Why profits pool at choke points, why brands pay a toll to reach customers they already have, and how to dig the canal instead
“A wealth of information creates a poverty of attention.” – Herbert Simon, 1971
1
One season, two lessons
In the space of a few months in 2026, the Strait of Hormuz taught the world most of what there is to know about choke points.
The strait is a narrow passage between Iran and Oman, and around a fifth of the world’s seaborne oil and gas trade moves through it. When conflict broke out in late February, nobody seized the oil and nobody owned the ships. The passage itself became the problem. By early April, only about 5% of pre-war shipping volume was getting through.
Then came the second lesson. The world had treated the strait as an international waterway and had never paid to cross it. Over the summer came proposals to charge. Iran sought fees of five to seven per cent of cargo value, Oman discussed around three per cent, and Washington wanted none at all. As I write, in mid-October 2026, they remain contested proposals rather than settled tariffs, and ships are again being attacked in and around the strait. But the instinct behind them is as old as trade. Whoever holds a narrow passage is tempted to do one of two things with it: close it, or charge for it.
Every consumer business has lived through both. A platform changes its algorithm and traffic halves overnight: the strait closes for a day. Auction prices rise year after year: the strait charges. The passage between a business and its customers runs through someone else’s water, and the keeper of that water decides when it opens and what it costs.
This essay is about that passage. It makes four arguments. Profits pool at choke points. For consumer brands, the tightest choke point is access to attention. Brands pay its toll twice: once to win a customer, and again to reach the same customer. That round-trip toll is the hidden tax at the centre of modern marketing, and the heart of this essay. And the economics are turning: for a growing number of businesses, digging a canal around the strait is becoming cheaper than paying to cross it.
2
The geography of profit
A choke point is a narrow passage in a valuable flow, where the alternatives cost enough that the passage acquires power out of all proportion to its size. Five properties decide how much power:
- A large, valuable volume must pass through.
- Few viable routes exist.
- Bypass cost. Going around is expensive, slow or inferior.
- Value capture. The keeper can earn from what passes.
- The flow stays, rather than finding another way.
Hormuz scores high on the first three. Its weakness, until this year, was the fourth: it conferred leverage, not revenue. The comparison that matters is with a passage that was built.
The Suez Canal opened in 1869. It was cut through desert to create something geography had not provided: a short route between Europe and Asia that spared ships the long voyage round the Cape of Good Hope. It has charged every ship that uses it for more than 150 years, and ships keep choosing it. Its pricing explains why. The Suez Canal Authority weighs the savings a ship achieves by using the canal against the alternative routes, alongside factors such as vessel type and freight rates, and when market conditions make the canal less competitive, it offers rebates of up to 75%.
The strait charges for the obstruction. The canal charges for the shortcut. One is resented and resisted; the other is chosen, ship by ship. The resistance to the Hormuz proposals shows the difference. The United States sanctioned the authority set up to run the waterway, and insurance wording now bars cover for shipowners who pay. Even the proponents avoid the word toll: Oman’s proposal is modelled on the Strait of Malacca, where a foundation collects voluntary contributions for navigation safety. A charge is accepted only when it pays for a service.
Figure 1. The strait and the canal. Both sit on valuable flows; only one is chosen.
The same geography shapes corporate profits. Others have mapped it well: Chris Miller on semiconductors, Henry Farrell and Abraham Newman on the networks of global finance, Ben Thompson on the aggregators that control digital demand. I will not repeat their work. One table is enough to show the pattern.
| Choke point |
What flows through it |
Its position today |
| ASML |
Machines for the most advanced chips |
The sole supplier of EUV lithography systems |
| Nvidia |
AI compute |
Strong, while its largest customers design their own chips to go around it |
| Windows |
PC software |
Once universal; bypassed by the web and the phone |
| Google Search |
Intent |
Ruled an unlawful monopoly by a US court in 2024; now being rerouted by AI answers |
| Apple App Store |
Apps and payments |
Its commission under regulatory assault in Europe and the United States |
| Visa |
Payments |
Durable: a chosen network whose fee is small relative to what it moves; in India, UPI built a public alternative |
Durable choke points are chosen networks whose fee is small relative to the value they move. Imposed tolls invite bypasses and regulators. Choke points need not be physical either. A machine, an operating system, a default setting, a protocol or a network can each become one.
