The Eighty-Five Million Problem: Reading the September 10 ETF Tape Like an Auditor

On September 10, the Farside Investors tape showed the eleven US spot Bitcoin ETFs printing a combined net outflow of roughly $120 million. The same day, the nine US spot Ethereum ETFs printed a combined net inflow of $34.7 million. Net across both complexes: about $85.3 million leaving the wrapper.

The headline writes itself. "Capital rotates from Bitcoin to Ethereum." "Institutions cool on BTC." "The rotation trade is on." I have a problem with that headline, because it is arithmetic masquerading as narrative. A net flow is a residual — the sum of thousands of individual creations and redemptions that cleared against a set of authorized participants who each had their own reason to be there that morning. It is not a sentiment index. It is not a directional signal. It is a settlement artifact that the market has agreed to read as a mood ring, and on September 10 the mood ring was lying, or at least telling the wrong story to the people who only read the total.

The numbers inside the number are what matter. And the numbers inside this particular number tell a story about single-actor concentration, fee-driven redemption, and a structural yield deficit — not about a rotation between two assets. The total is a cliff. The components are a scatter plot. If you only read the cliff, you are not analyzing; you are reacting.

I have spent the better part of a decade taking apart yield curves, liquidation incentives, and governance logic at the contract level, and the lesson that keeps repeating is that the aggregate is often the least informative layer of the stack. The September 10 print is a clean example. It is small — $85.3 million net is, in the context of the global crypto market, a rounding error — but it is structurally revealing. What it reveals is not where the money went. It reveals how little the flow metric actually measures, and how much of the ETF story has become a performance of liquidity rather than a record of it.

Let me walk the tape.

The Machine Under the Ticker

Before I disassemble the print, I want to be precise about what a spot ETF actually is, because the imprecision in the public discourse is not accidental. A spot Bitcoin or Ethereum ETF is not a protocol. It is not a smart contract. It is not a token with a supply schedule and a governance forum. It is a regulated securities wrapper — a trust or a commodity trust structure — that holds the underlying asset through a custodian and issues shares that trade on a primary exchange. The trust does not change Bitcoin's supply. It does not change Ethereum's issuance curve. It changes who holds the asset, at the margin, and how that holding is intermediated.

That intermediation is the whole game. The plumbing has three moving parts that matter for reading a flow print.

The first is the authorized participant — the AP. APs are the only entities permitted to create or redeem ETF shares directly with the trust. They do this in large blocks, called creation units, typically 10,000 or 40,000 shares depending on the fund. When an AP creates, it delivers the underlying asset (or cash, depending on the structure) to the custodian and receives shares. When it redeems, it hands back shares and receives the asset (or cash). Everything else — the tick-by-tick price on the exchange — is secondary market noise that arbitrage collapses back toward net asset value.

The second is the settlement mechanism. Here the structural detail most retail readers never see: the US spot Bitcoin ETFs, for the most part, operate on a cash creation and redemption model, not an in-kind model. That means when an AP creates shares, it does not deliver Bitcoin to the custodian. It delivers US dollars, and the fund's trading desk goes into the spot market and buys the Bitcoin, or executes a mirrored trade against CME futures and physical exposure. The same in reverse for redemptions. This was a concession to the SEC, which did not want the ETF complex directly touching the underlying rails, and it is a concession that has real consequences for tracking error, for market impact, and — critically — for how you should read a daily flow number.

The third is the fee layer. Each fund charges an expense ratio, taken daily out of the NAV. BlackRock's IBIT sits at 25 basis points. Fidelity's FBTC at 25. ARK's ARKB at 21, with a fee waiver. Grayscale's GBTC, the converted trust, sits at 150 basis points — six times the cheapest competitors and, in practice, the single most important variable in the entire Bitcoin ETF flow complex.

Those three parts — AP, cash settlement, fee — are the machine. When you read a flow print, you are not reading conviction. You are reading the output of a machine whose inputs are arbitrage spreads, funding costs, tax calendars, and fees. Trust is a variable, not a constant — and in the ETF wrapper, trust is literally denominated in basis points.

With that in hand, the September 10 print stops being a headline and starts being a forensic exhibit.

