Thus biasing submissions' language and content towards those most palatable to those open-weight models, creating further incentives to generate articles wholesale with them, and incentives against publishing null findings