3
The strait moves
In 1971 Herbert Simon made the observation that frames everything in this essay: a wealth of information creates a poverty of attention. Half a century later, AI is multiplying the wealth. It writes copy, generates images, builds software and personalises offers at almost no marginal cost. It cannot manufacture a single extra hour of human attention. So it deepens the poverty.
Clayton Christensen described what happens next. In The Innovator’s Solution he argued that when one layer of an industry becomes abundant and interchangeable, profits migrate to the adjacent layer that stays scarce. In consumer commerce, products, content and software are becoming abundant. What stays scarce is the moment a person chooses to look. Profit is migrating to whoever governs that moment.
None of this is new. What is striking is how often the strait moves. Across the last century, the passage between brands and buyers has shifted again and again: from the retailer’s shelf, to the publisher’s page and the broadcaster’s airtime, to the search box, the social feed, the app store and the marketplace. Today it is shifting once more, to the AI agent that will increasingly read, sort and act on a person’s behalf.
Figure 2. A century of straits. The passage between brands and buyers keeps moving.
In every era, brands rented the strait, and in every era its keeper became a Profipoly, my word for a business that captures an outsized and lasting share of an industry’s profits. Retailers charged for shelf space. Broadcasters sold airtime. Search engines auctioned intent, feeds auctioned attention, and marketplaces charged commissions and then sold advertising on top. Tim Wu’s The Attention Merchants tells the story of how attention was harvested. This essay is about who paid for it.
Straits also erode. Windows looked permanent until computing moved to the browser and the phone. Search looked permanent until AI began answering questions directly. Even Hormuz has pipelines running round it. A strait is a position in a flow, and flows move. That is the first reason for hope, and the first warning to anyone who would build a new one.
4
The round-trip toll
Picture a manufacturer that pays a canal to ship its goods to a customer, and then pays the canal again every time it wants to deliver to the same customer, even though it holds the customer’s address. Nobody would design a trade route that way. Consumer brands run their growth that way.
The first toll is fair. A brand that wants to be discovered by strangers has to go where strangers are: search, social, marketplaces. That is acquisition, and paying for it is the cost of being found.
The second toll is the strange one. The customer buys. The brand now has an email address, a phone number, perhaps an app install, a purchase history and permission to make contact. Months later, the customer drifts. To bring them back, the brand returns to the same platforms and bids again, often against its own competitors, to reach a person whose name it already knows. Then it does it again. Retargeting, reactivation and reacquisition are the polite names for this.
The keepers of the strait document it themselves. Google’s guide to Customer Match invites advertisers to upload their own customer data, and lists re-activating dormant customers among its uses. Nothing in that is improper, and some paid re-engagement earns its keep. The absurdity begins when buying back customers you already know becomes the default operating system rather than the exception.
Brands are the only travellers who pay a toll to bring their own customers home.
Figure 3. The round-trip toll. The brand pays the strait again and again, while its own route sits idle.
The toll is booked as a marketing expense, and its value expires with the interaction. Next quarter it is paid again. Worse, it is often counted as growth: a customer bought back from a platform appears on the dashboard as a customer acquired. The questions are simple. What share of the revenue you call new comes from people who were already yours? What share of your customers reach you only through someone else’s strait? I track the answer with a measure I call REACQ%: the share of customers bought back through paid channels instead of kept through direct relationships. It is a warning light, not a verdict; proving waste means comparing incremental outcomes and fully loaded costs across routes. In my writing on NeoMarketing I have reduced the remedy to a principle: Never Pay Twice.
Meanwhile the owned route sits idle. The email address, the app, the order history: each is a direct line that bypasses the strait. Brands collected these lines carefully and then used them as pipes for receipts, notifications and discount codes, not as places a customer might want to return to. The attention was never built, so it is rented back.
5
The lost canals
Owned attention sounds like a new idea. It is an old one that brands abandoned. A century ago, some of the most successful companies in the world dug canals.
The Michelin Guide, 1900. When the Michelin brothers published their first guide, France had fewer than 3,000 cars on its roads, and they gave away nearly 35,000 copies. The guide told motorists where to refuel, repair and eat. Its purpose was to make people drive more and wear out more tyres. It stayed free until 1920. A tyre company built a reason to travel, and its stars became a lasting, collectable status that has outlived every tyre Michelin sold that year.
