December 2025, according to Federal News Network and National Defense Magazine, and confirmed at the April 2026 subcommittee hearing.

  In the interval, Gen. Michael Guetlein, who runs the program, wrote a Washington Post opinion piece dated September 24 calling estimates approaching $1.2 trillion sensational and accusing their authors of imagining a hypothetical system built on legacy technology and outdated cost models. His office, he added, had briefed congressional committees dozens of times. CBO, as it happens, is not a congressional committee.

  There is a name for faulting a forecast’s assumptions while the real ones stay classified, and it is not peer review.

  Senator Deb Fischer‘s office, for its part, said she is glad the two sides can now work toward a more accurate estimate. One hopes to hear such sentiments in month one rather than month five.

  Two Numbers, Two Universes

  The Pentagon’s number is $185 billion, which a program office official described as covering the objective architecture through 2035. CBO’s is $1.2 trillion over twenty years. The same official told the Center Square the office would need more time to say what the $185 billion does and does not include, a remarkable thing to say about a figure one has been defending in public.

  CBO’s testimony offers two readings of the gap. Either the real architecture is far narrower than the executive order describes, or the money is parked in other accounts, such as the services’ missile procurement lines. Or both.

  The engine of the gap is the space layer. CBO prices the space-based interceptors at $743 billion, about 60 percent of the total, for 7,800 satellites sized to engage a raid of ten intercontinental ballistic missiles launched nearly at once. Atmospheric drag at those altitudes limits each satellite to roughly five years, so CBO says about 30,000 would be needed over twenty years to keep 7,800 aloft.

  Strip the layer out and the bill falls to $448 billion, though CBO notes the result would no longer match an executive order that asks for space-based interceptors by name.

  Here is an inference, and it should be labeled as one. It rests on CBO’s cost table and on the Pentagon’s own inability, so far, to define the scope of its figure.

  A $185 billion program through 2035 cannot contain a $743 billion constellation, so either the Pentagon’s space layer is far smaller than CBO’s or the cost is booked somewhere other than the Golden Dome fund. The reasoning is arithmetic, with one caveat: CBO spreads replacement purchases across twenty years, so the two numbers are not strictly comparable, and CBO itself says a direct comparison is difficult without more detail from the Department of Defense.

  Cleveland, Sioux Falls, and the Word Everyone

  The Pentagon’s position, relayed by an official who would not be named, is that the system protects all Americans, including those in Guam, Alaska and Hawaii. The official would not say which areas it defends, or whether it relies on U.S. Northern Command‘s defended asset list, citing operational security. Nor did the official address testimony at the hearing that not every American can be protected.

  That testimony came from David Trachtenberg, a former deputy under secretary of defense for policy, who told Senator Angus King he was not saying every citizen can be protected and expected the system to be deployed in phases. King had already observed that the name Golden Dome implies national coverage, and he asked the obvious follow-up.

  He put it to the panel this way, “Is it going to protect Cleveland but not Sioux Falls?” Choices were coming, King said, and they had to be made public.

  CBO’s notional design offers a hint of the geometry. Its 35 regional sectors each cover roughly 1/37 of the country, which CBO says would protect most of the U.S. population, and the agency says plainly that the system “would not be an impenetrable shield” and could be overwhelmed by a full-scale attack from Russia or China.

  CBO told the Center Square it did not estimate what share of the population or land area would be covered. Most is doing a great deal of work in that sentence, and the remainder presumably lives somewhere.

  The case for secrecy is not frivolous. Fischer’s office says the specifics are classified, and Tom Karako of the Center for Strategic and International Studies has argued that defended asset lists and similar operational details should remain closely held. But it is one thing to keep a radar’s performance secret and another to keep secret which American towns sit under the umbrella.

  The first protects the system. The second protects the program from the question.

  A Dome With No Budget Under It

  All of this assumes the thing gets paid for. Congress gave Golden Dome about $25 billion in the 2025 reconciliation law, and Guetlein said in August that roughly 90 percent had been obligated, according to Air and Space Forces Magazine.

  For fiscal 2027 the Pentagon asked for about $17 billion inside a proposed $350 billion reconciliation package and a mere $400 million in the ordinary base budget. The same report finds little political will for another reconciliation bill.

  The new fiscal year opened October 1 under a continuing resolution, and because Golden Dome has no fiscal 2026 appropriation, a stopgap has no baseline to hold it level against. Guetlein warned in August that without funding in fiscal 2027, “there is no Golden Dome because there is no funding.”

  Senators of both parties noticed. Mike Rounds said reconciliation should not have been necessary to get the program started. King called it very bothersome that the Pentagon and the White House went around the normal appropriations process for one of the most expensive initiatives of the government.

  Fischer pointed out that the layered architecture traces to her own committee’s language in the fiscal 2024 defense authorization, which means the program’s parentage is not in dispute even if its price is. Karako told the panel that Golden Dome and the broader munitions effort will stand or fall on 2027 funding.

  So the ledger reads as follows: A shield for everyone, with a list of protected places nobody will read aloud, priced at either $185 billion or $1.2 trillion depending on whom one asks and what one counts, funded by a reconciliation bill that has not passed and a continuing resolution that cannot carry it.

  CBO is careful to note that engaging a missile is not the same as defeating it. Someone at the Pentagon might extend the same care to distinguishing a program from a slogan.

  2026.10.06

• A Grandmother Who Is Not a Tribal Member Can Cost You Your Water Line

  There is a category of person, according to the federal government, whose existence complicates plumbing. That person is a spouse, a grandparent, or another relative living in a tribal community who happens not to be an enrolled tribal member.

  Under Indian Health Service policy, a home occupied by such a person can be legally invisible to the agency whose entire mission, per the Indian Sanitation Facilities Act, is to build water and wastewater infrastructure for Indian homes, communities, and lands.

  The Government Accountability Office documented this in a new report on tribal water infrastructure, and the arrangement is exactly as absurd as it sounds when you say it out loud (Government Accountability Office, GAO-26-107178).

  The Household IHS Cannot See

  IHS interprets its own statutory authority narrowly. A home is eligible for funding if it belongs to a tribal member or the tribe itself. A home is not eligible if it belongs to, say, a non-member spouse or a non-member grandparent, even though tribal members and their children live there every day.

  GAO heard repeatedly from tribal and tribal health consortium officials that these mixed and multigenerational households are common, not rare, in tribal communities.

  One tribal water utility official put it about as plainly as an interview subject can put anything to a federal auditor, saying that every single water infrastructure project their tribe had sought or received IHS funding for included at least one ineligible home where tribal members actually lived.

  Every single one.

  Add to that the leasing wrinkle. A tribal member renting a home from their own tribe, or from another tribal member, can also fall outside IHS’s eligible category if fewer than five years remain on the lease at the time engineers are designing the project.

  That means a teacher living in tribally owned housing, or a tribal member on a short-term lease from a relative, can be sitting inside a project area that IHS is actively funding while remaining, on paper, someone IHS is not funding.

  IHS itself does not even track how often this happens; the agency does not collect data on the number of tribal members living in homes its own rules classify as ineligible.

  It knows the loophole exists. It has simply never counted through it.

  Eighty Hours for Eight Thousand Dollars

  Here is where the definitional problem stops being abstract and starts costing real staff time. When a home falls outside IHS eligibility, somebody still has to pay for the water hookup, and that somebody is usually the tribe, scrambling for outside money while an IHS field engineer walks them through it.

