The big idea: A cartel for freelancers

The freelance market is facing existential risk from AI competition. Is the solution cooperation?

Like everyone else, I’m a little anxious about the constant predictions that AI will reshape the economy as we know it. Whatever the long-term equilibrium is, there’s one part of the economy where the impact is already visible: the freelance market.

Freelancers are a strange professional class of designers, writers, copywriters, developers, and consultants that mostly sell output on a project-by-project basis. This project nature is what makes freelancing uniquely susceptible to economic disruption; there’s a natural endpoint for clients to make changes based on changing circumstances. Unlike employees, freelancers have to keep selling their work on a spot market.

Early evidence suggests the introduction of ChatGPT corresponded with a 2% decline in monthly contracts on Upwork and a 5% drop in earnings. Worse, the impact on earnings hit high-skill freelancers hardest. The causes go beyond ChatGPT adding competition — AI can create the facsimile of expertise, making it hard to tell who genuinely is an expert and who is just good at prompting.

“Form a union” some people might say. But freelancers lack protection under the National Labor Relations Act, which regulates union organizing. Even if they didn’t, unions need a unified counterparty to negotiate with, which freelancers definitionally do not have. That’s why the Freelancers Union is more of an advocacy/health care organization — that’s really the only part of the relationship they’re able to operate in.

Unions aren’t the only form of cooperation though. What if freelancers tried to increase their earnings by forming a cartel?

Cartel model 1: OPEC, or controlling supply

The purest example of raising prices through coordination is oil.

The post-WW2 oil industry was controlled by the “Seven Sisters,” a collection of the 7 largest oil companies that collectively controlled ~85% of the world’s production. The Seven Sisters were not shy about using their bargaining power to cut fees to the nations that hosted their oil wells.

Starting in the 1970s, OPEC emerged as a real counter-force. Their first real show of strength came in 1973, when OPEC was able to dramatically increase oil prices from $3 a barrel to $12 by coordinating production cuts. But when everyone else reduces supply and prices increase, any individual member has an incentive to overproduce and cash in.

The responsibility for enforcing discipline falls to the swing producer, an entity holding enough spare capacity that they can influence prices at will. For OPEC, this is Saudi Arabia. When other countries cheat — or new entrants threaten OPEC, like US shale producers in the 2010s — Saudi Arabia can punish them by ramping up production to crash oil prices.

Freelancing is basically millions of small, repeated auctions for output, at a new price each time. Imagine if freelancers adopted a similar model, intentionally withholding work for projects to constrict supply and drive up prices. With fewer alternatives, these freelancers would be able to wring higher payments in response to collapsing supply.

It probably wouldn’t work. Even with relatively niche functions, there are a LOT of people you’d need to coordinate, most of whom would be happy to defect. Any real organization would be broken by offshore freelancers that are willing to provide lots of output at lower rates. Given the fragmentation, the coordination problems are impossible to solve.

Even if they could, the modern freelance market has a swing producer that lives outside of the cartel: AI. An LLM is happy to produce unending good-enough output at a low price. With those economics, even an imperfect AI substitute can drive down prices significantly.

For a copywriter, ChatGPT is the equivalent of the swing producer operating at max capacity, always, forever. Any attempt to withhold services just accelerates their replacement by Saudi ArAIbia.

So while FOPEC — the Freelance Organization for Pricing Equity and Compensation — is a great acronym, it may not be the right model for a freelancer cartel.

Cartel model 2: Licensing, or controlling permission

If you can’t control supply, what about just making it illegal to buy from anyone else?

The decentralized and fragmented nature of the freelance market makes it hard to imagine the OPEC model succeeding. But there is a model that’s been highly successful in protecting professionals with unique skillsets: professional licensing.¹

Limiting who is legally allowed to do a job is the ultimate hack for decentralized professional organizations. Professional organizations have already mobilized against AI competition in advanced knowledge work — the American Bar Association (ABA) is working on this for lawyers, pushing a bill in New York that would bar AI from practicing law. AI can’t act as a swing producer if they’re not legally allowed to produce.

Beyond strict bans, professionals can control access through market structure; the American Medical Association (AMA) has long lobbied to limit how many new doctors can enter the market, advocating for restrictions on medical schools, residency slots, and the scope of non-doctor care.

So is the path forward for freelance copywriters to make it illegal to write instruction manuals without a license? Licensing has crept downmarket from high-risk industries like medicine to lower-stakes professions like barbers and nail techs; I’d bet that some higher-skilled freelancers pull off lobbying for legal restrictions. I’m not sure what the copywriting equivalent of the bar exam is, but somebody will be able to develop it.