The Furrow, 1895. John Deere’s magazine wrote about farming, not tractors. Its circulation passed four million by 1912, and it is still published. Farmers kept it because it served them, which is why it served John Deere.
The Sears catalogue, 1888. Sears began in 1888 with a mail-order catalogue of watches and jewellery, and grew it into a general catalogue that, as the Big Book, was awaited each year by millions and brought the wider world into the isolated homes of rural America. Its Christmas edition, the Wish Book, followed in 1933. It was owned push, delivered by post, and it was kept: browsed for months, passed around, dreamt over. Sears did not pay anyone to reach those homes again. The catalogue was already on the shelf.
S&H Green Stamps, 1896. Merchants gave shoppers stamps with each purchase. Shoppers pasted them into books and redeemed full books for goods from a catalogue. At its 1960s peak, the S&H catalogue was touted as the largest publication in the United States, and the company claimed that three out of four American households collected the stamps. It was one book with many merchants: a cooperative attention currency, with an album.
Figure 4. The lost canals. Brands once built owned attention at scale.
Why did brands stop digging? Because renting got cheaper. Radio, then television, then digital advertising offered reach that was larger, faster and, eventually, measurable to the click. Owned canals needed editors, printers and postage. For most of the twentieth century and all of the digital era, renting won the make-versus-rent decision, and brands forgot they had ever built anything.
The lost canals also carry warnings. Green Stamps did not fade for a single reason. The great inflation of the 1970s turned shoppers’ attention to the lowest price, which undermined the stamps’ business model. Many supermarkets, squeezed by the energy crisis, began advertising lower prices instead of issuing stamps. The rewards themselves lost value, so collectors needed far more stamps for anything worthwhile. S&H went into decline and was sold in 1981. A shared currency survives only while it leaves both the shopper and the merchant better off. Half a century later, American Express tried a cross-brand programme called Plenti. It shut in 2018, and most members had bought from only one or two partner brands, which defeated the point of a coalition. A cooperative canal works only if people cross between the brands on it.
6
The Coase crossover
In 1937 Ronald Coase asked a question that would later win him a Nobel Prize: why do firms exist at all? His answer was that firms organise production internally when using the market costs more. Every firm draws its boundary where making and buying cost the same.
Attention has a boundary too. A brand can make attention, by giving customers reasons to return to its own channels, or rent it, by buying access through someone else’s strait. For a century the answer was obvious. Renting was cheap and scalable; making was expensive and slow. Both sides of that equation have now moved.
- Renting is getting dearer. Every competitor bids for the same customers in the same auctions, so the price of reaching them rises. AI answers are eroding organic reach in search. Messaging platforms charge per message, and their prices rise. Platform rules change without notice.
- Making is getting cheaper. AI brings the cost of producing a daily, personal, interactive edition close to zero. Email is an open protocol with no per-message toll. Interactive formats let customers act inside the message instead of clicking out of it.
The make-versus-rent equation for customer attention is being rewritten. For businesses able to build a voluntary habit, the crossover has arrived. For the rest, the race is to make it happen.
Figure 5. The Coase crossover as a scenario. Each business crosses at its own point.
One caution belongs here rather than at the end. An open protocol is not an open inbox. Email offers an open and inexpensive route to the customer, but the inbox is controlled by its provider and attention is controlled by the person. A durable attention asset must therefore outlive the message that first created it: progress, collections and relationships that persist after the email has been read.
The crossover also needs a boundary and a test. Making attention is not free. Relevance, editorial judgement, deliverability and trust all cost money, and the cost of a generic paragraph falling to zero says nothing about the cost of something worth a person’s time. The claim applies to repeat-purchase consumer businesses and to customers the brand already has; discovering strangers still runs through rented straits, and should. A business sustained by patents, contracts or switching costs may not need daily attention at all. The test is concrete: compare the fully loaded cost per incrementally re-engaged customer on each route, measured against a control group. For retailers, consumer brands, publishers, financial services and travel, any business whose profits depend on the second, third and tenth purchase, that comparison is now worth running. The crossover turns a structural threat into a choice.
7
Leverage, toll, fee, dividend
There are four ways to take value from a route, and what separates them is who keeps the value.
Figure 6. Four economics of a route. Value moves from the keeper to the traveller.
- Leverage is what Hormuz offered until 2026: the power to disrupt, with no revenue attached.
