This is why GET requests aren't supposed to have side-effects (linkage)
by at 11:54 AM
Google Web Accelerator preloads every link on a page, including logout/delete/etc. This actually isn't anything new in web-accelerator land, but apparently since Google is involved, people are suddenly treating the issue differently.
Comments
Yeah, it's annoying that people seem to worship it.
rel="nofollow" isn't as bad as a lot of people make it out to be. It may be a nonstandard extension, but it's entirely compatible both with browsers and with the spec, and, although it isn't a solution in and of itself, it certainly isn't the only step that Google's taking to improve search results. It isn't "putting the onus on websites for fixing their search heuristic", it's giving websites the option to tag links as "untrusted - don't consider this link for rankings". And, although comment spammers may not bother to check to see if links are marked as nofollow, it will make comment spamming less effective. Which is always a good thing.
Google's auto-link has also attracted a lot more attention than it deserves. What a lot of people don't seem to understand is that their links DON'T SHOW UP unless the user clicks a button on the toolbar, and, even then, they're visually distinct from the rest of the links on the page! It's a world apart from Microsoft's proposed linking, which would have automatically placed links into text that were indistinguishable from normally placed links.
The technically minded can come up with more useful 'accelerator' solutions, like running a local web and DNS cache, but setting up BIND and Squid isn't everyone's cup of joe.
I, for one, would gladly pay up to $150/month for even a 512k pipe (though the DSL service one town over goes as high as 4Mbit, and cable where available is up to 16), so I guess I'm not one of the whiners