Who would be the curators? If you open up curation to volunteers, it could be gamed by bad actors. However if you have too small a team, the results will be limited and biased. For example, favoring the English-speaking or tech-sphere web while ignoring large sections of the web with which the curators are unfamiliar.
Perhaps machine learning could help with scale - start with a human curated dataset, then train a model on it. However that could end up getting gamed too.