· 5 min read
The advertising that lasts forever is the advertising nobody sees
A Harvard study found 547 of the Million Dollar Homepage's links were already dead. Permanent ad space rots. Everything memorable in advertising history had a clock on it.
In 2005 a student called Alex Tew sold a million pixels on a single web page for a dollar each. It worked. The Million Dollar Homepage raised $1,037,100, the last thousand pixels going at auction for $38,100, and it became the internet's favourite parable about clever ideas.
The interesting part happened afterwards.
In 2017 Harvard's Library Innovation Lab went through the page link by link. Of the 2,816 links embedded in it, 547 were entirely unreachable — the sites simply gone. Another 489 redirected somewhere else, often to a domain resale portal. The dead links occupied 342,000 pixels. At the price paid, that is $342,000 of advertising pointing at nothing.
Those advertisers bought space forever. Forever turned out to be about a decade, and then their ad became a rectangle of colour linking to a parked domain.
That is the thing nobody says about permanent ad space: permanence is not a feature of the medium, it is a bet on the advertiser. The page kept its promise perfectly. The companies did not.
Everything memorable had a clock on it
Run through the advertising anyone can actually name.
The Super Bowl. Thirty seconds. For Super Bowl LX in 2026, roughly $8 million average, with late buyers reportedly passing $10 million — and that is airtime alone, before production. Thirty seconds. People discuss those thirty seconds for a week afterwards.
Times Square. A four-week flight on a standard digital board runs roughly $25,000–$40,000 for a one-eighth share; full-board takeovers go from $150,000 into seven figures. It expires. Somebody else gets it next month. That is precisely why being on it means something.
The back page of a newspaper. One day. Gone tomorrow.
Now the ones nobody can name: the sidebar banner, the footer link, the sponsor logo that has been on some page since 2019. Permanent, cheap per day, and completely invisible.
The correlation is not an accident. Scarcity is what makes attention legible. If anyone can have the space, and have it indefinitely, then having it says nothing about you. When the space is finite and timed, holding it is information: somebody wanted this slot enough to outbid everyone else, right now.
What the infinite shelf did to prices
The web made ad inventory effectively unlimited. Any page can grow another ad unit; any feed can insert another slot. Supply went to infinity, and the value of a unit went where supply-and-infinity always sends it.
The clearest measure is the click-through rate on a standard banner: around 0.05%. Five clicks in ten thousand impressions. Across all display formats it is nearer 0.46%, but the humble banner — the archetypal infinite-supply ad unit — sits at one in two thousand.
The Nielsen Norman Group has been running eye-tracking studies on this since 1997, and the finding has held for nearly three decades: people develop stable avoidance patterns around anything that looks like an ad, in the places ads usually are. Banner blindness is not a design problem you can creative your way out of. It is a learned response to infinite, permanent, unremarkable advertising.
Meanwhile Juniper Research puts ad fraud losses at $100.2 billion for 2026. When inventory is infinite and nobody is scarce, manufacturing more of it is trivially profitable, which is a fairly complete explanation of where that hundred billion goes.
Putting the clock back
So here is the design question I got interested in: what happens if you make internet ad space behave like a Super Bowl slot instead of a sidebar?
That is what BidSurvivor is. Twenty-four slots a day, one per hour. The highest bid holds an hour. Anyone can take it off you by paying more. When the hour ends it closes and the winner is fixed, permanently, in the record.
Three consequences fall straight out of the structure, and I did not have to design any of them:
It cannot rot the way the pixel page did. An hour is over in an hour. There is no ten-year-old rectangle linking to a parked domain, because nothing is held for ten years. The record of who held it stays; the ad space itself is always fresh.
Holding it means something. You beat everyone else who wanted 14:00 today. That is a statement with information in it, in a way "we have been in the footer since 2019" is not.
Losing is visible. Get outbid and you are off the board. The ticker says so by name. That is uncomfortable, and it is the source of the whole thing's tension.
Now the honest part, because I would rather you heard it here. Scarcity is not free. A board with twenty-four slots reaches vastly fewer people than an infinite feed. There is no targeting and no purchase intent — you are buying a moment of attention from whoever is looking, not a person who just searched for what you sell. Compared to search advertising it is a worse tool for most direct-response jobs, and I said so at more length here.
What it is good at is the thing the infinite shelf destroyed: being noticed. One brand, one hour, on a page where the whole point is watching who holds what.
The lesson from the pixels
Alex Tew's page is still up, twenty years on. It is a beautiful artefact and a genuinely brilliant piece of work. It is also a mosaic in which roughly one link in five goes nowhere, and where the eye slides off the whole thing precisely because it is a wall of permanent, undifferentiated, unchanging advertising.
He sold space forever. The lesson of the next twenty years is that forever was the wrong unit.
An hour might be a better one.
Sources. Million Dollar Homepage totals: Wikipedia. Link decay analysis: Harvard Library Innovation Lab, "A Million Squandered". Super Bowl LX pricing: eMarketer. Times Square rates: AdQuick. Banner blindness research: Nielsen Norman Group. Ad fraud: Juniper Research.