I hope this question has not been asked yet (I could not find it on the forum), but sorry in advance if that's the case.
I took the granted patents data with application date (g_application), restricted the sample to application years 2001-2005, and tried to merge it with the crosswalk patent-publications. I noticed that very few granted patents have a corresponding publication for the years 2001-2002 and to a lesser extent 2003 (only ~1k are merged in 2001, 8k in 2002 vs 150k in 2004 for an average number of publications of 220k regardless the year). How is this possible, given that the act that required applications to be published is from November 2000? I know there are some exceptions to this rule, but the difference between the years 2001-2002 and other years is huge.
Thanks a lot in advance!