How outbid.lol Re-invented "Attention Markets"

By the time the Outbid.lol copycats became difficult to keep track of on X, somebody had already started keeping track of them properly.

As of the morning of August 25, outoutbid.lol had verified 409 pay-to-rank boards inspired by the format. It found that 189 of their domains were registered within 48 hours of Outbid's own domain. Of the clones publishing readable payment totals, the median had taken $17.

That $17 is probably the most useful number to come out of the whole episode.

Jonathan Wilke launched Outbid on August 19 at 11:08 p.m. The product hardly needs an explanation: submit a product or website, pay for a position, and rank according to how much you have paid. If someone spends more, they move above you. Outbid's own site currently reports more than 1.33 million visitors since launch, with the leading position sitting at $17,000. The traffic was heavy enough that Wilke says he had to replace his analytics provider during the initial surge.

Within days, hundreds of people had recreated the basic idea.

Most of them did not recreate the result.

That makes Outbid more interesting than another viral indie-hacker success story. It is a unusually clean example of a problem that is becoming more important as AI lowers the friction involved in making software: the product can be copied much faster than the audience around it.

There is an old name for the market that forms around that problem.

"Attention market" sounds like a phrase invented for X in 2026. It isn't.

The underlying idea goes back decades. In a 1971 essay called Designing Organizations for an Information-Rich World, economist and computer scientist Herbert Simon wrote about what happens when information becomes plentiful. More information does not give people more capacity to consume it. It creates greater competition for the limited time people can spend paying attention.

Simon was writing long before social feeds, push notifications, recommendation algorithms or generative AI. The basic constraint has held up remarkably well.

The supply of information can grow almost without limit. The day stubbornly remains 24 hours long.

Economist David Evans later used the term attention markets explicitly in a paper first posted in 2017. His model describes platforms that attract people's time with content and then sell advertisers access to that attention. Television fits the model. So do many websites, search engines and social networks.

Outbid belongs to that lineage, but its design is unusual.

A conventional ad platform tries to make the advertising machinery disappear into the background. Users come for search results, videos, news, posts or entertainment. Advertisers compete behind the scenes for access to them.

Outbid puts the competition on the page.

The bid is public. The ranking is public. A company moving into first place is visible to everyone. The amount somebody is willing to spend becomes part of the reason people visit.

Advertising has existed on the internet for decades. Watching advertisers fight for the advertisement is a different experience.

Outbid itself now has a category labeled "Leaderboards & Attention Markets," which is a fitting description even if the term predates the site by years.

There is also a much older website that makes Outbid feel strangely familiar.

In August 2005, British student Alex Tew launched The Million Dollar Homepage.

Tew divided a webpage into one million pixels and sold them as advertising space for $1 each, generally in blocks of 100. Buyers could place a small image on their pixels and link it back to their own site. The concept spread far beyond what Tew initially expected, and the final 1,000 pixels were eventually auctioned for $38,100. The project brought in just over $1 million in gross revenue.

Looking at the page now is like opening a preserved slice of the 2005 web. Logos, casino ads, software companies and tiny graphics fight for space in an almost unreadable mosaic.

The important part was never the quality of the advertising. People visited because the page itself had become a story.

That attention made the pixels worth buying. New purchases made the page more interesting. The growing value of the page attracted more coverage, which brought more visitors and made the remaining pixels more valuable.

Outbid uses a similar loop without the Million Dollar Homepage's permanent inventory.

Tew sold a fixed piece of the page. Outbid sells rank. A desirable position can be taken away, which gives previous bidders a reason to return and gives spectators something new to watch.

The result looks less like a billboard and a little more like a game.

That distinction opens up a much larger design space than the current wave of nearly identical .lol leaderboards might suggest. A market for visibility could be organized around territory, tournaments, fictional stocks, voting, prediction, timed ownership or dozens of other mechanics. Those models are speculative, and most would probably fail. The useful idea underneath them is simpler: the process of allocating promotional attention can itself give people a reason to visit.

Outbid is a particularly good demonstration because the internet immediately ran an accidental experiment on it.

Hundreds of people copied the mechanism.

Wilke has said he built Outbid in roughly three hours using Grok in Cursor. More recently, he said the project was entirely built with Grok 4.6 in Cursor. That makes Outbid unusually representative of the current software environment: a small product can go from idea to public deployment in a single evening with substantial help from a coding model.

It would be a mistake to conclude that all 409 boards tracked by outoutbid.lol were AI-generated. The tracker cannot tell us how their source code was written, and domain registration timestamps are not development logs.

What it can tell us is how quickly the format propagated.

The tracker places the first known clone domain registration about 12.7 hours after Outbid appeared. Twenty-two arrived within the first day. By 48 hours, there were 189. At the busiest point, 20 domains were registered during a single hour.

That sort of copying was possible before generative AI. The Million Dollar Homepage produced plenty of imitators in 2005. What has changed is the amount of implementation work a person can hand off to a coding agent.

A pay-to-rank board is not a technically extraordinary application. It needs a frontend, a database, payment handling, ranking logic and deployment. None of those components are novel. Modern frameworks already provide much of the plumbing, and coding agents can assemble familiar pieces quickly.

That makes the clone data especially revealing.

Outoutbid's August 25 snapshot found a median of $17 among measurable clones. Eighty-three percent had taken less than $100. The tracker also warns that these figures are based on totals published by the boards themselves rather than audited financial records, so they should be read as reported payment volume, not independently verified revenue.

