I put New Scientist in the same category with OMNI magazine. Scientific American is so much better. I agree that NS publishes peoples' outlandish ideas, and do not put a lot of stock in their publications. I certainly would not publish with them.
I would hope that the rate of falsified work is very low, less than a few percent, and actually I would rather people did not falsify work - it will be discovered sooner or later, especially if someone tries to repeat the work, apply it, or test it.
I think there maybe a greater degree of misinterpretation, or faulty extrapolation. At least, that has been my experience. And I have seen highly detailed technical reports that contained faulty data and conclusions, e.g. from simple mistakes like a faulty calibration in the initial phase of the work, which was not discovered until after the final report was issued!
Yeah, and I wish they would stop doing that!