That is akin to not being able to accept the null hypothesis in statistics, as we assume innocence ant try to prove guilt (with the p-value of "a reasonable doubt"). While we should not explicitly disincentivize negative results, this doesn't mean that we shouldn't penalise p-hacking.