- A toll is a compulsory charge set by the keeper. The traveller pays because there is no other way through. The proposed Hormuz fees are tolls. So is the auction a brand must enter to reach its own customers.
- A fee is the price of a chosen shortcut. Suez and Visa charge fees. They are priced on the savings they create, and the traveller can always take the long way round.
- A dividend is what the traveller earns when it owns the route. Nobody else sets the price, and the returns accrue to the owner.
Marketing has lived almost entirely in the toll column. I call it the attention toll: pay for reach, receive a temporary chance to be noticed, pay again next time. The alternative is an attention dividend.
| |
Attention toll |
Attention dividend |
| What you pay for |
Reach |
Usefulness, habit and persistence |
| What happens to the value |
It expires after the interaction |
It accumulates across interactions |
| Next time |
Pay again |
Each return makes the next more likely |
| In the accounts |
An expense |
Behaves like an asset |
| Who sets the price |
The keeper of the strait |
The brand |
The asset is not the customer’s time. Nobody can own that. It is the recurring, permissioned relationship and the customer’s accumulated reasons to return: a daily edition they look forward to, a collection they are completing, a record of progress they would lose by leaving. I call this asset Owned Attention. Its operating measure is Real Reach: the share of known customers who have meaningfully engaged in the past ninety days, counted through actions rather than opens, which mailbox privacy features now inflate. Real Reach reveals the productive audience hidden behind the vanity of list size.
Put simply: rented attention is an auction; owned attention has no auction. Nobody bids against you for a place in a relationship you built.
8
One brand builds a pipeline; only a network digs a canal
There is a catch, and it comes from inside the argument. If every brand builds its own daily attention surface, a customer with relationships across forty brands receives forty daily emails. The first few brands gain. The rest recreate the overload they set out to escape. Human attention is fixed, and owned attention built brand by brand runs into a fallacy of composition.
The Gulf producers faced a version of this problem with Hormuz. They built pipelines around the strait, and the pipelines help. But a pipeline serves one shipper and carries a fraction of the flow. One brand can build a pipeline around the strait. Only a network can dig a canal.
A cooperative canal is a shared route to which many brands contribute and which each one makes more valuable. On the consumer side, it looks like the Green Stamps book rebuilt for the digital age: one album, many issuers, where a person’s collection, progress and rewards grow because the album spans brands. On the commercial side, it is a network through which a brand can reach a customer it cannot reach directly, on the trusted surface of another brand or publisher that customer still pays attention to.
Figure 7. The cooperative canal has two layers: an album the person controls, and permissioned routing paid on outcomes.
A shared album is necessary, but it is not enough. Consolidating what people keep is not the same as consolidating what they receive. Forty issuers in one album can still mean forty emails a day. The network has to make attention itself more efficient. People choose which issuers they follow and how often they hear from them, and can take one optional curated edition that brings several issuers together with updates from their own collection, instead of forty separate demands to open.
The commercial layer has its own condition. Reaching a customer through another brand’s or publisher’s relationship must rest on authorised, privacy-preserving matching. Possessing an email address is not permission.
History supplies the design rules:
- The album must be better because it spans brands. Plenti failed because members never crossed between partners.
- Both sides must stay better off. Green Stamps faded once a price cut was worth more to shoppers, and cost merchants less, than a stamp.
- Brands keep their customer relationships. The canal carries the traffic; it does not take ownership of the cargo.
- People control their collections and their attention. Participation is chosen, and leaving is possible.
- The operator earns on service, not on access. The moment it charges for passage, it has become a strait.
The order of travel follows: owned first, cooperative second, rented last. A brand reaches a customer through its own channels where it can, through a trusted partner’s where it cannot, and through the auction only when both fail.
9 Agent-proof attention
The strait is moving again. Personal AI agents will increasingly stand between people and their inboxes, feeds and shopping. They will summarise, compare, renew and buy. In an earlier essay, The Inbox Has Two Readers, I described this as the inbox gaining a second reader: the agent, alongside the person.
The agent will absorb most functional attention. A renewal notice, a price comparison, a delivery update or a routine reorder no longer needs a human to look at it. Any brand whose relationship consists of functional messages will find its attention intercepted, summarised and acted on without a person ever seeing its name.
But agents can’t be fans. Some things lose their point the moment they are delegated: completing a collection, playing a game, discovering something new, belonging to a group. Nobody asks an agent to finish their collection, any more than they ask it to watch a match for them. Attention given for its own sake is agent-proof.