Concentration: The 65% Nobody Mentions

The Bitcoin complex bled roughly $120 million on the day. The instinct is to spread that bleeding evenly across the eleven funds and call it a broad retreat. The tape does not support that. On the day, ARKB — ARK's fund — accounted for about $78 million of the outflow. GBTC accounted for about $27.2 million. IBIT — BlackRock's flagship — accounted for about $19.5 million. The remainder was scattered small redemptions across the tail.

Do the division. ARKB alone was approximately 65% of the entire Bitcoin ETF outflow for the day. GBTC was roughly 23%. IBIT was about 16%. Three funds, one direction, wildly different magnitudes.

This is the single most important detail in the print, and it is the one the headline erases. A $120 million outflow driven by one fund is not a $120 million outflow. It is a single large redemption — or a cluster of redemptions from one channel — wearing a market-wide costume. When 65% of a data point originates from one node, you are not measuring a system. You are measuring a participant.

I have seen this pattern before, in a different context. In 2017, while reverse-engineering the governance logic of the 2x2 DAO against its incomplete Solidity codebase, I found that the outcome weights in their voting mechanism could be manipulated by a single actor through an integer overflow in the weight accumulator. The public dashboard showed distributed governance. The code showed a single point of control. The lesson stuck: the map the market looks at is rarely the territory the code actually occupies. The September 10 tape is the same species of error. The dashboard shows "Bitcoin ETFs," plural, as if the complex were a diversified institution. The underlying reality is a handful of AP desks, each of which can move the aggregate by an order of magnitude with a single decision.

Why does the concentration matter for your read? Because the interpretation flips entirely depending on which fund moved. A broad-based outflow across all eleven funds would suggest a genuine shift in institutional appetite — a reassessment of the asset, a macro rotation, a risk-off impulse. A single-fund outflow suggests something much more mundane and much more local: a rebalance, a profit-take, a tax-loss harvest, a single allocator moving size, or an AP unwinding a hedged position. The first is signal. The second is plumbing. The September 10 print is the second.

And there is a further tell. ARKB's outflow was large enough to dominate the complex but not large enough to trigger visible arbitrage dislocation — the fund's price did not decouple from NAV in any meaningful way, and the spot market absorbed the flow without a violent candle. That tells you the flow was anticipated and intermediated, not panicked. Institutional redemptions that represent genuine conviction change tend to leave a footprint in the basis and in the options skew. This one did not. It was a desk doing desk things.

The GBTC Treadmill

Now the GBTC line: roughly $27.2 million out. If you have watched this fund for the past year and a half, this number does not surprise you, and its non-surprise is itself the insight.

GBTC began life as a closed-end trust with no redemption mechanism, which allowed it to trade at a persistent premium to NAV during the bull cycle. When it converted to a spot ETF in January 2024, that premium collapsed and, at times, flipped to a discount, and the trust unlocked the one thing it had never had before: an exit. Since then, GBTC has been a structural source of Bitcoin selling pressure, almost every day, because of a single variable — its 150 basis point fee.

This is not a mystery and it is not a scandal. It is a straightforward consequence of price competition in a commoditized product. When the underlying exposure is identical — one Bitcoin is one Bitcoin, whether it sits in IBIT or GBTC — the only axis of differentiation that matters to an allocator is cost. A 150 basis point fee against a 25 basis point competitor is a 125 basis point annual drag, which over a multi-year holding period compounds into a material performance gap. Sophisticated allocators do not tolerate that drag when they can get the same exposure cheaper. So they redeem from GBTC and, if they want to maintain exposure, they buy IBIT or FBTC or ARKB. The Bitcoin does not necessarily leave the complex. It changes wrapper.

This is the distinction that the flow headline destroys. GBTC redemptions are often not net Bitcoin selling. They are net fee arbitrage. An allocator redeeming $27 million from GBTC and rotating it into a 25 basis point fund removes roughly $27 million from one line item and adds a similar amount somewhere else — sometimes on the same day, sometimes within the week. If you only look at the red fund and not the green fund, you conclude that institutions are selling. If you look at the whole complex, you often find a wash. The map is not the territory.

I want to be careful here, because there are days when GBTC redemptions are genuine selling — when the redemption proceeds go to fiat, to Treasuries, to a different asset class entirely, or to the sponsor's own liquidity needs. The point is not that GBTC outflows are always benign. The point is that the flow number cannot distinguish between a wrapper swap and an exit, because both look identical on the tape. The ETF flow metric is blind to intent. It records movement, not motive. Silence is the only audit that matters — and the silence of the tape on intent is the loudest thing in this print.

