A recent BBC event aimed at solving problems faced by women in tech flew Amy Cuddy in to give design and professional advice to the participants. The BBC story repeatedly celebrated Cuddy for giving the "second most watched TED talk ever". It also talked up the "Wonder Woman power pose" as a way to change your life.
Notably not present at the event were Cuddy's co-authors Carney and Yap, who have been vastly more receptive to the replication question, and less quick to promote uncertain results in non-academic channels.
Scientific falsehoods are almost always more impressive than the truth - they aren't constrained to realistic effects. At a certain point it starts to look like we're actively rewarding people for doing bad research and then rewarding them further for refusing to correct course.
Science is about finding approximations of truth that are least influenced by the local context.
The whims of a particular scientist's ego (especially if you are that scientist!) make for a rather nasty contextual influence.
Of course, this is a difficult ask, but constantly striving towards this ideal is absolutely essential.
But you also can't prevent people from learning about how the world works: the rewards and penalties that will affect their careers.
Even if you could, teaching students a myth about what a career in science is like would basically be malpractice.
But does this make sense? Can a scientist survive without doing this? Can a scientist get very far without results right now?
Is this a problem of the scientist, or perhaps the social/economic expectation?
New results are significantly more difficult and expensive to achieve in modern science, as the easier stuff has already been conquered in centuries and decades past.
The harder problems that science now works on means that distinguishing between actual results and incorrect conclusions is also much more difficult.
The growth of science as a percentage of the population and economy has also finally slowed down a few decades ago, after centuries of rocket-like growth. It is now a maturing enterprise, instead of a high-growth startup.
This means increased competition for resources and mindshare from those wanting to do science. Not surprisingly, all proxy measures and heuristics for evaluating scientific success are thus more likely to be abused.
I don't think we have fully adapted to these fundamental changes.
Disclaimer: I'm not a historian or researcher, but I've heard aspects of the above from a variety of sources, and it makes a lot of sense to me.
Of course, that's much easier for people in STEM than it is for social scientists.
Talking about the stakes is in direct conflict with the scientific ideal of focusing on the science without talking about its effects on the scientists involved.
For example, from Gelman's response: "I just wanted to emphasize that there’s no reason her career, or even her famous Ted talk, has to rely on a particular intriguing idea (on there being a large and predictable effect of a certain pose) that happened not to work out." Well, except that maybe she wouldn't have gotten a Ted talk without it?
He's saying, hey, this paper shouldn't have been as important as it was. And according to the scientific ideal, we shouldn't get too attached to any particular results. But in practice, putting bad results aside sometimes requires a significant change of identity. Forgetting about that, even for a moment, is always going to come across as tin-eared.
The scientific ideal is more compatible with low stakes than high stakes. But talking about the game as if it's low-stakes (or ignoring the stakes altogether) when actually it's high-stakes is always going to make discussions focused on the science alone come across as unsympathetic.