It's interesting to hear, mostly because I've never told authors to claim more (and plenty of times to please I beg of thee claim less).

I just meant this is semantics when you consider other limitations of arriving at this estimate... And tbf, I actually don't have major qualms with designating something fraud even if intentions can't be proven. As editor, I'm interested in reliability of content & intentions don't matter for this

It is because often there are ways to do this appropriately or properly. With a lot of this type of assessments this is what scares me the most: how much language we use matters. Which is probably true of everything, but hits different when you discuss crises of the entire facet of the society.

...Because it is easy to think that question "have you ever manipulated data?" is pretty clear, but actually it could mean different things to different people. I mean, there is a reason retractions always mention *inappropriate* manipulation of images...

I ran out of steam a bit (for obvious reasons), but he also covers previous surveys of researchers asked if they ever massaged data etc. and how these are likely under estimates. And here too I couldn't help but wonder - are they? ...

Well, there is a formal definition of this at the start of the preprint, but as I pettily pointed out, but this definition most of these numbers would not be estimates of fabricated or falsified research, so...

Ha. Well James says that this is lower bound, and that most scientific sleuths he surveyed thought it is higher. Do you think it is low though? Perhaps for once I am the least cynical person in the room.

Reading notes: if you got thus far, thanks and don't hold it against me. But also go and listen to that Everything Hertz episode I mentioned to hear the authors speak of this work on his own terms:

…even if it is a crisis of something. And even if 15% of papers were filled with fake data, it still doesn't mean that 15% of science is fake. Science - as a somewhat systematic system of generating knowledge - will, I believe, do just fine. 41/41

And so yes, we shouldn't ignore possibly fraudulent research - but should also not treat it in isolation from other problems of academia, nor present as a new issue. But also, even generously assuming that 1 in 7 is close to real value, we shouldn't be framing this as a crisis of science… /40