The pressure to get positive results is just too high to be compatible with good science. When you have people's whole life and career on the line all the time (publish or perish) then people are going to do what they have to stay employed.
Like Gresham's law bad science drives out good, because it is much faster to do bad science than good. Those that try to maintain quality can't pump out as many papers as the bad and so lose on the grant treadmill.
Any solution that does not address the incentives is doomed to fail. Not fixing this problem will kill science.
And that pressure to get positive results is extra high right at the beginning of scientists careers, when they should be learning best practices! They do a big project for their doctorate and if it turns out poorly, they can be screwed. There's their chance to move on with their careers flying out the window. Maybe the already had their next gig lined up, but it's time to put that on hold :( Or, maybe if we just make an extra assumption, or twist and turn the data just this way, using this model and these covariates... Oh! Here's something significant! Here's your PhD welcome to science.
The pressure doesn't stop at any stage in your scientific career. It is bad at the PhD stage (get results or no PhD), at the postdoc stage (get results or no tenure track position), at the tenure track stage (get results or no tenure), tenured (get results or no grants).
About the only point you can actually slow down enough to care about quality is at the emeritus stage, but even then if you don't produce your lab will be moved into a broom cupboard.
One of the things I make my students do is to try and design projects so that a myriad of answers are all potentially interesting, such that the likelihood of getting a result is high, but the particulars of the research are responsible for what that result is.
Yes I used to do something similar. All my students had two projects - one was high risk, but if it succeeded glory, the other boring, but no matter the result, publishable.
Yep it is disgusting. The end result is going to be science is seen the same as the Tour de France. Unlike the Tour science is not something we can live without.
The destruction of the reputation of science is probably the most dangerous thing going on in our society right now. If we lose science we have nothing.
The incentives for research absolutely needs to change, and I'm really happy that it looks like more and more people are both speaking up about it, and starting initiatives to try to improve.
We surely don't know the answer yet, but pre-registration of experiments/methods to get a guaranteed publication is one quite interesting.
But there are others, and I applaud any initiative in these areas, because it's at it core about the hardest problem of all:
How do we overcome bias in a large system under a lot of financial pressure?
It's such a worthwhile issue to pursue, but also usually quite thankless, so anyone fighting for it is doing something great even if it's in seemingly small ways!
What kind of failures should we celebrate? There are two kinds that I can see: a failure to produce a result because of a lack of skill or knowledge and a failure to produce a result because of a demonstrable or provable reason. The last one is actually a success because that kind of failure produces new knowledge. But we have a culture in far too many fields where this is seen as not publishable.
I wouldn't go as far as to say failure is celebrated, but it should be OK to fail. Your career should not be over because you tried something hard and it didn't pan out.
I almost think we should have a magic card you get to play once in your career that resets your track record. This way you could take risk without totally destroying your career.
Like Gresham's law bad science drives out good, because it is much faster to do bad science than good. Those that try to maintain quality can't pump out as many papers as the bad and so lose on the grant treadmill.
Any solution that does not address the incentives is doomed to fail. Not fixing this problem will kill science.