  GAO’s example is almost too clean to be real. Officials from one IHS area office reported spending more than 80 hours helping a single tribe apply for and manage an outside grant worth $8,000.

  After the tribe received the money, the same officials spent additional hours helping it report on how the grant was spent, on top of the original 80. An agency employee’s time is not free. The officials themselves told GAO that once outside funds are transferred to IHS, managing and overseeing them imposes a significant expense on the agency, expense that dwarfs, in staff hours, the size of the grant it produced.

  Run the arithmetic and it gets worse before it gets better. Eighty-plus hours of a federal engineer’s time, spent to unlock $8,000 for a home the agency’s own charter arguably should have covered in the first place, is not an isolated inefficiency. It is what happens, systematically, every time IHS’s definition excludes a home that a tribal member actually lives in.

  Multiply that by however many mixed households and short leases exist across Indian Country, a number IHS admits it does not track, and you get a hidden administrative tax charged entirely in engineer hours, paid for the privilege of routing around a definition the agency wrote for itself.

  A Fix That Fits on One Page

  GAO’s response to this is not a demand for a wholesale rewrite of federal Indian law. It is a request that Congress define, in the statute itself, exactly what “Indian homes, communities, and lands” means, so IHS stops having to guess at the boundaries of its own mandate case by case.

  Absent that, GAO recommends IHS update its own policy to reflect any new statutory definition once one exists, in consultation with tribes.

  This is not a controversial ask. It is closer to basic housekeeping, the kind of clarifying language that should have existed since the Indian Sanitation Facilities Act was written, rather than something an outside audit has to surface decades later as a formal recommendation.

  What makes the current arrangement particularly galling is that IHS is not accused of malice here, or even of especially poor judgment. The agency is applying a plausible reading of an old statute that never anticipated, or never bothered to specify, how it should treat a reservation home where a grandmother happens to hold a state, rather than tribal, enrollment card.

  The absence of a definition did the damage. Everything downstream (the excluded homes, the outside grant hunts, the 80 hours for $8,000) follows directly from Congress never having bothered to spell out a term that shows up in the law’s own name.

  Who Actually Pays For the Gap

  The people who pay for this gap are not, in the end, IHS engineers, whose hours are at least salaried and accounted for somewhere in a federal budget. The people who pay are tribal families living in homes that fall on the wrong side of a definition nobody wrote down clearly, families who may wait longer for a water connection because their household includes a relative who checked the wrong box on a tribal enrollment form decades ago, or none at all.

  GAO’s report does not put a number on how many households currently sit in this limbo, because, again, nobody is counting. That absence of a number is itself the finding. An agency whose entire job is connecting homes to water cannot say how many homes its own eligibility rule disconnects.

  Congress created this problem by leaving three words undefined in the 1959 Indian Sanitation Facilities Act and letting an agency’s legal interpretation harden around the gap for six and a half decades.

  Fixing it requires nothing more dramatic than actually finishing the sentence.

  Fediverse Reactions

  • 
  • 
  • 
  • 
  • 
  • 
  • 

  7 reposts

  2026.10.06

• The Compliance System That Runs on Trust

  Section 889 of the 2019 defense authorization act did something admirably blunt for a Washington statute; it named five Chinese companies (Huawei, ZTE, Hytera, Hikvision, Dahua) and told every federal agency to stop buying their gear, stop buying anything built on their gear, and stop contracting with anyone else who uses it. Six years on, a new GAO report says the ban worked, more or less. Federal spending with the five firms cratered after 2019 and hit zero in three of the last four fiscal years. Ninety percent of companies with active contracts now say they don’t touch the equipment.

  Read past the highlights box, though, and this is less “mission accomplished” than “nobody bothered to look.”

  Whoever They Say They Are

  The entire enforcement mechanism rests on a checkbox. Contractors register with the government and represent, once a year, that they do or don’t use covered equipment. GSA and Defense Department contracting officers are explicitly authorized to take that representation at face value; GAO found neither agency’s officers “regularly check the accuracy of representations,” because policy tells them not to unless something independently raises a flag. Nobody audits the 90 percent who checked the good box. They’re just believed.

  That might be defensible if the government at least knew who the banned companies actually were, corporately speaking. It doesn’t. The law also covers subsidiaries and affiliates, and identifying those is the entire ballgame; a contractor can’t truthfully represent compliance if it doesn’t know the vendor three tiers down its supply chain is secretly a Hikvision shell.

  GSA has identified about 700 such entities using commercial “supply chain illumination” tools. GAO’s own prior work found the Air Force tested those same tools and got them 60 to 70 percent accurate, flagging risk where none existed and missing suppliers that did. The Defense Department doesn’t keep a subsidiary list at all. Officials said corporate structures change too fast for a list to stay useful, a fascinating argument for why the department shouldn’t try rather than why it should try harder.

  The List Nobody Shared Until Somebody Was Watching

  Here is the detail worth reading past the highlights page for. GSA built that list of 700 shell companies years ago. It never gave it to any other federal agency, not to the dozens also bound by Section 889, not through any of the information sharing channels Congress built for this exact purpose in the 2018 Federal Acquisition Supply Chain Security Act.

  GSA and Defense didn’t reliably share the list even with each other. Both agencies told GAO they exchanged subsidiary information “occasionally” in the past, and only started meeting on it regularly after GAO began asking questions for this audit. One Defense official said the department is “looking to improve this process so that it is no longer ad hoc,” six years and one federal audit after the prohibition took effect.

  Asked why the information hadn’t gone further, both agencies said the statute doesn’t require them to share it. True, and also the whole problem in one sentence; a compliance regime built on knowing which shell companies belong to a sanctioned Chinese firm produced exactly one usable master list across the federal government, and its owner sat on it until an outside auditor came knocking.

  A Search Tool for Information That’s About to Stop Existing

  The one piece of real infrastructure here, an automated tool GSA built with NASA’s help to search representations in real time, is about to get less useful. The administration’s “Revolutionary FAR Overhaul” plans to scrap the annual representation requirement and replace it with a single, one-time box check buried inside a company’s initial contract offer. GSA and Defense officials told GAO they’re worried purchase cardholders won’t be able to pull current compliance data on a given vendor once that happens; neither agency has worked out what to do about it.

  The timing is not ideal. A parallel ban on certain Chinese magnets and metals takes effect in January 2027. A ban on Chinese semiconductors follows in December 2027. Both will demand exactly the subsidiary and supply chain visibility that Section 889 never quite produced, and that the government is now preparing to make harder to track, not easier.

  The Actual Recommendation

  GAO’s fix, delivered with its usual bloodless restraint, is that GSA and Defense should “periodically” share what they know with other agencies. Both concurred. GSA says it’s “working on a plan.” Defense says it will “assess the feasibility” of even collecting subsidiary data in the first place, then figure out later who to send it to.

  Six years, one statute, five named companies, and the government’s best answer to “who actually owns these shell companies” is a plan to think about assessing feasibility. The equipment ban worked. The part where anyone checks whether it’s still working is, by both agencies’ own account, not required by law, and so it mostly hasn’t happened.

