You are 100% correct. The problem is, even without the incentive of money and employment there's still the incentive of fame, and negative results simply won't

You are 100% correct. The problem is, even without the incentive of money and employment there's still the incentive of fame, and negative results simply won't get the same attention and impact as positive results. We can give people with negative results PhDs and publications, we can continue funding them. But at the end of the day, people remember and researchers cite the team who invented the LLM, not the many teams who tried and invented the less-successful models.