The Anticipat research database continues to comprehensively cover all legal grounds of rejection considered by the Patent Trial and Appeal Board (PTAB) for its ex parte appeals decision. This includes the more exotic like statutory subject matter Section 101 cases to the much more common issues like obviousness (Section 103).
Since July 25, 2016 to February 28, 2019, there have been around 24,448 obviousness decisions from the 29,102 total decisions meaning that nearly 84% of all appeals involve obviousness. Our observation is that obviousness is the most common ground of rejection to be decided at the Board. In this post, the data considered excludes decisions where the outcome involved a new ground of rejection based on obviousness as these typically only form a small fraction of the cases.
Of the 24,448 total decisions, 12,369 were affirmed, 8,386 were reversed, and 2354 were affirmed in part. Thus, the wholly reversed rate (for all claims in a case based on obviousness) was about 34%. The at least partially reversed rate (at least one claim in the case was found patentable) is about 44% (43.9%).
What is interesting to note is the breakdown by technology center in the USPTO. The technology centers contain the various art units to which patent cases are assigned based on how the claimed technology in each patent application is classified by the USPTO. Here is the summary of obviousness cases broken down by technology center based on cases decided between July 25, 2016 to February 28, 2019.
TC 1600 (biotech/pharma): Total decisions: 2333; 1907 total obviousness decisions. 1151 affirmed, 132 affirmed in part, 492 reversed. Wholly reversed rate is 25.8% and at least partially reversed rate is 32.7%.
1700 (chemical): Total decisions: 3950; 3633 total obviousness decisions. 2103 affirmed, 280 affirmed in part, 1041 reversed. Wholly reversed rate is 28.7% and at least partially reversed rate is 36%.
2100 (computer/electrical): Total decisions: 2976; 2575 total obviousness decisions. 1467 affirmed, 228 affirmed in part, and 739 reversed. Wholly reversed rate is 28.7% and at least partially reversed rate is 37.6%.
2400(computer/electrical): Total decisions: 3164; 2791 total obviousness decisions. 1531 affirmed, 276 affirmed in part, and 852 reversed. Wholly reversed rate is 30.5% and at least partially reversed rate is 40.4%.
2600(computer/electrical): Total decisions: 2760; 2458 total obviousness decisions. 1420 affirmed, 241 affirmed in part, and 677 reversed. Wholly reversed 27.5% and at least partially reversed rate is 37.3%.
2800(computer/electrical): Total appealed decisions: 1979; 1648 total obviousness decisions. 813 affirmed, 128 affirmed in part, 583 reversed. Wholly reversed 35.4% and at least partially reversed rate is 43%.
3600 (business methods/software): Total appealed decisions: 5941; 4212 total obviousness decisions. 1842 affirmed, 418 affirmed-in-part, 1755 reversed. Wholly reversed rate is 41.7% and at least partially reversed rate is 51.6%.
3700 (medical device/mechanical): Total decisions: 5734. 5022 total obviousness decisions. 1929 affirmed, 625 affirmed-in-part, 2195 reversed. Wholly reversed rate is 43.7%. At least partial reversal rate is 56.1%.
We note that the data has not shifted more than a couple of percentage points from our review of the data as reported in the past despite the increase in number of decisions. This indicates that, with respect to obviousness, by and large the Examiners (and the two supervisory Examiners involved in the Appeal Conferences) are very consistent in 1) picking the same kinds of bad decisions to take on appeal and the Board is 2) agreeing with applicants that this is the case at the same consistent rate. While some things like the particular cases being taken on appeal today have changed, the behavior by the USPTO has stayed markedly the same.
As a general observation, in the private sector a process producing 34% outright defective parts and 44% partially defective parts as determined by its own internal quality control process (the Board) would likely immediately regarded as being a low quality, unpredictable process (and would probably put a company out of business). At Anticipat, we believe that the ex parte appeals statistics are the closest and best end of line quality control indicator of the quality of the USPTO’s patent examination process. However, since these statistics are just an end of line quality control indicator, trying to change Examiner behavior solely using the ex parte appeals statistics will not solve the quality problem—reducing examiner variability will require the USPTO identifying meaningful inline statistical data monitors (pre-appeal) that could be used to reduce the variability in the examination process. The inline data monitors are what the USPTO could use to reduce the current levels of examination variability–and the end of line data (the ex parte appeals statistics) will show the effect too.
Could such a statistics-based process be implemented at the USPTO? Certainly–but it can only begin when the agency acknowledges that statistics like these reflect unmistakably on the quality and predictability of the patents currently granted. Tightening the USPTO’s distribution to result in lower ex parte appeals reversals will inevitably (through the operation of statistics itself) result in more predictable and better quality issued patents. Until then, the only predictable thing is that USPTO Examiners will continue to be reversed by the PTAB at these double digit rates for obviousness.