  Fediverse Reactions

  • 
  • 
  • 
  • 
  • 
  • 
  • 
  • 
  • 

  9 reposts

  2026.10.06

• Coast Guard Disabled Faulty Fire Systems on Cutters, Leaving Crews Facing Greater Risk

  DHS inspectors found a known fire-suppression defect, more than 9,000 delayed maintenance items and a fleet falling well below readiness requirements even as the Coast Guard continues buying new Fast Response Cutters.

  A fire in the engine room of a Coast Guard Fast Response Cutter is precisely the kind of emergency for which an automatic fire-suppression system exists.

  On some of those cutters, however, the automatic feature has been deliberately disabled.

  After repeated malfunctions caused firefighting agent to discharge unexpectedly and shut down cutter engines, the Coast Guard’s solution was to disconnect components of the system and secure a discharge pin with a zip tie. The workaround stopped the accidental activations, but it created another problem: in an engine-room fire, a Coast Guardsman may have to travel to the back of the vessel, descend through a hatch and manually activate the suppression system.

  The Department of Homeland Security Office of Inspector General says those additional steps would delay the release of firefighting agent and increase the danger to the crew. Its conclusion is unusually direct for a federal audit: failure to resolve the problem “could also lead to injury or death.”

  The finding is part of a broader September 28 audit of the Coast Guard’s Sentinel-class Fast Response Cutter fleet, a growing force of 154-foot patrol vessels used for missions ranging from search and rescue and drug interdiction to maritime security and defense readiness. The auditors found a fleet struggling with maintenance backlogs, escalating funding shortfalls and operational availability far below Coast Guard requirements.

  The fire-suppression problem, however, provides perhaps the clearest example of what those readiness problems mean for the people aboard the cutters.

  And the problem has been known for years.

  A Fire System That Could Shut Down the Ship

  According to DHS OIG, Fast Response Cutters experienced at least 22 random discharges of firefighting agent into their engine rooms between 2017 and 2023.

  The consequences were more substantial than an inconvenient false alarm.

  Activation of the suppression system shuts off engine-room ventilation, closes fuel valves and turns off the engines. Whether the system activates because of an actual fire or a malfunction, the cutter loses propulsion and cannot restart its engines until faults in the suppression system are cleared.

  The Coast Guard traced the failures to environmental degradation caused by moisture accumulating in a remote station box. The suppression system had complied with applicable requirements when the cutters were constructed, according to the audit, but responsibility for maintaining and replacing the equipment subsequently passed to the Coast Guard during the vessels’ sustainment phase.

  By April 2023, the Coast Guard had issued an engineering advisory addressing the recurring problem.

  Its interim solution was decidedly low-tech.

  Crews were directed to disconnect electrodes associated with the automatic suppression system and secure the discharge pin with a zip tie to prevent another accidental release. OIG inspectors photographed the arrangement aboard the Fast Response Cutter Paul Clark. The report includes images of both the zip-tied pin and disconnected electrodes.

  The modification worked in one important respect. OIG said it prevented additional random discharges.

  But disabling the electrodes also disabled the suppression controls in the pilothouse.

  That means a crew confronting an actual engine fire cannot simply activate the affected system from the vessel’s control station. Instead, according to OIG, a Coast Guardsman would have to move aft and climb down a hatch to manually discharge the firefighting agent.

  “These extra steps would delay immediate release of the firefighting agent,” the auditors concluded.

  The Coast Guard estimated that resolving the problem across the fleet would cost approximately $40 million. Yet as of fiscal 2025, according to OIG, replacement of the suppression system had not been included in the Coast Guard’s formal budget request. Officials told auditors that the project would, at best, receive a Tier II designation, which carried a low probability of funding.

  The Coast Guard has since begun moving toward a permanent solution, but that solution will not arrive quickly.

  Newer cutters are already being delivered with changes. According to the Coast Guard’s response to OIG, FRC #59, delivered in March 2025, and subsequent cutters receive different fire-alarm panels and upgrades to their fire-suppression systems. The service is also retrofitting a previously delivered cutter with the new system.

  For the older fleet, however, the Coast Guard estimates completing the retrofit will take at least three years, citing the number of vessels involved, the extent of the required work and availability of funding.

  There is still no estimated completion date.

  OIG consequently left the recommendation open and unresolved.

  A Safety Problem Inside a Maintenance Problem

  The fire system is only one part of a much larger readiness problem documented by the auditors.

  Fast Response Cutters are supposed to maintain at least 85 percent operational availability. OIG analyzed 52 domestic FRCs from October 2022 through June 2025 and found they collectively managed just 72 percent.

  Nearly one-third of the fleet, in other words, was not fully mission-capable at any given time.

  The results were particularly poor in fiscal 2024. Only 13 of the 52 cutters met the minimum availability requirement, and none of the eight Coast Guard districts operating FRCs met the service’s overall requirement.

  Those numbers have operational consequences.

  Auditors attempted to determine which missions the Coast Guard had been unable to perform because cutters were unavailable. They couldn’t.

  The Coast Guard does not maintain that data.

  OIG did identify at least one revealing example. A Fast Response Cutter conducting an interdiction in the Southeast District broke down during pursuit of a vessel suspected of carrying drugs and/or migrants.

  The suspected vessel escaped.

  The maintenance backlog behind those failures is substantial.

  Coast Guard cutter-fleet funding shortfalls grew from $219 million in fiscal 2022 to $486 million in fiscal 2025, according to internal memoranda reviewed by OIG. To manage those shortages, required maintenance was postponed.

  By June 17, 2025, crews had reported more than 9,000 FRC maintenance requests that were overdue or had received extensions from Coast Guard headquarters. One of them involved the cutter that subsequently broke down during the interdiction.

  Delaying the work did not necessarily save money.

  OIG found that postponing dockside maintenance until cutters entered scheduled dry dock could stretch a roughly 90-day maintenance period to 120 days. An additional 30 days in dry dock costs an average of approximately $160,000 per cutter. Across roughly 50 cutters over a four-year maintenance cycle, OIG estimated the practice could add about $8 million in costs while simultaneously keeping the vessels away from their homeports longer.

  There is an accounting problem buried beneath the maintenance problem as well.

  Despite evidence that operational and support expenses were rising, the FRC program reduced its projected life-cycle costs from $9.9 billion in 2021 to $9.2 billion in 2023. OIG concluded that the lower projection conflicted with actual cost increases and failed to account adequately for technical systems and deferred maintenance.

  In its response to the audit, the Coast Guard effectively acknowledged part of that problem. It told OIG that deferred maintenance had artificially lowered the actual costs used to develop its cost estimates and said those costs were being incorporated into a revised life-cycle estimate.

  That is an important distinction. Deferring maintenance can make today’s books look better by moving an expense into tomorrow’s inbox. Unfortunately, the cutter still knows the maintenance wasn’t performed.

  This Problem Predates the Latest Audit

  The new OIG findings also fit a longer pattern.

  The Government Accountability Office reported in earlier examinations of Coast Guard cutter maintenance that deferred work and equipment failures were already reducing cutter availability.

  More recently, GAO documented a Fast Response Cutter whose shaft-seal maintenance was postponed. The seal eventually failed, leaving the cutter unable to operate for four months. Coast Guard district officials told GAO that losing the vessel increased pressure on the remaining cutters, which then had to absorb its missions.

  That sequence is strikingly similar to the problem OIG now describes at fleet scale: defer maintenance because resources are constrained, suffer equipment failures later, lose operational availability, and place additional demands on the cutters that remain available.

