Most of the work hasn't been attempted to be reproduced. Those statistics are skewed by really high profile flashy "high impact" publications. There are reproducibility issues, but you must also keep in mind those stats are created by polling a specific subset of scientists.
I have yet to be unable to reproduce ANY experiment that I have tried from the literature. Everything has been appropriate.
The majority of it is not bullshit. Some of it is, exactly how much isn't known. Certainly not a majority.
Can't take your word mate, nor your specific field. Evidence shows otherwise.
Except it really doesn't.
I will make another post about it then. This time it will be more technical as well. Let's step up this conversation.