If you don't have a filtering mechanism moderators will burn out and more interesting projects may never be seen. If you've run a bunch of LLM agent sessions concurrently you run into the human review bottleneck problem. That's all this is on a larger scale.