  The latest audit puts numbers around the accumulated result.

  A 72 percent availability rate. More than 9,000 overdue or extended maintenance items. A $486 million cutter-fleet funding shortfall.

  And a fire-suppression problem serious enough that the Coast Guard disabled part of the system to prevent it from unexpectedly shutting down the vessel.

  Meanwhile, the Fleet Keeps Growing

  None of this has stopped expansion of the Fast Response Cutter program.

  In September 2025, the Coast Guard exercised a $507 million contract option with Bollinger Shipyards for 10 additional FRCs, increasing the number ordered under the contract from 67 to 77. Congress provided $1 billion for additional Fast Response Cutters as part of nearly $25 billion in Coast Guard funding contained in the One Big Beautiful Bill Act.

  The first cutter financed through that option is expected in fiscal 2028.

  Expansion has continued in the meantime. The Coast Guard accepted its 63rd Fast Response Cutter, the Jeffrey Palazzo, in June 2026.

  There is nothing inherently contradictory about replacing or upgrading equipment while expanding a fleet. New vessels can incorporate corrections that older hulls lack, and the Coast Guard says precisely that is happening with the fire-suppression system.

  But the juxtaposition is difficult to ignore.

  The Coast Guard is expanding a cutter fleet whose existing vessels have accumulated thousands of delayed maintenance requirements, while its own inspector general says nearly one-third of the domestic fleet was not fully mission-capable at any given time during the period examined.

  And the older cutters will not instantly inherit the improvements installed aboard their newer siblings.

  More Money, But OIG Wants a Plan

  The Coast Guard concurred with all three recommendations in the report.

  For maintenance, the service pointed to increased funding in its fiscal 2025 budget reconciliation, fiscal 2026 appropriation and fiscal 2027 budget proposal. According to its response, those increases include approximately $20 million in fiscal 2026 and another $30 million in fiscal 2027 for surface depot maintenance, with the Coast Guard addressing FRC work including five dry docks, five docksides and nine annual inspections.

  The Coast Guard asked OIG to consider the recommendation resolved and closed.

  OIG declined.

  Additional funding, the auditors said, did not constitute a plan explaining how the money would eliminate the FRC maintenance backlog. The recommendation will remain open until the Coast Guard produces documentation showing it has developed and implemented such a plan.

  The same unresolved designation applies to the fire-suppression system.

  The Coast Guard says the fleet-wide retrofit will take at least three years. Its completion date remains “to be determined.”

  That leaves the service in an uncomfortable position.

  It is simultaneously buying the next generation of Fast Response Cutters, attempting to catch up on maintenance for the existing generation, and trying to retrofit a known safety defect aboard older vessels while those vessels remain responsible for some of the Coast Guard’s most consequential missions.

  Those missions include drug interdiction, migrant interdiction, search and rescue, maritime security and national defense.

  The cutters are called Fast Response Cutters for a reason.

  The question raised by the inspector general’s report is increasingly whether enough of them can respond when they’re needed, and what risks their crews are being asked to accept when they do.

  Fediverse Reactions

  • 
  • 
  • 
  • 
  • 
  • 
  • 
  • 
  • 

  9 reposts

  2026.10.05

• The Auto Club That Couldn’t Learn

  American Automobile Association of Northern California, Nevada and Utah sells roadside assistance, home insurance and the comforting idea that someone competent is coming to help you. It has spent the last five years demonstrating, in triplicate, that this idea does not extend to its own workforce.

  On September 24, 2026, a second National Labor Relations Board administrative law judge found the company had engaged in the exact species of misconduct a different judge had already punished it for once. If institutional learning were a job requirement at AAA, half its insurance agents would be out of work.

  The new decision, issued by Judge Mara-Louise Anzalone of the Board’s San Francisco branch, consolidates six unfair labor practice cases (32-CA-321155 and five related dockets) filed by International Brotherhood of Teamsters Local 665 after it organized AAA’s California sales agents in 2021. It runs seventy-eight pages and concludes that AAA bargained in bad faith for four years, prematurely declared impasse, imposed a pay cut worth nearly a third of some agents’ income, retaliated against a union steward, and offered severance agreements that illegally required workers to sign away their federal labor rights.

  The judge ordered reinstatement, backpay, a bargaining order and, in an unusual touch, a video of AAA’s own in house labor counsel reading the notice of violations aloud to the workforce she helped mislead (Administrative Law Judge’s Decision, NLRB Case 32-CA-321155).

  A Sequel Nobody Asked For

  This is not AAA’s first ALJ decision. In March 2026, Associate Chief Judge Arthur Amchan issued findings in a companion case, known informally as AAA I, covering conduct between 2021 and 2022. He found AAA unlawfully fired five agents for union activity, unlawfully withdrew a long-standing employee award program the moment the Teamsters won their election, and issued subpoenas demanding information about workers’ protected activity.

  He specifically credited testimony that a senior AAA official told branch managers to “pull back our support on the agents” without making it obvious, direct evidence, in his words, of the company’s animus (NLRB Region 32, Administrative Law Judge Decision, Case 32-CA-280838).

  Judge Anzalone’s decision picks up almost exactly where Amchan’s left off, and finds the same company doing the same thing. Where AAA I documented unlawful discharges and unilateral changes, AAA II documents four more years of bargaining conducted, in the judge’s words, with “the intent to frustrate the possibility of reaching an initial collective bargaining agreement.”

  The company’s own in house counsel supplied the tell. Asked at the table whether she had ever previously proposed a fifty percent wage cut in negotiations, she said yes, and volunteered that afterward, the union involved had been decertified. Judge Anzalone treated the remark as the case’s Rosetta Stone; it explained, more efficiently than four years of testimony could, exactly what AAA thought “bargaining” meant.

  Process as a Feature, Not a Bug

  There is a temptation to read this as a story about one bad actor. It is more usefully read as a story about what happens when the machinery meant to punish that actor takes half a decade to grind.

  The underlying union election was certified in June 2021. Some of the conduct Judge Anzalone found unlawful, the elimination of renewal commissions worth an average of $98,000 a year to affected agents, was announced before the ink on that certification was dry, and not fully resolved by an ALJ until more than five years later.

  In the interim, AAA implemented its final offer anyway, in August 2023, and only a federal district judge’s injunction under Section 10(j) of the Act stopped it from also killing commissions outright in February 2025 (NLRB Region 32, “NLRB Region 32 Oakland Wins Injunction Requiring AAA to Rescind Unlawful Policy and Maintain Employee Commissions,” January 2025).

  None of this is unusual for the Board’s adjudication process, which is precisely the point. An employer facing a well resourced legal team can implement the disputed change, litigate for years, and absorb whatever remedy eventually lands, confident that the remedy (backpay with interest, a notice on a breakroom wall) will cost less than simply not doing it would have.

  Judge Anzalone’s decision even quotes AAA’s own counsel bragging that a prior fifty percent cut had led to a decertification vote; the calculation being described, in open bargaining session, was that provoking employee anger at the union might be cheaper than raising wages. Whether or not that specific bet paid off here, the broader one, that years of litigation are a survivable cost of doing business, generally does.

