They had a good bound of the long-tail distribution when x>=3/5=0.6. Now someone extended that result to x>=13/25=0.52. (The long term objective is to prove a stronger version for x>1/2=.5.)
affected distribution is always x>3/4, both before and after
what's measured is upper bound on number of zeroes <y, relative to y
it was <y^(3/5), now it's <y^(13/25)
it says nothing about absence of zeroes, but the density result already affects prime distributions
I'm wondering if this result indicates we have a new method to eke out important signals from noise that otherwise get smoothed out.