The other week @jdevalk asked if you can use nofollow on html link tags. I didn't know if that would work, so i spent a disgusting amount of time on running different systems' unittests to figure out. Following tweets explain why it took so long and what the answer is.
First, the answer: yes, you can use nofollow on <link... tags in the form of rel="alternate nofollow", and that will prevent Google from using the link from the href attribute. If you don't specify a nofollow, the URL from href will be extracted as a weightless outlink.
Second, we have many systems that can do URL extraction that may be used for discovery. For instance, Googlebot itself can do some minimal extraction, rendering can do advanced extraction, HTML parser can do extraction. That's why it took me over a week to have an answer.
Third, the sitemaps processing system i still have to test (*), cos it does find the link tags, too, but for that particular system i suspect a robotstxt rule would be better.
* - it was written by someone who's been a c++ arbiter for years; i barely understand their code.
* - it was written by someone who's been a c++ arbiter for years; i barely understand their code.
