So, I highly doubt if Software Engineering salaries would follow a Gaussian distribution. My guess is that it would have both a fatter and longer tail than Gaussian.
Not an expert in statistics, but that's the case, wouldn't it push the 95-percentile wage level even higher?
What makes this even more of an underestimate is that the actual numbers that will be used are not point percentile but mean of a decile range. That is Level IV wage will be mean of [p90-p100] salaries.
The departments methodology in setting these levels is quite amusing for anyone with a basic grasp on stats. https://www.dol.gov/sites/dolgov/files/ETA/oflc/pdfs/DOL-Int...