Zero Reviews
Trust between strangers has always been bought with small, cheap first jobs. AI is absorbing those jobs. Who should now pay for the first chance?
Trust between strangers is usually bought in small amounts. You lend someone a book before you lend them a car, and you order one logo from a designer on the other side of the world before you hand over your whole brand. The first transaction is always the expensive one, though not in money. Someone has to go first without evidence, and whoever goes first carries the risk that the other turns out to be careless or dishonest. The question nobody likes to ask out loud is who should carry that risk.
For years, open freelance marketplaces answered it quietly. The newcomer carried it. A person with a laptop and a skill and no reviews would offer work at a price low enough that a buyer abroad could afford to gamble on a stranger. If it went well, the buyer came back and the second order was bigger. Trust across a border got built one cheap job at a time, and the cheapness was the newcomer's down payment.
That arrangement is now coming apart. Fiverr reported last week that its active buyers had fallen 21.9% in a year, to 2.7 million, and according to Calcalist the losses were concentrated in lower-complexity work such as basic copywriting, simple graphic design and entry-level programming. Spending per remaining buyer rose, and large technical projects grew. Micha Kaufman, the founder, said the results "reflect a market that is changing faster than expected, driven by rapid AI adoption." The small, cheap first job, the one a buyer used to risk on an unknown, is the job a general-purpose model now does in seconds.
So the down payment no longer buys anything. A buyer who once needed a stranger for the small task now needs nobody, and the large task goes to someone with a record. The person with no record has lost the only currency she had for acquiring one.
One answer says that's simply how it goes, and it has a real moral core. Nobody owes a stranger a chance. A buyer's money is her own, and choosing a cheaper, faster tool is an exercise of a freedom anyone would claim for themselves. Trust is valuable precisely because it has to be earned; a system that forces people to trust newcomers they otherwise wouldn't, through subsidies or quotas or algorithmic nudges, cheapens the signal for everyone, including the newcomers who do earn it. And some of the vanished work was poorly paid and taught little. If it goes, perhaps people are pushed toward work that pays and teaches more.
I accept most of this about individual buyers. A small business owner choosing a model over a freelancer has done nothing wrong, and I wouldn't want anyone lecturing her otherwise.
The leap comes in the next step, from "no individual owes the newcomer anything" to "nobody does." The platforms built very large businesses on that cheap first rung. Their whole value lay in turning strangers into trusted counterparts at scale, and the raw material was newcomers willing to underprice themselves to get started. They benefited from a ladder whose bottom rung someone else paid for. Whoever profited from that rung, and from the technology now removing it, takes on at least some duty to help build the next one, perhaps through verified skills tests that stand in for a first review, or through visible guarantees that let a buyer risk a newcomer without risking herself. The cost of the first trust has to land somewhere. I'd rather it landed on the parties who gained most from the old arrangement than on the person with the least to spare.
There is a catch in my own proposal. A platform that subsidizes first chances is still choosing who gets them, and its choices will follow its own incentives. The newcomer might end up trading one gatekeeper she could get past by being cheap for another she can't get past at all. That risk is real, and it is still better than a ladder with no bottom rung, because a gatekeeper can be pressured, audited and compared, and an empty space cannot.
Somewhere tonight a seller is finishing a profile: six sample logos, a starting price, a short paragraph in careful second-language English, and under the name, "0 reviews." For a long time that line was a starting point. If it is to stay one, her first customer can no longer be a stranger gambling twenty dollars. It has to be the marketplace that spent a decade collecting those gambles, putting its own name behind her before anyone else will.
— The Andes