IBIT's Rounding Error

Then there is BlackRock. IBIT, the largest and most liquid of the Bitcoin ETFs, printed roughly $19.5 million of net outflow. I want to be blunt about this number: it is a rounding error.

IBIT's assets under management are measured in the tens of billions. A $19.5 million daily flow against that base is approximately one basis point of movement — the kind of fluctuation that a single creation or redemption basket can produce by itself, and the kind of fluctuation that is statistically indistinguishable from zero given the fund's normal daily volume. To read $19.5 million out of IBIT as evidence of institutional cooling is like reading a single person leaving a stadium as evidence that the team lost. The crowd is still there. The lighting is still on.

The contrast matters because it is diagnostic. In a genuine institutional retreat from Bitcoin — the scenario the headline implies — you would expect the deepest, most liquid, most institutionally-embedded fund to show it first. IBIT is where the largest, most long-horizon allocators sit. It is the fund of choice for registered investment advisors, for pension-adjacent mandates, for the kind of capital that moves on annual rebalancing schedules, not on daily sentiment. If that cohort were leaving, IBIT would be leading the charge.

It was not. It was near-flat. The outflow was concentrated in the two funds with the highest fee sensitivity (GBTC) and the most tactically-traded investor base (ARKB). That is a structural signature, not a sentiment signature. The long-horizon money stayed. The short-horizon money adjusted. The headline called the second thing the first thing.

In the void, only the immutable remains — and what remains on this tape is the stability of the institutional core, behind the noise of the tactical periphery.

The Ethereum Side and the Missing Yield

Now the other half of the print. The Ethereum complex printed roughly $34.7 million of net inflow, led by ETHB at about $22.9 million and ETHA — BlackRock's Ethereum fund — at about $9.7 million. On its face, green is green. On closer inspection, the Ethereum ETF complex has a structural problem that no amount of inflow can hide, and it is a problem that will define the product's competitive position for years.

The problem is that the US spot Ethereum ETFs, as currently structured, do not stake. The underlying ETH sits with the custodian, idle, earning nothing. No staking, no staking rewards, no yield. The ETF delivers price exposure to Ethereum and nothing else.

Sit with that for a moment. Ethereum is a proof-of-stake network, and one of the defining characteristics of holding ETH natively — as opposed to holding a price-tracking wrap of it — is the ability to participate in consensus and earn issuance. On-chain, the yield is not trivial. It is several percent annually, and for a long-horizon allocator with a discount rate, a yield stream is not a garnish. It is the difference between a product that compounds and a product that merely tracks.

So the Ethereum ETF, structurally, ships with an embedded opportunity cost. A pension mandate deciding where to allocate ETH exposure faces a direct comparison: the ETF, which tracks price but forgoes staking yield, versus a native staking position, or a liquid staking token, or a regulated staking product, each of which tracks price and captures yield. On a risk-adjusted basis, the ETF's exclusion of staking is a permanent handicap. It is a product that is structurally inferior to its own underlying — not because of execution, but because of a regulatory constraint that forbids the fund from doing the one thing that makes Ethereum economically distinctive.

This is why the ETH inflow number, even when green, should be read with a discount. It tells you that some allocators want exposure. It does not tell you that the product is competitive. It tells you that the allocators either cannot access staking directly — a real constraint for many regulated institutions — or are willing to forgo the yield for the convenience of the wrapper. The former is a captive demand pool. The latter is a shrinking one.

I have spent time on the other side of this trade. In 2024, working with a European fintech on a zk-SNARK implementation for their KYC process, the hardest negotiations were never technical — they were with the legal teams who feared the opacity of cryptographic proofs and the yield structures they might enable. Translating a cryptographic guarantee into a compliance-safe framework took eight months. That experience taught me that the binding constraint on crypto product design is almost never the code. It is the legal envelope around the code. Code compiles; people break — and in the Ethereum ETF case, the code compiles fine. It is the envelope that is missing the staking clause, and the envelope is the product.

If and when the SEC permits staking within the ETF wrapper, the competitive landscape for ETH products changes overnight. Funds that can stake will be able to offer a yield-enhanced version of the same exposure, and funds that cannot will be structurally obsolete. Every ETH ETF that lacks a staking pathway is, in effect, a product with a built-in expiry option on regulatory change. The inflow on September 10 is not a verdict. It is a stay of execution.

