It’s not my idea, but rather Yahoo’s. The quicklinks can be pretty useful.

Great question regarding Twitter and Facebook. The nice thing about quicklinks is that they add a lot to a search results page without taking up a lot of space – and they are useful.

It would be interesting to see a study of the language used in tweets or Facebook status updates that introduced links, to see how well they described the links being pointed towards. I’d like to see that.

Thanks. I think it’s a good idea using tags from sources like that as well.

Agree with you on discounting words that might not be found on the page. The “language model” for the pages mentioned in the patent filing would be where something like that might be prevented. They would look at the content of the page, as well as meta information like the page title, and other links pointing to the page, and queries that the page was being found for (which where clicked upon), to create that language model. Chances are, a “toread” tag wouldn’t score very high as a potential label for a quick link because it wouldn’t match up well with the language model for the page.

The idea of using tags to identify related searches might not be a bad idea either, especially if it were used in combination with something like the language model mentioned.

As for tags like ‘toread’ I’d think that search engines would discount words that are not found on the page itself (preventing “miserable failure” style Google bomb).