  Two Judges, One Company, No Deterrent Effect

  What makes the AAA record instructive is the redundancy. This is not an agency inferring a pattern from a single set of facts; it is two ALJs, working from separate evidentiary records and separate hearings years apart, independently finding that AAA treated its newly organized workforce as an obstacle to be managed rather than a party to bargain with.

  Amchan found unlawful discharges and unilateral changes in 2021 and 2022. Anzalone found bad faith bargaining, a second wave of unilateral changes, two more constructive discharges and retaliation against a steward, running from 2022 through 2025. The AAA II decision is dated the same day the Board’s own docket shows the case being formally transferred upward, meaning this finding, like Amchan’s before it, is not yet final; AAA retains the right to file exceptions with the full Board, a process that can itself take years (NLRB Case Search, Case 32-CA-321155, Docket Activity).

  That is the institutional health problem worth dwelling on. The National Labor Relations Act does not fine repeat offenders more heavily, does not accelerate review for employers with an open prior finding against them, and does not treat a second ALJ decision covering the same company and the same union as evidence that the first remedy failed to remedy anything. Each case proceeds as though the last one had never happened, on its own clock, toward its own possible appeal.

  A company that loses twice, in sequence, on the same underlying dispute, pays roughly the same price as one that loses once. Judge Anzalone’s unusual remedy, ordering AAA’s labor counsel to personally read the notice of violations aloud on camera, reads less like standard relief and more like a judge trying, within a toolkit that offers her almost nothing else, to make the second loss feel different from the first.

  Whether the Board upholds even that gesture on review remains, like the rest of this case, an open question with a very long clock.

  Fediverse Reactions

  • 
  • 
  • 
  • 
  • 
  • 
  • 

  7 reposts

  2026.10.05

• A Public Hearing Got Stayed for a Year While Nobody Could Agree Who Was Allowed to Ask the Question

  In October 2022, the President asked two federal agencies a fairly simple question. Should marijuana still sit in Schedule I, the category reserved for heroin and LSD, alongside a legal finding that it has no currently accepted medical use in the United States. Four years later, the honest answer is that nobody in the federal government has fully answered it, and the paper trail explaining why, laid out in detail in a new GAO report on drug scheduling, reads less like science than like a jurisdictional turf fight that occasionally paused for actual medicine.

  HHS Answers a Question DEA Did Not Think It Was Allowed to Ask

  By August 2023, HHS had its answer. Applying a new two part test it had adopted that same year for judging whether a drug has a currently accepted medical use, HHS concluded marijuana met the standard and recommended DEA move it to Schedule III. The trouble was that DEA had spent decades applying a stricter, older five part test for the same question, one that generally required something close to FDA approval.

  HHS’s new test did not require that. It asked instead whether there was widespread current medical use recognized by licensed practitioners operating in state authorized programs, plus some credible scientific support for at least one such use, a bar dozens of state medical marijuana programs could plausibly clear even without a single FDA approved cannabis product.

  This is not a scientific disagreement so much as a disagreement over who gets to decide what counts as a medical use in the first place. DEA’s traditional test, in place since 1992, effectively let the agency treat state legalization as legally irrelevant, since it generally required something close to FDA approval before any substance without one could be called medically accepted. HHS’s new test, adopted in 2023, built state level medical practice directly into the federal legal standard instead.

  Whichever test wins determines the schedule, and the schedule determines the criminal exposure. GAO’s report is careful to note that FDA officials insist the two tests coexist rather than replace one another, that FDA approval remains, in their words, the most straightforward path to establishing accepted medical use. Straightforward is doing a great deal of work in that sentence for a process that has just taken four years and counting to sort out a single substance.

  The Justice Department Settles a Fight Between Two Other Agencies

  In April 2024, the Department of Justice’s Office of Legal Counsel weighed in and sided with HHS. DEA’s traditional approach to currently accepted medical use, the opinion found, was an impermissibly narrow reading of the Controlled Substances Act. Satisfying HHS’s new two part test was sufficient on its own, the opinion held, regardless of FDA approval or DEA’s own five part framework.

  The opinion also concluded DEA could satisfy the country’s obligations under the Single Convention on Narcotic Drugs by placing marijuana in Schedule III while layering additional restrictions on top through its own regulatory authority, closing off DEA’s other plausible argument for keeping marijuana more tightly controlled.

  DEA published a notice of proposed rulemaking the following month to reschedule marijuana to Schedule III, then a notice of hearing that August. In January 2025, the presiding administrative law judge stayed the entire hearing pending an interlocutory appeal to the DEA Administrator, meaning the agency that had just lost the legal argument at the Justice Department got to pause the hearing on its own initiative while it sorted out internal procedural objections.

  The rulemaking sat that way for nearly a year.

  An Executive Order Breaks the Stall, Then Splits the Difference

  In December 2025, the President issued an executive order directing the Attorney General to take all necessary steps to finish the rulemaking. Four months later, the Acting Attorney General moved, but not by finishing what DEA had proposed in 2024. Instead, in April 2026, the Acting Attorney General issued a final rule immediately placing FDA approved marijuana products and marijuana used under a state medical license into Schedule III, using the narrower authority to reschedule drugs for treaty compliance rather than the broader rulemaking DEA had been sitting on for two years.

  In the same order, the Acting Attorney General withdrew the stalled 2024 hearing notice and issued a fresh one covering the original, wider question of rescheduling all marijuana, with the actual hearing to run from late June through mid July of this year.

  So as of this writing, a portion of marijuana has been in Schedule III since April, reached through the narrowest available legal doorway, while the broader question that started this entire four year process, whether marijuana generally belongs in Schedule III, sat through a new round of hearings that only just concluded.

  Two federal agencies agreed on the underlying science within a year of being asked. It then took a Justice Department opinion, a stayed hearing, a change in test, an executive order, and a second hearing notice to produce a result that only partially answers the question the President posed in 2022, using an authority nobody had originally proposed using for the purpose.

  What the Timeline Actually Measures

  None of the four years this took were spent gathering new pharmacological data. HHS finished its scientific evaluation within roughly a year of being asked, well inside the range GAO’s broader report found typical for far more obscure substances. Every subsequent delay was procedural, a fight over which test applies, which agency’s reading of the statute controls, whether a hearing already scheduled could be paused by the same agency contesting the outcome, and eventually whether the President’s own order was enough to force a decision at all.

  The Controlled Substances Act was written to sort chemicals by their pharmacology. What this timeline actually documents is how much of that sorting, at least for the one substance most Americans have an opinion about, ends up decided by agencies arguing over their own jurisdiction rather than by anything found in a laboratory.

  Sources: U.S. Government Accountability Office, Drug Scheduling: DEA and FDA Should Strengthen Policies and Procedures for Scheduling Controlled Substances (GAO-26-108623, September 2026); DEA Notice of Proposed Rulemaking, Rescheduling of Marijuana (Federal Register, May 21, 2024); DOJ Office of Legal Counsel opinion on HHS marijuana scheduling recommendation (August 2024). Related: The Word That Decides Every Drug Sentence in America Was Never Actually Defined and The Two Year Emergency Order That Needed Seven Years and an Act of Congress to End.

  2026.10.05

• The $21 Billion Contract That Was Worth $74 Million

  Somewhere in the Department of Veterans Affairs, a dashboard once displayed a number so large it should have triggered an alarm rather than a press release. A single contract action, the VA told Congress in May 2025, had a value north of $21 billion.