Figure 8. Two readers. The agent takes the functional; the person keeps what is chosen for its own sake.
So brands need two capabilities. For the agent, machine access: structured, trusted, actionable messages that an agent can read and act on with the customer’s permission. For the person, experiences chosen for their own sake. The battle to own the consumer’s agent belongs to the largest technology companies in the world. Brands should not fight it. They should become the counterparty that agents trust and the experience that people choose.
10 How to dig a canal
Building a canal does not require a monopoly. Suez has competitors: the Cape route, rail and air freight. It earns because it is the better route for most ships, most of the time. A canal is built in four stages.
Figure 9. Four stages of a canal. The fourth is a Profipoly, reached without forbidding other routes.
- A cheaper route. Comparable results at lower cost than renting.
- A better route. Results the alternatives cannot match: customers reached who were otherwise lost, actions completed that would otherwise stall.
- A compounding route. Each new brand, publisher and participant makes the route more valuable for everyone already on it.
- A structurally advantaged network. The combination of cost, accumulated consumer value, network density and outcome intelligence becomes hard to copy, and returns stay high.
Suez also supplies the pricing rule. Earn a share of the advantage you create, leave the traveller better off, and cut your price when you are not the better route. In marketing, the operator should earn a performance carry only on verified incremental value above a pre-agreed baseline, measured against a control group. No lift, no carry. The ceiling is set by the traveller’s real alternative, which is not always an advertising auction. For recovering a lapsed buyer, it is the cost of reacquiring them through paid media; for completing a stalled renewal or verification, it is the profit the brand would otherwise forgo. A canal priced this way does not need defending. Its customers defend it.
The danger comes later. Every successful route eventually tempts its keeper to become a strait: to raise prices because leaving is hard, to obstruct the alternatives, to charge for passage rather than for the shortcut. Hormuz in 2026 shows how that ends: sanctions, insurance barriers, coalitions of the unwilling and pipelines round the edge.
Charge for the shortcut, never for the obstruction.
11
How this thesis could fail
A thesis that cannot fail is not worth holding. This one rests on three separate claims, and each has its own kill condition. They fail independently: a network that never forms would not invalidate a profitable direct route, while a habit that never forms would invalidate almost everything.
- The attention claim fails if meaningful return does not persist once the novelty fades and without ever-larger incentives. Then there is no asset, only a new kind of message.
- The economics claim fails if the fully loaded direct route cannot beat the rented route on incremental contribution per customer, measured against control groups.
- The network claim fails if adding issuers and participants creates no measurable extra value for consumers or brands. Then the canal is merely a coalition, and coalitions have a poor record.
Three threats could move the crossover even if all three claims hold:
- The open lane narrows. Mailbox providers or AI agents could turn the inbox into a pay-to-be-seen space, and email’s gate would become a toll.
- The incumbents cut their tolls. This is the most likely response. Platforms can lower the price of reaching existing customers to keep brands in the auction. The strongest answer to a new canal is a cheaper bridge built by the incumbent.
- The canal operator turns into a strait. If whoever runs the network starts charging for access rather than results, brands will build around it, as they should.
Each of these is observable within a few years, and each deserves closer watching than any argument in this essay.
12
The Strait Audit
Every business has a strait. Most have never named theirs. Three questions do it:
- What is your strait? The passage through which most of your growth flows today.
- What is your round-trip toll? What you pay to reach customers you already have.
- What would your canal be? The route you could build, alone or with others, that makes the toll unnecessary.
Then apply the five properties twice, because they measure two different things. For today’s strait, a high score means exposure: much of your growth passes through it, you have few alternatives and leaving would be costly. For the canal you could build, a high score means defensibility: a rival would find it hard to copy.
Figure 10. The Strait Audit. One column measures exposure, the other defensibility.
The reading is direct. High exposure and a weak canal is the most dangerous position, and the one most consumer businesses occupy today. High exposure and a strong canal is an opportunity waiting for investment. Low exposure means the strait is not yet your problem. Check again next year, because straits move.
Herbert Simon saw the scarcity half a century ago. AI has made it acute: as the supply of content becomes limitless, the passage to a person’s attention becomes the most valuable water in the economy. The keepers of today’s straits know it, and they will charge accordingly.
Every business has a strait. You can keep paying to cross it, or you can dig the canal.
