Hmmm
Something in the researchers paper that isn't mentioned in the Reg's article:
"Weinberg et al. [57] demonstrated that these mitigations
are not enough—that web attackers can still creatively
leak history information. They used interactive tasks (e.g.,
CAPTCHAs) to trick users into disclosing history information,
inferred the color of links from screen reflections
in webcam images, and used re-paint events"
I have seen CAPTCHAs used on several deceptive and/or malicious sites.
My initial thought was this was used as a sort of "gateway' to protect their sites from being accessed by researchers using automated web scrapers.
I see that they may be using CAPTCHAs for other purposes as well.