Is there any mechanism to prevent gathering entropy through links?

[sparhami]

Lets say I have a site called "tackcuriosity.com", which is a feed of items that users can click on. When the user loads the page, I grab n bits of entropy, which is under the budget.

When a user clicks on an item, I open a new page to "tаckcuriosity.com/?item=abc&session=foo" (in this example, 'a' is replaced with the unicode lookalike 'а' or any other such substitution in the domain, using subdomains, completely different domains), then grab n bits more entropy on the new domain, which I combine on a server with the previously recorded entropy using a linking session identifier in the URL.

Another example would be news sites that have a tap to continue reading button, which could also perform this trick. The page could load a new page with the article expanded, but use it as a chance to grab more entropy. Really any time a user needs to interact (or you can force them to interact), you can take a chance to grab more bits (with a poorer user experience, but that hasn't stopped these sorts of things so far), then tie together the identifying information.

If you have something like the portals proposal, I believe you could go to your intermediary domain, grab entropy, then use the portal to display content on the original domain to the user. Or maybe you can even use a portal to make the whole thing look seemless to the user (unless they share a limit like iframes).