It’s a patchy solution, likely to (at best) exist at the state level with plenty of cheating using multi-state entities or just ignoring the rules. Most freelancers are remote anyway; at best you’ll see “freelancers from Colorado will not be considered without a license to do graphic design” on job postings.

Cartel model 3: A clearinghouse, or controlling trust

There’s another route that freelancers can take: capturing the platforms.

Platforms like Upwork, Toptal, and Fiverr need to manage a careful balancing act between keeping sufficient freelancers while bringing in businesses. Historically there’s been a love-hate relationship between platforms and freelancers; platforms want to ramp up fees, freelancers want access to work without paying a platform tax.

But if AI collapses freelancer earnings, the business model for two-sided marketplaces breaks down — you can’t collect meaningful fees on both sides if buyers are paying peanuts to an AI-augmented amateur. Preserving market value for verified human labor is the priority for both freelancers and platforms.

With a common enemy, it might be prudent to transform the market from atomized freelancers to a clearinghouse. As the sorting layer of the freelance market, platforms could become a trust clearinghouse: verifying freelancers, charging clients a premium for human-guaranteed work, and setting rating floors and quality standards across the market.

In exchange, clients get verified identity (this person is real), credentials and audits (this was made by a human), dispute resolutions (this person should get paid), and predictable rate norms (this person should get paid this much). The platform acts as both a licensing body and enforcement mechanism, controlling what work gets done and, to a lesser extent, how. Given increasing AI backlash, this kind of clearinghouse of labor might be genuinely valuable to some buyers.

A central certification agency to attest quality and self-regulate its members sounds sort of like an old historical idea: a guild.

Enforcing a guild

Once you build a guild-like system, the challenge becomes preventing defection among the participants. Freelance platforms actually have an ability to turn this into a more traditional cartel model; there aren’t that many freelancer platforms, and they can mutually agree to align standards to drive up fees (and their percentage cut).

What if the freelance platforms created a membership system that works across platforms to ensure enforcement: FOMO (Freelancer Organization for More Oversight)?

To get credentialed, freelancers would need to put up a meaningful bond: say, $1,000 to start alongside a share of future earnings. The fee both acts as a financial buy-in and a barrier to those that don’t intend to abide by the rules.

Certification through FOMO would give them membership, usable across platforms as a license to sell freelance work in their particular expertise. Members of the cartel would be required to adhere to specific rate guidelines, limit off-platform freelance work, and agree to AI use transparency. If a user breaks these rules, they receive a fine out of their bond that compensates everybody else in the network.

This still leaves the issue of off-platform transactions: what’s to stop somebody from doing extra hours somewhere else, or off-platform? Or worse, what if a new FOMO-free platform emerged to offer the cheapest freelancers to buyers?

Social enforcement is probably the key to stopping individual defection. A culture of distrust of outsiders, like any guild worth its salt, will help ostracize freelancers who work outside the guild. Updates to the LinkedIn project section become a trigger point for public shaming.

If that isn’t enough, the freelancer guild might start demanding increased surveillance of activity. Bank app monitoring and computer keylogs might need to become a standard part of the onboarding process for freelance marketplaces.

But that doesn’t stop new platforms from entering. And if new platforms show up, the response might be how lots of historical guilds reacted: direct action. FOMO-enrolled freelancers might start to resort to boycotts for businesses that use them, public shaming campaigns, and even mafia-esque threats. Much like the luddites, some graphic designers might be arrested for data center arson while fighting AI-augmented competition.

Once the cartel is formed, the economic logic starts to look less like an economic cartel and more like a criminal cartel, complete with muscle. Maybe it’s better if freelancers just have to deal with a fragmented market.

Official idea rating

1/5. While a cartel for freelancers would likely benefit the freelancers — especially since it’ll make low-wage, offshore competition much harder — it’s quite unfair for everyone else involved. Somebody trying to start freelancing will have new obstacles to entry. Enforcement is never going to be 100% fair, and the cultural changes seem unpleasant at best.

That said, with new technologies attacking both sides of the marketplace platforms will likely rethink their business models. It seems like there’s opportunity for platforms to find new ways to elevate freelancers as their incentives align. At a minimum, human-verified badges are almost certain to happen.

Freelancers will need to adapt to a world where impersonating their expertise will be easier than ever, and low cost competition will only become better and more available. But this probably looks more like specialization than coordination; the presence of AI splits the market too harshly to sustain real coordination.

Trying to build a cartel just feels like reinventing a guild with more steps.

1. Licensing has another benefit. Under the Sherman Antitrust Act, this kind of organization by independent contractors is straightforwardly illegal. Although the Biden FTC pushed back on this, it hasn’t been legally tested enough to validate. But professional organizations operate in a legal gray area, where these constraints are arguably voluntary.