huertanix / haystac

Hayst.ac obfuscated web search.
http://www.hayst.ac
11 stars 0 forks source link

Duplicate Clickthroughs / Async Fail #23

Open huertanix opened 14 years ago

huertanix commented 14 years ago

Probably the biggest issue/bug in the whatisthisidonteven department is that when doing an obfuscated Google search, the searches themselves show up in the right randomized order in Google Search History, as do their clickthroughs, but a large portion of the clickthroughs show up again after the searches in a huge batch with no searches associated with them. If a clickthrough shows up twice, its a huge red flag for spotting artificially-generated clickthroughs.

The GM_xmlhttpRequest function is totally asynchronous. This is why both the search term generation and search functions are set up in such a Escher-tastic way that the function calls itself after its done doing stuff. The GM_xmlhttpRequest call for clickthrough generation doesn't do this, however. All clickthroughs are going through on tests with fixed numbers of searches and clickthroughs, so its not like the function is jumping ship before the page can load, although if somehow the page is being fetched twice (not sure how, this is all inside an anonymous function), it would explain why there are different session keys on the clickthrough URLs for the duplicates:

Earlier preceeding clickthrough with search associated: http://www.google.com/url?sa=t&source=web&ct=res&cd=16&ved=0CFwQmAEwDw&url=http%3A%2F%2Fwww.huffingtonpost.com%2F2010%2F06%2F03%2Frue-mcclanahan-dead-golde_n_599256.html&ei=iJAMTPPDNKHONO_wnLYE&usg=AFQjCNEgtsZ6nP3jYmK6GMxVUQDGK5G-1w&x2s5ulNiXXpRy6YNQ

Succeeding clickthrough URL w/o a search associated: https://www.google.com/history/url?url=http://www.huffingtonpost.com/2010/06/03/rue-mcclanahan-dead-golde_n_599256.html&ei=P5UMTJbOEZjSswLzx5TOCQ&sig2=69S92YDI7H7z93jB2Rqnfw&ct=g

This is probably the most puzzling problem I've ran into in this project. Help us, internets!

huertanix commented 14 years ago

Aw, poo. The second link wasn't copied into this issue correctly; That's the Google History redirect URL for the web search redirect URL because YO DAWG, I heard you like redirect URLs...

huertanix commented 14 years ago

No answer on Quora. I'll ask Vincent from TrackMeNot; He may have done some research in the area.