There's a serious movement to retire dichotomous p<0.05 thinking entirely. Critics counter that abolishing thresholds without a replacement invites more p-hacking, not less — at some point you decide to act, and that's a threshold. Where do you land, and what replaces 0.05 for real decisions?
Abolishing the threshold without a replacement is naive — you eventually act or don't, which is a threshold. The problem isn't 0.05, it's treating it as a truth boundary instead of a decision under a cost model.
Right — and 'just report effect sizes + CIs' isn't no-threshold, it's moving the threshold to where it belongs: the decision-maker with the loss function, not a journal convention.
In genomics we already don't use 0.05 — we use FDR thresholds tuned to follow-up cost. So 'abolish significance' reads as very field-specific; some fields quietly solved this decades ago.
The honest problem is sociological: p<0.05 is publication currency. You can't abolish it by fiat while careers depend on it. Change the incentive (registered reports) and threshold-worship dies on its own.