More effective random algorithm; q=n^(1−δ) vs. q=n^(1/2−δ)

This would appear to be an update of the paper that lots of people have written about. The main discussion is here:

https://news.ycombinator.com/item?id=11719543 (157 comments)

Other submissions, some with a comment or two, here:

https://news.ycombinator.com/item?id=11754074

https://news.ycombinator.com/item?id=11725991

https://news.ycombinator.com/item?id=11723319

https://news.ycombinator.com/item?id=11715677

https://news.ycombinator.com/item?id=11713454

https://news.ycombinator.com/item?id=11712636

https://news.ycombinator.com/item?id=11709112