A series of recent decisions of the U.S. Supreme Court on patent law, culminating in the 2014 decision in Alice v. CLS Bank, Int'l., which prohibited software that generically applied an abstact idea that is not otherwise patentable, have made it dramatically more difficult to obtain patents, particularly the subset of patents called "business method patents" which include most software patents.
In January of 2004, only a little more than 2% of patent applications were rejected (on Section 101 grounds which governs what is patentable). By July of 2015, that percentage is about 15%.
Before Alice in July of 2014, about 31% of business method patent applications were rejected on Section 101 grounds (already a major increase from 2007 when the U.S. Supreme Court adopted a more expensive definition of "obviousness" under Section 101 for patent law purposes in KSR International v. Teleflex, and 2010 when the U.S. Supreme Court in Bilski v. Kappos articulated a new (and functionally more restrictive) legal standard for granting software patents (although not as restrictive as the federal circuit case it reviewed which is linked), while affirming that software patents could still be obtained). After Alice, 82% of business method patent applications were rejected.
Indeed, there have been at least 15 different major U.S. Supreme Court cases from 2004 to the present in which the U.S. Supreme Court has reversed pro-patent holder rulings from the Federal Circuit.
And, as I have previously noted, in patent infringement cases that are litigated until a result on patent validity is reached on the merits, roughly half of all litigated patents are found to be invalid (which isn't necessarily as shocking as it seems, because patent cases in which there is no strong patent invalidity defense routinely settle before a ruling on the merits of that issue is rendered).
This may have something to do with the fact that roughly 98% of patents filed with the PTO contain mistakes.
Tax strategy patents were also abolished, by statute, effective September 16, 2011 as part of the American Invents Act, which was the most significant overhaul of U.S. patent law since 1952. But, this law itself, outside of the area of tax strategy patents, didn't significantly change the scope of what kind of ideas could be patented.
On the whole, the rulings are a good thing, gradually and in a relatively nuanced manner, weeding out a lot of laxly granted patents (particularly in the business method category) that undermined innovation by creating legions of patent trolls whose tolls had to be paid to invent anything half innovative. Since patents last a couple of decades, many of the patents granted under the earlier lax standards remain in force, and this dramatic change in the legal standard for patentability over the last dozen years or so also helps explain why so many patents infringement lawsuits give rise to invalidity findings at trial.