The Rotation Thesis Fails Arithmetic

The most seductive misreading of September 10 is the rotation thesis: that capital left Bitcoin and went to Ethereum. It is a clean narrative, and it is arithmetically broken.

Bitcoin ETFs lost roughly $120 million. Ethereum ETFs gained roughly $34.7 million. If the same capital were moving from one to the other, the magnitudes would roughly match — you would see red on one side and green of comparable size on the other, netting to near zero or to a modest residual. What we actually see is a $120 million drain and a $34.7 million fill. The green absorbs less than a third of the red. The residual — roughly $85.3 million — did not rotate anywhere within the crypto wrapper complex. It left, or it moved to something outside these two products, or it moved between wrappers in a way the flow data cannot see.

So whatever happened on September 10, it was not a rotation between BTC and ETH ETFs. It was a Bitcoin-side outflow event and, separately, a smaller Ethereum-side inflow event. The two are not a causal pair. The headline welded them together because they occurred on the same calendar day, which is the same logic that says two unrelated things that happen on a Tuesday are connected.

This matters because the rotation narrative has real market consequences. It sells newsletters. It moves the ETH/BTC ratio in the short term as traders front-run the imagined flow. And it is a story the industry has strong incentives to keep telling, because rotation implies an active, interconnected, sophisticated market — a market where capital is constantly reassessing and reallocating, where there is always a next trade. The reality on this tape is quieter: a large localized redemption and a small localized subscription, unrelated in cause and mismatched in size.

Decentralization is a promise, not a guarantee, and the same is true of the rotation narrative. It sounds like a structural feature. On closer inspection, it is a coincidence being sold as a trend.

The Plumbing That Shapes the Print

I want to go one level deeper, because the cash creation/redemption model on US spot Bitcoin ETFs produces a specific and underappreciated distortion in how daily flows should be read.

Under an in-kind model, an AP creating shares delivers Bitcoin and receives shares. The Bitcoin moves from the AP's balance sheet to the custodian and does not touch the spot market. No market impact, no tracking drag, minimal trading cost. Under a cash model, the AP delivers dollars, and the fund's trading desk must go buy the Bitcoin in the spot market — or, more commonly, execute a mirrored trade that replicates physical exposure. That trade has a cost: the bid-ask spread, the market impact, the execution timing risk. The fund bears that cost, and the cost shows up as tracking error.

Now think about what this does to a flow print. In a cash model, a large daily flow — in either direction — forces the fund's desk into the market to hedge or to source the asset. That trading, in turn, can move the spot price, which can move the basis, which can trigger further flows. The flow number you read at end of day is not simply a record of investor decisions. It is partly the residue of the fund's own mechanical trading. The tap and the meter are not fully separable. The measurement instrument moves the thing it measures.

This is why I distrust single-day flow prints as signals. They sit at the intersection of investor intent and mechanical hedging, and the two are entangled. A $120 million outflow on a cash-settled complex is partly an investor decision and partly a desk's unwind of the associated hedges. If you read it as pure investor conviction, you are attributing to allocators a decision that the plumbing is generating on their behalf.

There is a second-order effect that I think is more important, and it connects to a structural forecast I have been making for some time. The Ethereum ETF complex, like the Bitcoin complex, is downstream of a settlement layer whose economics are themselves in flux. The blob space introduced by EIP-4844 and Dencun gave rollups cheap data availability, and that cheapness has driven an explosion in blob demand. My contention is that this demand will saturate blob space within roughly two years, at which point rollup data costs will rise again and, with them, the effective gas economics of the entire L2 stack. When that happens, the cost structure that makes certain Ethereum-native yield and DeFi strategies attractive shifts, and the relative appeal of the ETF wrapper — which cannot capture any of that yield anyway — shifts with it. The September 10 inflow is a momentary snapshot of an ecosystem whose data economics are mid-transition. Two years from now, the same dollar entering an ETH ETF will be competing against a materially different on-chain cost structure.

