Over time, I've found that the total number of datasets on data.gov is not gradually getting bigger but sometimes smaller than before. e.g. the number is 190000+ in 2016-04 but now (2016-10) it's only 180000+. Why do the number of datasets seem to be shrinking?
Common reasons are because agencies have made updates to their data inventories and deleted duplicate records, or grouped similar datasets -- such as multi-year of the same data, into a collection, which counts as 1 in the dataset total. Dataset total is around 190K, total resources represented if you separate out collections is 7 million.