Still, the distribution is hard to miss.

Rebuilding the mechanism was cheap.

Rebuilding the crowd around it was not.

That is the part of the Outbid story that matters beyond Outbid.

AI coding tools are increasing the number of people who can turn an idea into functioning software and decreasing the time required for many ordinary web-development tasks. That does not make engineering trivial, nor does it mean AI improves productivity in every software project. Complex systems still require architecture, debugging, operations and judgment.

For small web products, though, implementation is becoming less effective as a moat.

A founder can copy a leaderboard. They can copy its layout, its bidding rules and its payment flow. They cannot ask a coding agent to generate the million people who saw the original.

As more software becomes buildable, more products end up competing for the same people.

That is where an old economic idea starts feeling current again.

There is another reason Outbid belongs in a conversation about AI and cybersecurity.

Its official About page says the initial traffic surge broke its analytics provider. That is a fairly benign failure as viral failures go, but it shows how quickly the operating assumptions behind a small project can become irrelevant.

A side project can be written for hundreds of visitors and wake up to hundreds of thousands.

At the time of publication, I could not find credible evidence that Outbid itself had suffered a distributed denial-of-service attack. It would be irresponsible to imply otherwise.

Its most obvious historical predecessor was not so lucky.

In January 2006, shortly after The Million Dollar Homepage completed its million-dollar run, attackers demanded money from Alex Tew and launched a distributed denial-of-service attack when he refused to pay. A DDoS attack overwhelms a service with traffic from many systems so legitimate visitors can no longer reach it. Tew publicly confirmed that his site had become extremely slow or unavailable and that the FBI was investigating the extortion attempt.

The coincidence is useful.

The Million Dollar Homepage was valuable because it became visible. Its visibility also made it a more attractive target.

Twenty years later, a solo developer can reach that stage much faster.

Modern hosting makes sudden growth easier to absorb than it was in 2006, but it also introduces new failure modes. An application that automatically scales in response to traffic may stay online while generating an unpleasant infrastructure bill. A payment-heavy product needs to think about fraud, webhook replay and duplicate transactions. A public submission system has to consider malicious links and abuse. Dependencies such as analytics, payment processors and databases can fail before the application server itself does.

For an indie product with any realistic chance of going viral, some basic preparation is no longer excessive: put a content delivery network or comparable edge layer in front of the origin, use managed DDoS protection where appropriate, rate-limit expensive endpoints, make payment operations idempotent, configure spending alerts and know which third-party service will hit its limit first.

None of that guarantees resilience. It does make success less likely to become its own outage.

AI has made it possible for one person to create and launch a surprisingly capable web product in an afternoon. It has not removed the operational work that begins when strangers start using it.

That gap is going to produce more incidents.

The current pay-to-rank frenzy will cool off.

Outoutbid's data already makes it hard to believe that hundreds of nearly identical boards have durable businesses behind them. Twenty-six measurable clones had not recorded a payment at all in its latest snapshot, and most remained below $100.

The leaderboard format does not need to survive for the broader idea to matter.

Outbid took something normally hidden inside advertising systems, the competition over who gets seen, and made it visible enough to become entertainment. Once that idea is separated from the specific interface, there are plenty of other ways to experiment with it.

Some will look like games. Some will resemble markets. Others may use voting, tournaments, territory ownership or limited-time competitions. The useful versions will need something the weakest Outbid clones do not have: a reason for people who are not buying promotion to visit anyway.

That is the difficult part.

An attention market with advertisers and no spectators is just a room full of people trying to sell things to each other.

Outbid briefly solved that problem by becoming the story on X. The escalating bids, screenshots, founder updates and rapid succession of copycats gave people something to follow even if they had no intention of buying a position.

The next successful version will need its own reason to deserve an audience.

That brings us back to the $17 median.

Four hundred nine boards can reproduce a software format in less than a week. The original can still hold the cultural center of the trend.

Generative AI makes that imbalance more important, not less. When the cost of producing another application falls, the valuable thing may increasingly be the community, reputation, distribution channel or strange little internet phenomenon around it.

Herbert Simon's observation survived mainframes, television, the web, search engines and social media. It will survive generative AI too.

We can make far more things than anyone has time to look at.

Outbid just put a price on the fight over which one gets seen.

Herbert A. Simon, Designing Organizations for an Information-Rich World, 1971. Simon's original discussion of attention scarcity in an information-rich environment.

David S. Evans, The Economics of Attention Markets, first posted October 2, 2017; revised April 2020. Economic treatment of platforms that acquire consumer attention and sell advertisers access to it.

Jonathan Wilke / Outbid.lol, About. Official launch date, visitor counter, highest-bid figure and account of the analytics-provider failure. Accessed August 25, 2026.

Outoutbid.lol, The outbid.lol clone wave, counted, August 25, 2026 snapshot. Live dataset and methodology for clone count, registration timing and self-published payment totals.

CBS News, Sean Alfano, Pixel Perfect, January 11, 2006. Contemporary reporting on Alex Tew and The Million Dollar Homepage.

The Guardian, Bobbie Johnson, Million dollar homepage under hack attack, January 18, 2006. Contemporary coverage quoting Tew's confirmation of the DDoS attack and extortion attempt.

Relve, Neelam Khan, A Guy Built a $1 Pay-to-Rank Board in 3 Hours and Broke the Internet, August 24, 2026. Secondary reporting on Wilke's use of Grok in Cursor and the initial build.