My biggest 3C1R is 55cM, except for an endogamy case that's 64cM, so maybe this is a good example of the differences between testing companies - mine is with AncestryDNA. Looking at Blaine's data (not just his chart, which lumps all testing places together, uses means and 99th percentiles) I see that for 3C1R, AncestryDNA has a median of 33cM, but lies between 9cM and 100cM 90% of the time. For GEDmatch, it's 33cM, and between 8cM & 100cM. For FTDNA, it's 66cM, and between 30cM & 147cM.
I hear AncestryDNA decided to exclude some very common genes a few years ago, before I tested, so that's why they tend to be smaller. I'm not sure it makes sense that it would be HALF of FTDNA's, though. Probably, there aren't enough FTDNA data points for 3C1R.
Anyway, I see the same thing with my brother's results, and there are a few that I have and he doesn't, and vice versa. Sometimes the ones that are biggest for me are among the smallest for him, and vice versa.
But your first statement needs to be qualifies a little. For 3C and beyond (maybe even from 2C1R), the values start at practically zero, but the highest value you're likely to see drops. So in that sense, the variability drops.
But what you're thinking in terms of is perhaps the percentage difference vs the typical value. The parent/child values are probably within 10% of the nominal value (guessing), while for 5C it might vary plus or minus 100% from the nominal value.