Job boards had twenty years

A fair trial

The online job board has now had more than twenty years, essentially unlimited capital, and several complete generations of technology — search, mobile, social graphs, machine learning — to become good for the person looking for work. Twenty years is a fair trial. And the honest verdict from the seeker's side of the glass is: applications vanish into silence, listings are stale or phantom, volume is rewarded over fit, and the experience of using these systems ranks among the most demoralizing activities a professional adult routinely performs.

When a product category fails its users for two decades despite abundant talent and money, the interesting explanation is never incompetence. Nobody stays accidentally bad at something for twenty years. The failure persists because, for the business, it is not a failure at all.

Follow the invoice

Every marketplace serves whoever pays it; this is close to a law of nature. And in the job board model, the money comes from the demand side — employers paying to post, to promote, to search the database, to license the tools. The seeker pays nothing, which settles the question of what the seeker is. Not the customer. The inventory.

Once you see the invoice, every chronic frustration of the seeker experience stops being mysterious and starts being revenue-consistent. Ghosting persists because closure is a cost center that no paying customer demands. Phantom and evergreen listings persist because listing volume is the product the paying side is buying. The application spray is not discouraged, because a large pile of applicants is precisely what was sold. The seeker's wasted evenings never appear on any income statement, so no roadmap has ever had a line item for them.

“You are not the customer of a job board. You are the shelf.”

Why iteration cannot fix it

This is why twenty years of feature releases have changed the surface and not the experience. Better search, better filters, better matching — every improvement is an improvement to the shelf, deployed to extract signal from seekers more efficiently on behalf of the side that pays. The model can be polished indefinitely without once pointing at the seeker's actual interests, because the seeker's interests are not what funds the polishing.

And the timing of the harm is what makes this worse than ordinary consumer-tech misalignment. The person on the supply side of this marketplace is frequently in the most vulnerable financial and psychological season of their adult life. A model that treats a person in crisis as inventory is not a neutral business decision that happens to annoy its users. It is a structural answer to the question of whose interests count — and the answer is: not yours.

So the conclusion worth writing down is not that job boards are evil. It is that they are finished — finished in the sense that the model has fully expressed what it is capable of. Nothing important remains to be discovered inside it. What the person in transition needs cannot be built as a feature of a marketplace that sells them. It has to be built as a different kind of thing, with a different answer to who is served.

What user-sovereign infrastructure requires

Call the alternative what it is: infrastructure that belongs to the person in transition. Not a marketplace with kinder copy — a different structure. Three requirements are non-negotiable, and each one exists to make a specific betrayal impossible.

First, zero-knowledge storage. The raw material of a transition — the career history, the finances, the drafts, the 2 a.m. notes — should be encrypted such that the platform holding it cannot read it. Not will not: cannot. This is the only privacy promise that survives an acquisition, a subpoena-happy business pivot, or a quarterly revenue miss, because it removes the platform's ability to monetize the data even from its own future self. A promise enforced by mathematics does not depend on anyone staying virtuous.

Second, access by consent. The searchable-database model — where recruiters mine profiles and the person finds out, if ever, when the spam arrives — inverts the correct power relation. In a user-sovereign design, visibility is a grant, not a default: the person decides who sees what, scoped and revocable, the way you hand someone a key rather than publishing your address. An inbox you cannot close is not access. It is exposure.

Third, proof over polish. If the system carries any signal at all to the hiring side, it should be verified evidence of capability — dated practice, demonstrated skill, artifacts with provenance — rather than another self-description ranked by keyword. Polish is now free to manufacture and therefore worthless to rank on; ranking on it anyway just re-runs the old model with extra steps. Evidence is the only signal that respects both sides: the seeker is judged on what they can actually do, and the employer reads receipts instead of ad copy.

The category, named

Notice that these three requirements are not features missing from job boards. They are negations of the job board's revenue model — which is exactly why no incumbent will ship them, and why this is a new category rather than an upgrade. Career transition infrastructure: tooling funded by the person it serves, structurally incapable of selling them, that treats the season between roles as a whole thing — nervous system, money, evidence, narrative — rather than as attention to be brokered.

The test for anything claiming the category is short. Who pays, and for what? Can the platform read what you stored? Who turns your visibility on and off? Does the signal it carries prove anything? Ask those four questions of any tool that wants to help with your transition, including ours. Structure is what actually protects you. Everything else is copywriting.

The job board was a reasonable thing to build with the incentives of its era, and it had its twenty years. The person in transition owes it nothing further — least of all their most vulnerable season. That season deserves infrastructure that answers to them. It is finally being built.