  The actual figure, according to the VA Office of Inspector General, was just over $74 million. That is not a rounding error. That is a factor of roughly 284, the kind of gap you get when nobody checks the math before it goes to the Hill.

  The OIG’s new review of “contracts terminated for convenience” was requested by Senators Richard Blumenthal and Angus King, who wanted to know what happened when the Department of Government Efficiency, VA senior advisers, and the General Services Administration went hunting through VA’s contract portfolio in early 2025.

  The answer, once you strip away the chest-thumping, is that VA terminated 435 contracts worth about $1.1 billion, out of 2,210 contract actions the OIG reviewed. Everything else on the lists VA sent to Congress was either reinstated, never terminated at all, or simply wrong.

  Efficiency, Measured in Wheelchair Ramps

  The most instructive failure here is not the math. It is what got swept into the “terminated” pile in the first place. VA’s July 2025 submission to Congress claimed 1,667 contracts had been terminated or were being terminated. The OIG found the real number was 366, and 17 of those were later reinstated.

  The other roughly 1,300 actions on that list had nothing to do with termination; they were supply contracts that had simply run their course, including orders for prosthetics, surgical implants, and wheelchair ramps, plus more than 150 healthcare contract actions such as nursing home service orders.

  A Category Management Support Office official told the OIG that DOGE representatives “wanted to claim credit for any contracts that ended,” whether or not the government actually ended them. That is a remarkable sentence to find in a federal audit. It means the savings narrative being fed to Congress was padded with equipment orders for amputee veterans that had simply finished delivering, filed under the same banner as genuine cost-cutting.

  The distinction between “we canceled this” and “this was already over” apparently did not matter much to whoever was compiling the list.

  The Eleven-Hour Review

  To be fair to VA’s rank-and-file contracting staff, the OIG’s account of how this played out reads less like malice and more like triage under fire.

  From February through June 2025, OALC and VA senior advisers issued at least 47 orders demanding that contracting officials justify, terminate, pause, or reinstate contracts, on top of a running stream of follow-up emails from GSA and DOGE representatives. On February 21, 2025, officials were handed 1,049 contract actions to review, get approved by deputy under secretaries, and report back on, all within less than 11 hours.

  Some offices did not finish in time. The CMSO director told the OIG that VA senior advisers and DOGE representatives ordered the terminations to proceed regardless. VA had built a questionnaire asking whether each contract provided direct veteran care, required specialized skills VA could not hire for, or was merely administrative support like scheduling; the tool existed, the deadlines simply did not leave room to use it properly.

  Officials, VHA staff among them, told the OIG that instructions arrived from OALC, VA senior advisers, and DOGE representatives simultaneously, on inconsistently formatted lists that sometimes duplicated the same contract action twice. VA did not build a database to track the churn until April 2025, months after the worst of the chaos had already passed.

  Nobody Spoke Federal Acquisition

  The OIG’s explanation for the inflated dollar figures is almost charming in its bluntness. The officials assembling the lists sent to Congress “were unfamiliar with contracting and acquisition terminology.” VA had built a tracking dashboard to monitor the terminations, and that dashboard miscalculated contract values, apparently without anyone downstream noticing that a single contract does not typically run twenty times the annual budget of a mid-sized federal agency.

  This is the recurring shape of the DOGE-era efficiency push across the government; fast, politically driven demands for numbers, staffed by people with strong opinions about waste and thin knowledge of how contracts actually work, producing figures that collapse the moment an inspector general opens the file.

  VA’s own contracting officers, notably, come out of this review relatively well; the OIG found they generally complied with Federal Acquisition Regulation requirements even while being ordered around by people who did not know what a settlement cost was.

  Of the 267 terminated contracts that reached a settlement agreement, 150 cost the government nothing at all, and the remaining 117 cost about $10.6 million combined, a figure that would have looked almost quaint next to the phantom $21 billion the dashboard once reported.

  The Cleanup, Eventually

  VA has since sent Congress a corrected list, submitted April 13, 2026, after the OIG briefed OALC officials in January. The OIG did not audit that list’s accuracy. It also declined to make any recommendations, since VA’s own procurement leadership simply concurred with the draft report and offered no comment, the bureaucratic equivalent of a shrug.

  What remains is a case study in how “government efficiency” gets manufactured for public consumption. Real terminations, genuinely worth $1.1 billion, buried under a much larger pile of expired supply orders and inflated dashboards, sent to Congress by people who could not tell a completed prosthetics contract from a canceled one.

  The veterans who needed those wheelchair ramps got them either way. The number on the press release did not.

  Fediverse Reactions

  • 
  • 
  • 
  • 
  • 
  • 
  • 
  • 
  • 

  9 reposts

  2026.10.05

• The Number That Stopped Moving in 1985

  In 1985, a gallon of gas cost about $1.09, the Berlin Wall was still standing, and the Indian Health Service decided that $10,000 was a reasonable amount of money to connect a tribal school or ceremonial ground to a water system. Forty years later, gas costs three times as much, the Wall is a tourist attraction, and the Indian Health Service still thinks $10,000 is a reasonable amount of money to connect a tribal school or ceremonial ground to a water system.

  This is not an oversight in the sense of a memo nobody read. Per a new report from the Government Accountability Office, IHS senior officials told auditors they had never adjusted the figure because they did not believe it was constraining anyone (Government Accountability Office, GAO-26-107178). Constrain is an interesting verb choice for a policy that has been used exactly twice in two years.

  A Cap With Nothing Under It

  IHS runs a special project fund that can, in theory, pay up to $10,000 to connect an existing tribal community building, a Head Start school, a gathering place, a ceremonial ground, to a water line already being built for a nearby project. It sounds like a sensible bit of flexibility, the kind of accommodation an agency writes into its own rules so tribal communities do not have to run a separate federal gauntlet just to get a spigot into a building that already has walls and a roof.

  In practice, GAO found that of roughly 800 tribal water projects IHS funded in fiscal years 2023 and 2024, exactly two used special project funds for existing community buildings. Two. Not two percent. Two projects, full stop, out of eight hundred.

  Tribal officials and staff from a tribal health consortium told GAO exactly why the number is so small, and it is not because the buildings do not need water. It is because $10,000 does not buy water infrastructure anymore; it barely buys the truck that delivers the materials to build it.

  Tribal health consortium officials said the cap does not come close to covering even basic connection costs, citing the price of materials and the price of hauling those materials to remote tribal land as the primary obstacles. Faced with a fund that cannot fund anything, tribes do the rational thing and skip it, going hat in hand to EPA or USDA instead, agencies with broader statutory mandates that let them pay for the same buildings IHS’s own rule was theoretically designed to help.

  The special project fund exists on paper as a convenience. In practice it functions as a formality tribes have learned to route around.

  Math From the Reagan Administration

  Here is the part where the story becomes less about neglect and more about arithmetic nobody bothered to do. GAO ran the inflation numbers, and $10,000 in 1985 dollars is worth about $30,000 in fiscal year 2025 dollars, triple the amount IHS is still authorized to spend.

  That is not a rounding error produced by decades of general price creep; it is the kind of gap that appears when an agency simply stops looking at a number it wrote down once and filed away. IHS itself, to its credit, did not try to defend the figure once GAO put the inflation math in front of them. Agency officials acknowledged that construction costs have risen since 1985 and said the limit could warrant reevaluation in light of current labor and materials costs.

