I assume that this is a side effect of more dynamic crawling. A lot of the pages on these sites are created by a user search and then the site links to this term as a popular search or whatever and as the page is linked to then it can enter the index. (This was discussed recently)
If google can (does already?) start crawling and indexing urls like below then this will open the floodgates (already open?) for dynamically created pages to enter the index.
If you then take the q= result and dynamically change a few terms in <h> tags <b> tags etc I am sure it can get through the duplicate filter.