Yes, it's backwards. The most common way this is quantified is using Fraction of Attributable Risk (FAR) [0]. Even if the mean of the temperature distribution is just shifted slightly upward with no change in shape, the higher the threshold, the larger the fraction of risk that is attributable to the change in mean. This holds for all but pathological distributions, and doesn't even require increased variability or anything like that.
Throwaway as well because I've worked with Cliff Mass and am not interested in being at the receiving end of the retaliation that he's known for doling out.
Cliff is kind of a pariah among meteorologists and climate scientists. He _loves_ that fact, and revels in this idea that he's the lone genius who got it right while the rest of the community got it wrong on climate change. But he's not viewed as someone who makes credible statements, particularly as they pertain to questions about climate.
Many of his blog posts are deliberately misleading or contain incorrect information. Others I'd categorize as "You're not wrong... you're just an asshole." He has a decent understanding of some of the meteorological phenomena that are unique to the Puget Sound region, but even there I take every word he writes with a grain of salt. Sure, read his blog to understand how the snow is going to be this weekend at Crystal. But I'd leave it at that.
[0] Figure 1.1f here: https://www.ipcc.ch/site/assets/uploads/sites/3/2019/11/05_S...