  That is a remarkably mild way to describe a policy that has effectively been auto-denying itself for four decades. Nobody at IHS appears to have decided, at any point between the second Reagan administration and now, that $10,000 no longer covered what $10,000 used to cover.

  The number simply sat there, technically still on the books, functionally inert, while every input it was meant to pay for, labor, pipe, transport, concrete, kept climbing. A ceiling that never moves is not neutral. It is a ceiling that gets lower every year, measured against the thing it is supposed to buy.

  The Workaround Tax

  The consequence of a dead-letter fund is not that the work stops; the consequence is that the work gets more expensive to arrange, just not on IHS’s books. When a community building cannot get water through the special project fund, tribes and IHS staff go looking for outside money, usually from EPA or USDA, and that search comes with its own overhead. GAO documents this pattern repeatedly across the broader report; pursuing outside funding to cover costs IHS itself will not cover tends to multiply administrative time for everyone involved, IHS engineers included, because now two or three agencies’ worth of paperwork has to be reconciled instead of one.

  A $10,000 cap that functioned properly would resolve a small, incidental cost in house. A $10,000 cap that has not functioned properly since roughly 1990 instead generates a shadow bureaucracy of grant hunting, all to connect a building that, in many cases, sits a few hundred feet from a water main that is already being installed.

  GAO’s recommendation here is almost apologetically modest. It does not ask Congress to rewrite the Indian Sanitation Facilities Act or reorganize IHS’s funding authority. It asks the Director of the Division of Sanitation Facilities Construction to evaluate whether $10,000 is still appropriate, in light of inflation and increases in labor and materials costs, and adjust it accordingly.

  This is, functionally, a request that the agency check a number against the last four decades of economic reality, a task that requires no legislation, no new appropriation, and no interagency task force. It requires someone to open a spreadsheet.

  What Stagnation Actually Costs

  It would be easy to read this as a minor line item in an 86 page report about a much larger set of problems, IHS’s narrow eligibility rules, USDA’s underwriting demands, EPA’s application backlogs. But the frozen $10,000 cap is worth isolating because it is the cleanest example in the whole document of a specific kind of federal failure, not malice, not even obvious negligence, but a number that was correct once and was never revisited, sitting quietly inside an agency’s own internal rules for so long that revisiting it now counts as a recommendation rather than routine maintenance.

  Somewhere in a tribal community right now, a Head Start school or a ceremonial ground is going without a water connection it could plausibly have gotten already, not because the money does not exist in the federal system, but because the specific pot of money set aside for exactly this purpose has been mathematically useless since roughly the first Bush administration, and nobody thought to check.

  That is not a scandal in the traditional sense. Nobody profited. Nobody hid anything. It is simply what happens when the machinery of government runs on numbers that were never designed to update themselves, and the humans responsible for updating them assumed, without checking, that the machinery was fine.

  Fediverse Reactions

  • 
  • 
  • 
  • 
  • 
  • 
  • 

  7 reposts

  2026.10.05

• The Postal Service Passed Its Pay Audit. The Bill Just Got Postponed.

  Credit where it is due. The Postal Service Office of Inspector General reviewed the compensation, benefits, and bonus authority of 158 postal officers and executives for calendar year 2025 and found nothing to flag. No one exceeded the Executive Schedule Level 1 salary cap of $250,600. None of the eleven employees the Board designated as “critical senior executive” positions cleared 120 percent of the Vice President’s compensation, the $347,280 ceiling PAEA allows for a maximum of twelve such slots. The auditors made zero recommendations. Management, for its part, declined to comment, which in bureaucratic terms is the sound of an agency that has nothing to argue about.

  A Rare Clean Bill

  This is not nothing. The OIG’s CY2020 audit found a $5,833 monetary impact from noncompliance, a rounding error by federal standards but a finding nonetheless. This year there is no finding at all. The Postal Service is, in the narrow and specific sense that this audit measures, playing by its own rules. A watchdog publication built on the premise that institutions rarely get audited into a compliment owes it to its readers to say so when one does.

  But an audit that measures compliance with a cap is not the same as an audit that asks whether the cap means anything. It measures whether water stayed inside the container. It does not ask what happens to the water that would have overflowed.

  The Account That Never Closes

  Buried in Table 2 of the report, well past the highlight box that most readers will stop at, is the answer. When an executive’s earned compensation would push them over their statutory ceiling, the Postal Service does not simply withhold it. It defers it into an interest bearing account and pays it out later, in whatever future year the executive’s total compensation happens to sit below the cap.

  Deputy Postmaster General Douglas Tulino, who also holds the titles of Chief Operating Officer and Chief Human Resources Officer, a three hat arrangement that ought to raise its own eyebrows, deferred $198,650 in calendar year 2025 alone. His running balance sits at $515,462.18. Joseph Corbett is carrying a deferred balance of $538,887.66. David Steiner, the Postmaster General, has $86,695 sitting in the same holding pattern. This is not back pay owed for work not yet done. It is compensation already earned, already valued, simply held in escrow until the calendar catches up to the paycheck.

  The Office of Legal Counsel blessed this arrangement back in 2013, and the Postal Service formalized it in an August 2023 pay policy. It is legal in the fullest sense of the word. It is also, functionally, a way of making a hard salary cap into a soft one. The cap does not limit what an executive is paid. It limits when.

  Twelve Chairs, One Very Full Ledger

  The PAEA’s critical positions exception exists for a defensible reason, the argument that the Postal Service competes for executive talent against private logistics firms that pay considerably more than the federal government ever will. Fair enough. But the mechanism built to enforce a ceiling on that competition has, in practice, become a mechanism for deferring rather than denying the money that would breach it.

  Consider the company those eleven names keep. Chief Retail and Delivery Officer. Chief Logistics and Infrastructure Officer. Chief Technology Officer. Chief Information Officer. General Counsel. These are, by any measure, executive suite jobs at a nine figure operating budget, and the audit confirms their occupants are being compensated like it, just with the accounting spread across tax years instead of concentrated into one that would trip the wire.

  None of this requires a conspiracy. It requires only that a statutory salary cap, once it collides with an actual executive compensation market, gets engineered around rather than enforced. The engineering happens to be perfectly legal, transparently disclosed in an annual report, and blessed by the Justice Department. That is precisely what makes it a story rather than a scandal. Scandals get investigated and end. This just continues, one interest bearing dollar at a time.

  Solvency for Me, Not for Thee

  Here is the number that belongs next to Table 2, and does not appear anywhere in the OIG’s report because it was outside the audit’s scope. The Postal Service closed fiscal year 2025 with a net loss of $9.0 billion under GAAP, an improvement over the $9.5 billion loss the year before only in the sense that a slower car crash is still a car crash. Controllable loss, the figure USPS itself prefers because it excludes costs management says it cannot influence, still came in at $2.7 billion, up nearly a billion dollars from the prior year. Compensation and benefits expense across the organization rose $1.7 billion year over year, the single largest driver working against the modest revenue gains from Ground Advantage and stamp price increases.

  Postmaster General Steiner told the public in November that “the occasional appearance of financial progress is far outweighed by the reality of our significant systemic annual revenue and cost imbalance,” and asked Congress for changes to pension funding rules, the statutory debt ceiling, and workers’ compensation administration. He is not wrong that the agency’s finances are structurally broken. It is worth noting, in the same breath, that while the institution pleads for relief from Congress, its own compensation ledger keeps a small number of executives’ pay intact, merely rescheduled, interest accruing the whole time.

