Last week I posted a bunch of stats about 101 rejections in the age of Alice and Mayo. The data I presented suggested that, for most art units, Alice and Mayo seem to have had a relatively small impact on allowance rates and average number of rejections. I did, however, concede that my analysis could be flawed and I invited suggestions for improvement. A few took me up on that invitation. A few more took to the Interwebs to label me an “efficient infringer” lobbyist who was likely paid “tens of thousands of dollars” for spreading a “misleading narrative.” ( which was very flattering considering that these posts effectively cost me thousands of dollars in lost billable hours).
Turns out, those ready to string me up may have had a point (although, to be fair, they did not actually provide any constructive suggestions for how to use the data to prove their point). Awash in a sea of borderline-incomprehensible rants were claims that the data for certain art units had to be wrong based on firsthand experience. The fact that some of the same art units appeared in multiple totally-unrelated comments (that is, from different people in different forums/channels) made me realize a potentially glaring weakness in my analysis.
Coverage of the Office Action Data Set
Breaking down the Office Action data set by art unit requires that coverage of the data set is consistent across art units. The authors of the data only analyzed coverage of the data set by number of applications and filing year – not by art unit. If certain art units are underrepresented, obviously that could skew the results.
The heatmap below shows coverage of the data set by art unit (y axis) and filing year (x axis). The value in and color of each cell represents the percentage of public applications (as reflected in PAIR) in the office action data set. Darkest green corresponds to 100% coverage, darkest red corresponds to 0% coverage.
Coverage of the USPTO Office Actions Data Set by Art Unit and Filing Year
As you can see, the coverage starts to get pretty inconsistent for applications filed in 2015. And here is where I really have to give an acknowledgment to whoever “BP” is on the Patently-O comments who said “the situation in art units such as 2850s, 2860s and 2120s is bad, really bad.” Turns out those art units are pretty poorly represented in the data set. So Mr. or Mrs. BP, thanks for the constructive comment and let’s hope the USPTO updates the data set to we can what is going on in those art units (For those that really want to see the data set updated, maybe let the office of the chief economist know).
The office actions data set is a great start, but considering the poor coverage for 2015 and later, it is probably not very useful (yet) for assessing the impact of Alice on prosecution. That said, allowance rates in general are still at or near all time highs, so I am not totally eating my words 🙂 Actually, on that note, one of the common pieces of feedback from last week’s post is that allowance rates are high because filings are down in the most-affected art units. I’ll try to assess the merits of that claim in the next post.