The Fee-Layer Blind Spot

Every forensic pass eventually lands on the fee layer, because fees are where the quiet decisions live. The ETF complex's fee war is well known — IBIT and FBTC at 25 basis points, ARKB waiving fees to near-zero temporarily, GBTC stubbornly at 150. But the interesting structural point is not which fee is lowest. It is that the fee layer is the only place where an ETF wrapper can differentiate, and that fact has a dark consequence for the flow data.

Because the exposure is identical, the only reason an allocator moves funds is cost. And because cost moves are mechanical, they are largely predictable. A rational allocator with a multi-year horizon and a fee-sensitive mandate will, given the opportunity, move out of the expensive wrapper and into the cheap one. That move shows up on the tape as an outflow from the expensive fund and, if the allocator maintains exposure, an inflow to the cheap fund. It is a wrapper migration, not a capital event.

The blind spot is that the market reads wrapper migration as capital flight. If the migration is large enough, it can trigger its own narrative — "institutions are leaving," "demand is softening" — which can move price, which can trigger genuine redemptions from allocators who take the narrative as evidence, which can deepen the outflow. A fee-driven wrapper migration, misread as sentiment, can bootstrap a real sentiment event. The algorithm saw the crash, not the pain — and here, the algorithm is the flow tape, and the pain is the allocator who redeems because the tape told them to.

The Manufactured Liquidity Story

I have to name something directly, because it sits underneath all of this. The ETF flow narrative is not the only story in crypto that manufactures itself out of a measurement artifact. The broader version is the recurring claim that "liquidity fragmentation" is a crisis — that capital is dangerously dispersed across too many chains, too many rollups, too many wrappers, and that the solution is consolidation into a unified liquidity layer, usually sold as a new product.

I do not buy it. Liquidity fragmentation is not a market failure. It is the definition of a competitive market. Capital dispersed across venues is capital that has chosen, for its own reasons, to be in those places — for yield, for latency, for regulatory comfort, for counterparty preferences. The "fragmentation" framing reframes that choice as a bug, which conveniently positions the framer's consolidation product as the fix. The narrative is not derived from the data. The data is being arranged to sell the narrative. Logic holds until the ledger bleeds, and the ledger here is the balance sheets of the projects that fund the fragmentation panic.

The same pattern applies to single-day ETF flows. The $120 million headlined outflow and the $34.7 million headlined inflow are real numbers, but the narrative welded around them — rotation, cooling, divergence — is a product being sold, not a finding being reported. The finding is much smaller and much less dramatic: one fund had a large localized redemption, another complex had a small localized subscription, and the aggregate net was a rounding error against the global crypto market's daily turnover. The story is bigger than the event. That is the signature of a manufactured narrative.

The Positive Signal Nobody Prices

I want to end the analytical core with a signal in this data that almost nobody is pricing, because it runs against the industry's steady pessimism about Bitcoin's long-term security model.

Bitcoin's security budget is a function of two things: the block subsidy, which halves every four years and trends toward zero, and transaction fees, which are supposed to replace it. For most of Bitcoin's history, fee revenue has been a negligible sliver of miner income — a few percent at best, often less. The structural fear, the one that gets written up every cycle, is that as the subsidy decays, fees will not grow fast enough to sustain the hash rate, and the network's security assumptions will quietly erode.

Then came the inscription wave — Ordinals, BRC-20, the entire apparatus of writing data to Bitcoin's blockspace. Whatever you think of the aesthetics, the economic effect was undisputed: it produced demand for Bitcoin blockspace on a scale the network had never seen, and with that demand came fee revenue. In periods of inscription congestion, fees briefly became a double-digit percentage of miner income. That is not a rounding error. That is the first credible evidence that the fee market can, under the right demand conditions, actually matter.

I will go further, and I will state this as a structural opinion rather than a data point: without the inscription wave, Bitcoin's security model would already be in visible trouble. The subsidy decay is scheduled and inexorable. The fee alternative was theoretical. The inscriptions turned the theoretical into the observed. They gave miners a second revenue source with real blockspace demand behind it, and they did it precisely in the window where the security-budget debate was transitioning from academic to urgent.

The relevance to the ETF discussion is this: the ETF complex is often described as the thing that "legitimized" Bitcoin and brought institutional demand. That is true as far as it goes, but it describes the demand for the asset, not the demand for the network's security. The ETF wrapper holds coins and pays miners nothing directly. The fee market is what pays miners, and the demand that actually sustains the fee market has come, in the recent cycle, from blockspace consumers — inscriptions, and the broader data-demand trend — not from ETF allocators. The institutional demand and the security demand are two different demand curves, and the industry conflates them constantly. If you want to assess Bitcoin's long-term viability, watch the fee share of miner revenue. The ETF tape tells you where coins sit. The blockspace tape tells you whether the network can afford to keep them safe.

