My first reaction was "This person isn't in a hard science like physics or engineering. Maybe medicine?" Ayup--medicine.
Style can be annoying but lack of context is deadly.
"To our surprise, false positives were low (3%)." is not much of an improvement. "To our surprise, false positives were low (3%) as we expected closer to 10%." IS an improvement.
However, that statement in a hard science is going to cause people to start asking some questions: Is 3% actually low? Why was the expectation so high to start? Did you screw up your error bars? Is it really 3%? etc.
"It is difficult to find untainted samples." Because is missing. In addition, this comment is rarely germane. I, the reader, don't really care. That's your problem. Either find the samples to get to statistical power or get a different method. As the reader, I only care if you did something weird in order to sidestep the problem--that I want to know about.
All of the recommended things still retain "weaseliness" while trying to sound like they don't.
Finally, I would love to have his problem of merely cleaning up some weasel words when most students I know (even at the PhD level) still have trouble stringing together coherent arguments and understanding where the holes and weaknesses are.
FYI, the author used to be a programming languages theory researcher before changing fields to biomedicine. So I think you’re reading into the field a bit
CS doesn't really classify as a "hard science" either, to be fair.
All my physics professors, for example, were anal retentive about error bars--you had to put your error estimates in the notebook in ink before you took measurements. If something was "better than expected" you were about to have a painful journey figuring out why.
Style can be annoying but lack of context is deadly.
"To our surprise, false positives were low (3%)." is not much of an improvement. "To our surprise, false positives were low (3%) as we expected closer to 10%." IS an improvement.
However, that statement in a hard science is going to cause people to start asking some questions: Is 3% actually low? Why was the expectation so high to start? Did you screw up your error bars? Is it really 3%? etc.
"It is difficult to find untainted samples." Because is missing. In addition, this comment is rarely germane. I, the reader, don't really care. That's your problem. Either find the samples to get to statistical power or get a different method. As the reader, I only care if you did something weird in order to sidestep the problem--that I want to know about.
All of the recommended things still retain "weaseliness" while trying to sound like they don't.
Finally, I would love to have his problem of merely cleaning up some weasel words when most students I know (even at the PhD level) still have trouble stringing together coherent arguments and understanding where the holes and weaknesses are.