A Mercor contributor described a project with fewer than 25 people and something that sounded almost luxurious: no fighting over tasks, a quiet channel, and colleagues who collaborated. The work also matched their specialty. In their account, this was the ideal project.
Then, in the same Reddit discussion, a different contributor described a different small internal team. They had been told the work could last a year. It ended after a little over two months.
Both experiences matter. A small team can offer something valuable while it exists. Its size alone cannot tell you how long the work will continue.
What has made an AI project worth your time—or made you reconsider it? Reply to this email with the question you want World Model to investigate next. If you use your private clawded.ai recommendations page, tell me what fits and what misses. Please leave out confidential project details.
Recent reader conversations keep bringing me back to three practical problems: contracts with little available work, standards that change between assignments, and expertise that does not quite match the tasks on offer. These are different problems. Applying for another project without identifying which one you have can reproduce the same frustration.
Meanwhile, a Reddit discussion about a project with thousands of participants describes crowded support channels, repeated questions, and access delays. Some contributors still enjoyed the work. Headcount, organization, and satisfaction were not moving together in any simple way.
These accounts are self-reported experiences, mostly from Mercor discussions, not a representative survey. They cannot establish that every additional trainer makes a project worse. They do suggest a useful question: has the project added people faster than it has added work, reviewers, and support?
My working principle is to judge the part of the project you actually experience. A large program can contain a well-supported specialist group. A small pilot can have unclear instructions and an empty queue. Slack membership is a poor substitute for knowing who reviews your work and how tasks reach you.
In a small collaborative team, your opportunity may be to help clarify the work. In a crowded production project, your priority is to make your judgment easy to review while protecting your time. Those conditions require different habits.
Below, paid readers get two practical playbooks, two messages to adapt for reviewers, and a first-week checklist for deciding how much attention a project deserves. Paid membership also includes clawded.ai: a private page of AI job recommendations informed by your background and feedback. It helps you explore alternatives; it cannot guarantee task supply inside a project.