The Contrarian Read: The Tape Is Lagging, Not Leading

Here is the counter-intuitive angle, and it is the one I would stake my read on: daily ETF flow data is a lagging indicator presented as a leading one, and the market's treatment of it as leading is the actual risk.

The convention is to treat a flow print as a signal of institutional intent that will propagate forward. Big inflow, expect price up. Big outflow, expect price down. Traders front-run the print. Analysts build narratives on the print. The print becomes self-fulfilling for a day or two.

But trace the causal chain. Why did the allocator redeem on September 10? Not because of September 10. They redeemed because of a decision made days or weeks earlier — a rebalancing trigger, a tax calendar, a liquidity need, a mandate change, a fee migration. The flow print is the settlement of a decision that has already been made. By the time the tape prints it, the information content is largely spent. The price has already adjusted, the desk has already hedged, the allocator has already acted. You are reading yesterday's intent in today's ink.

The self-reinforcing layer is where it gets dangerous. Because traders treat the lagging print as leading, they trade on it, which moves price, which makes the print appear predictive, which reinforces the convention, which draws more traders to trade on the next print. The signal is manufactured by the market's belief in the signal. This is not a crypto-specific phenomenon — it is the same reflexivity that turns any widely-watched metric into a partly self-fulfilling one, from the VIX to the nonfarm payrolls. But in crypto, where the flow metric is young, thinly understood, and heavily narrated, the reflexivity is more acute and the underlying information content is lower.

So the contrarian conclusion is simple and uncomfortable: the September 10 print, and every print like it, is best read as a record of completed plumbing, not a forecast of future exposure. The signal you can actually act on is not the number. It is the composition — which fund, how concentrated, what fee layer, what settlement mechanism. The total is theater. The components are data.

And the deepest blind spot is that everyone is watching the wrong instrument. The ETF tape measures where coins are held. The things that actually determine whether those coins are worth holding — the fee market, the blob space economics, the regulatory envelope around staking — are measured elsewhere, less visibly, by fewer people. The crowd is reading the wrapper. The wrapper is reading nothing back.

Takeaway: What to Watch Next

I do not want to close with a summary, because a summary would imply the analysis is complete. It is not. This is a structural read of a single day's residue, and structure evolves.

So here is what I am watching, in order of signal quality.

First, the persistence of ARKB's outflow. One day of concentrated redemption is noise. Three consecutive weeks of it is a signal about a specific allocator base, and it will tell you whether the September 10 event was a rebalance or a genuine reassessment of the tactical cohort.

Second, the GBTC treadmill's slope. If GBTC's redemptions slow, it means the fee-driven wrapper migration is approaching exhaustion — the price-sensitive allocators have already left, and the remaining base is either captive or indifferent to the 125 basis point gap. If they accelerate, it means the migration is still in early innings and the structural selling pressure has further to run.

Third, and most important for the next cycle: the regulatory status of staking within the Ethereum ETF wrapper. Every month that staking remains excluded is a month the ETH ETF complex is structurally inferior to its own underlying. The day that changes — if it changes — the flow dynamics of the entire Ethereum complex will reprice, and so will the relative appeal of native staking, LSTs, and restaking products. Watch the envelope, not the ticker.

Fourth, the blockspace economics that nobody is connecting to this conversation. The blob space saturation clock is ticking, and the fee market's trajectory — on both Bitcoin and Ethereum — is the deepest variable in the entire stack. The ETF tape tells you where the coins sit. The blockspace tape tells you whether the chains that hold them can stay secure and cheap. Of the two, only one of them matters in five years.

The September 10 print will be forgotten by next week. The structures it revealed — concentration, fee-driven migration, a yield deficit, and a market trained to read residuals as prophecies — will not. We coded the escape, but forgot the exit — and the ETF complex, for all its regulatory legitimacy and institutional gloss, still cannot tell you the difference between an investor leaving and a wrapper changing hands. Until it can, every flow headline you read is a story about a number, told by people who profit from the story.