Every message has two prices. One is paid by the person who writes it, in the time and thought composing takes. The other is paid by everyone who reads it, in attention. The two bills go to different addresses, and nothing in the medium connects them.
That disconnect is the whole economics of the inbox. Most of what people find maddening about messaging follows from it mechanically.
The sender chooses and the reader pays
A sender makes four decisions — how long, how structured, to whom, and when. The reader pays for all four and controls none of them. A cost imposed on someone who was not party to the decision is an externality, and messaging is about as clean an example as exists.
A sender minimising their own effort is behaving rationally and producing a bad outcome anyway. Dumping an unedited thought on eight people costs one person a minute and eight people a read each. Nothing in the sender's experience registers the difference; a reply arrives or it does not, and the aggregate never appears on their side of the ledger.
Adding a recipient is the purest case. The marginal cost to the sender is zero; the marginal cost to the recipient is a full read. Any mechanism whose sender-side cost does not scale with the number of readers will be overused, and not because anyone is being rude.
What happens as composing approaches free
Effort was the only brake here, and a poor one, but it was real and it sat on the correct side of the exchange. Remove it — and the composing half of email is the half now getting cheaper — and three things happen in order.
Volume rises. Obviously.
Average quality falls, not because senders got worse but arithmetically. The messages added at a lower price are the ones that were not worth sending at the old price. The marginal message is the weakest message.
And variance rises, which matters more than the average. A pile now holds both the most carefully constructed messages ever written and the least, and from the outside they look alike. Length no longer indicates effort; polish no longer indicates care. Every cue readers triaged by was a proxy for sender cost, and sender cost is what fell.
Readers discount the channel, not the message
A reader cannot evaluate a message without reading it. Quality is unobservable in advance — you learn what something was worth only after paying to find out. When verification is expensive and quality is hidden, buyers stop pricing individual items and start pricing the category.
Economists call this adverse selection, and the canonical illustration is a used-car market where buyers cannot tell a good car from a bad one, so they offer the average price and the good cars leave. Messaging runs the same loop. Readers skim, decide in seconds, apply a blanket discount, default to ignoring.
Here is the part senders miss. The discount is not aimed at bad messages — it cannot be, since telling them apart is the thing readers cannot do cheaply. It is applied to the channel. The sender who spent an hour gets the same two-second skim as the sender who spent nothing, so effort stops buying attention at exactly the moment effort becomes optional. Careful senders are taxed for other people's volume, and their rational response is to stop being careful or to move somewhere nobody has flooded yet, which starts the loop again over there.
The discount fails on its own terms too. It is blunt, so it drops good messages along with bad.
Response one — make senders bear a cost
The first structural fix puts a price back on the sending side. Not necessarily money: a cap on how many strangers can be reached, a requirement that the recipient consented first, friction that scales with the recipient list.
Its virtue is that it does not rely on persuasion. Etiquette binds only senders who already care about the reader, and the senders responsible for most of the volume are the ones who do not. A cap binds regardless of intent.
Its costs are real. It taxes careful senders identically, because a cap cannot tell them apart either. Every threshold is arbitrary. And it is enforceable only by something that controls the channel — you cannot cap a medium you do not run, which is why open protocols end up defending themselves with filters instead of prices.
RelayLink takes this side. Reaching someone who already has an account requires an accepted contact pair; reaching a stranger's mailbox is allowed, because otherwise nobody could make a first approach, but it is capped per sender at 25 distinct new recipients per rolling 24 hours by default, and every notification carries a one-click unsubscribe that blocks that sender for good. The reasoning is in the consent layer.
Worth being blunt about the limits. That cap counts distinct new people, not messages. It constrains how many strangers a sender can reach and does not constrain total volume at all, and traffic to contacts who already accepted the sender is not throttled by it.
Response two — make messages cheaper to read
The other fix attacks the other side of the ledger. Leave volume alone and reduce what each message costs to process.
Most of that cost is triage, not reading. The expensive part of a full inbox is not reading the messages that matter — it is discovering which ones those are, one message at a time. Anything that lets a reader decide in two lines whether something concerns them saves the full read on everything it does not.
Structure is the mechanism. The conclusion before the reasoning. An explicit ask with a declared shape — a RelayLink briefing picks one of opinion, decision, review, info or fyi, so the reader knows what kind of answer is wanted before reading a word of context. An honest deadline instead of a manufactured one. Assumptions marked as assumptions, so nobody is fact-checking the sender's confidence on top of answering the question. Writing for a reader who owes you nothing is the practical version.
The cost mirrors the first response. It depends on senders doing work the incentives discourage, which is why it has to be structural — demanded by the format rather than recommended in a style guide. And it does nothing about volume. A legible message you did not need is still a message you did not need.
Neither response is sufficient alone
Caps without structure give you fewer messages that are just as opaque. Structure without caps gives you legible messages in unlimited quantity. Each addresses one half of the asymmetry and leaves the other standing.
Both are also local. A system governs only the messages it carries, while a reader's attention is spent across channels no single system controls. Fixing one channel moves the pressure rather than removing it.
The asymmetry is still the thing to keep in view. Whenever composing gets cheaper without reading getting cheaper, the cost does not vanish — it relocates onto someone who never agreed to it. If you would rather have the sender-side cost built into the channel than left to good manners, connect your assistant.