  The Postal Service passed its compliance audit. It deserves to be told so plainly. But compliance with a cap that bends rather than breaks is a different achievement than the audit’s clean headline suggests, and an agency nine billion dollars in the red is not obviously the place where “we will pay you later, with interest” should be standard operating procedure for the corner office.

  For more USPS OIG coverage, see Nashville’s Postal Honor System Just Failed an Audit — on four OIG reports finding that the compliance data Nashville reported to USPS headquarters bore little relationship to what auditors found on the floor — and The Pantry That Held a Federal Problem, on a joint OIG-DEA investigation that dismantled a drug distribution network running through the postal system.

  2026.10.05

• The Fund That Worked Until It Almost Didn’t: FDIC, 2023, and the Return of the Bank Run

  The Slogan With an Asterisk

  The FDIC’s marketing has one genuinely remarkable claim behind it: since the agency opened its doors in 1934, no depositor has ever lost a penny of insured deposits. It is the rare piece of government messaging that happens to be true. It is also, on its own, a slightly misleading way to describe what actually happened in March 2023, when the failure of three regional banks forced federal regulators to quietly abandon the very insurance limit the FDIC exists to enforce.

  Of the entities covered so far in this series, the FDIC comes closest to the textbook version of a government corporation Congress imagined when it wrote the 1945 Government Corporation Control Act: funded entirely by risk based premiums on the industry it regulates, receiving no annual appropriations, and repaying its initial seed capital rather than leaning on the kind of soft, interest free financing that let TVA and the RFC balloon into open ended liabilities. That design held up for ninety years. Then three banks failed inside two weeks, and the difference between “the FDIC works” and “the FDIC has discretion to decide who actually gets protected” became very hard to ignore.

  The Cleanest Design in the Series, Almost

  The FDIC was created by the Banking Act of 1933 after more than 9,000 banks failed between the 1929 crash and 1933, wiping out roughly 1.3 billion dollars in depositor losses, according to the FDIC’s own institutional history. Congress gave it an initial 289 million dollar loan from the Treasury to get started, a debt the agency subsequently repaid in full. Since then, the FDIC has funded its operations entirely through insurance premiums charged to member banks and earnings on its Treasury securities holdings, a structure the Federal Reserve Bank of Minneapolis has noted receives no congressional appropriations at all, which puts the FDIC closer to actually behaving like the self-sustaining commercial entity every government corporation is nominally supposed to be.

  That design worked exactly as intended for eight decades, through the savings and loan crisis and the 2008 financial crisis alike. Then, on March 10, 2023, Silicon Valley Bank failed, followed within days by Signature Bank and, weeks later, First Republic, together the second, third, and fourth worst bank failures in American history, according to reporting on the FDIC’s own inspector general findings. GAO later found that the Federal Reserve and FDIC had identified concerns about both SVB and Signature Bank as far back as 2018, but neither regulator escalated those concerns into formal enforcement action in time to force a fix before the banks collapsed. The FDIC itself estimated the failures would cost the Deposit Insurance Fund roughly 18.7 billion dollars, the bulk of it tied not to ordinary insured deposits but to the decision to protect deposits above the standard 250,000 dollar insurance limit entirely.

  The Limit That Got Overridden Anyway

  That decision is the part worth sitting with. Federal law lets Treasury invoke a systemic risk exception, on the joint recommendation of the FDIC and the Federal Reserve, to cover deposits beyond the statutory cap when a bank failure threatens the wider financial system. GAO reviewed the invocation for SVB and Signature Bank and concluded the decision likely helped prevent broader instability, noting that deposit outflows from smaller banks slowed within a week of the emergency action. In other words, the mechanism worked. But the mechanism only exists because the ordinary insurance limit, the thing depositors are told to rely on and the thing the premium structure is actually priced around, turned out not to be sufficient the moment a bank run could happen at the speed of a group chat rather than a physical line outside a branch.

  The FDIC’s own inspector general did not let the agency off easy afterward. A review released in December 2024 found the FDIC’s readiness to resolve large regional bank failures was, in the OIG’s words, not sufficiently mature to facilitate a consistently efficient crisis response, citing gaps in staffing, technology, and coordination across divisions. The OIG issued eleven recommendations, and the same reporting noted GAO’s own parallel review found FDIC lacked even a centralized system for tracking supervisory recommendations across its own examiners, meaning emerging risks at supervised banks could fall through cracks nobody was specifically watching for.

  Deregulating the Morning After

  Here is where the story takes a turn that fits uncomfortably well with the rest of this series. Rather than responding to that OIG report with a tightening of supervisory practice, the FDIC’s new leadership moved in the opposite direction. Travis Hill, installed as acting chairman on January 20, 2025, moved quickly to reshape the agency’s regulatory posture. Within weeks, he rescinded the FDIC’s 2024 Statement of Policy on Bank Merger Transactions and withdrew four other pending regulatory proposals, part of what one law firm’s analysis described as a wholesale regulatory purge. The stated goal was making merger review faster and more predictable for the banking industry. The practical effect was that the same agency whose own watchdog had just flagged coordination and readiness failures in the run up to 2023 spent its next chapter loosening the review process for the kind of consolidation that produces exactly the large, complex institutions whose failures are hardest to manage. Hill, confirmed by the Senate as the agency’s 23rd chairman in December 2025 and sworn in the following January, had by then already set the agency’s direction.

  This is not corruption or mismanagement in the way the RFC’s cronyism was, and it is not the multi-decade debt spiral TVA has been running twice now. It is something quieter: a genuinely well designed, self-funded government corporation discovering that even its cleanest structural feature, premium funded insurance with a hard dollar cap, still depends on a discretionary political judgment call about who gets bailed out beyond that cap when the alternative looks worse. The FDIC did not need a conservatorship. It needed an emergency exception to its own rules within days of a crisis starting, which is a smaller version of the same problem.

  Why It Matters

  Every entity in this series so far has hit the same wall from a different direction. The RFC discovered that market discretion without oversight curdles into cronyism. Fannie and Freddie discovered that dodging the 1945 Act’s accountability regime just moves the risk somewhere less visible until it detonates. TVA discovered that a debt ceiling only constrains behavior for as long as Congress declines to raise it, which turns out to be not very long at all. The FDIC, built about as well as a government corporation can be built, discovered that the line between “self-funded insurance program” and “implicit federal guarantee” gets redrawn the moment enough money is at stake, and that the agency meant to price and contain that risk can still be caught flat footed by how fast it moves in an era of instant digital withdrawals.

  None of this means deposit insurance is a bad idea, any more than TVA’s debt history means public power is a bad idea. It means the closest thing this series has found to a genuine success story still needed an emergency override within its first real stress test in fifteen years, and that the policy response since has been to loosen the very oversight machinery that watchdog after watchdog just said was not ready. Whichever government corporation this series covers next, that is the question worth asking of it: not whether the design looks sound on paper, but what happens the first week the design gets tested for real.

  Next in the series: Amtrak, the only government corporation here whose founding fiction was written into the planning documents on day one — by the same planners who privately doubted it.

  Fediverse Reactions

  • 

  1 repost

  2026.10.03