The America Invents Act (AIA) was a great leap forward in the effort to improve patent quality in the US, including the creation of three new post patent issuance challenge procedures: post grant review (PGR), covered business method review (CBM), and inter partes review (IPR). The US Patent & Trademark Office (USPTO) has regularly reported some basic statistics related to the new challenge procedures. In an attempt to determine whether these new challenge procedures are serving their intended purpose of improving patent quality, I compiled and now report on additional statistics characterizing the parties and patents associated with completed challenge proceedings and correlated those characterizations to the nature of the outcomes.
The statistics reported by the USPTO focus on the petitions filed for the Patent and Trial Appeal Board (PTAB) to consider in determining whether to institute a challenge proceeding. Well over 1000 such petitions have been filed; the PTAB has instituted challenge proceedings in response to over 70% of the petitions. The USPTO has provided significantly fewer statistics with respect to the outcomes of the challenge proceedings that have been instituted. The outcomes are the end results of the challenge proceedings and, ultimately, should be the best indicator of what the proceedings are accomplishing.
As of May 31, 2014, 13 CBM proceedings and 128 IPR proceedings reached a resolution of some kind. The completed IPR proceedings are particularly valuable because the number is large enough to be statistically interesting yet small enough to make practical the manual extraction of characterization information which cannot be extracted electronically. Note that the completed proceedings include 53 final decisions, 14 adverse judgments requested by the parties, and 61 settlements. I included all of these, to the extent possible, in the statistics I present here. All of the data I present about post patent issuance challenges is from this pool of completed challenge proceedings – I made no attempt to characterize the larger data set of petitions to institute such proceedings, or proceedings for which no outcome has been reported.
Although it is too early to draw firm conclusions from the statistics, there are interesting trends that merit watching. A bulleted summary of findings and more detailed explanation follow.
Summary of Findings
- As of May 31, 2014, outcomes were achieved in 13 CBM proceedings and 128 IPR proceedings, ending in 53 final decisions, 14 adverse judgments requested by the parties, and 61 settlements.
- Seventy-six percent of patents that have undergone IPR proceedings are the subject of patent litigation (not necessarily involving the challenger in the IPR proceeding).
- Twenty-three percent of patents that have undergone IPR proceedings had been transacted (sold/assigned) after patent issuance and outside of merger and acquisition activity,
- PAE patents constitute 44% of those undergoing challenge by way of IPR, even though PAE’s only own approximately 2% of all patents.
- While 37% of all patents are software patents, only 30% of patents that have undergone IPR challenge are software patents.
- The PTAB found at least one claim invalid in 87%-95% of the cases overall. For PAE patents, the PTAB found at least one claim invalid in just 79% of the patents. The PTAB found at least one claim invalid in 100% of patents not owned by PAEs.
- For CBM, the PTAB has found at least one claim invalid in 100% of the completed cases.
- While at least one claim in 60% of patents identified in petitions to institute CBM proceedings are challenged on 35 USC 101 grounds and at least one claim in nearly 40% of patents identified in petitions to institute CBM proceedings are challenged on 35 USC 112 grounds, the PTAB only found at least one claim invalid on 35 USC 101 grounds in 27% of patents in completed CBM proceedings and found no claims invalid on 35 USC 112 grounds. In other words, in 73% of the completed CBM cases, the PTAB found at least one claim invalid on 35 USC 102 or 103 grounds.
Parties and Patents in Completed IPR Proceedings
The rise of patent assertion entities (PAEs) and the premise that PAE patents are of low quality inspired many of the provisions of the AIA, including the new post patent issuance challenge procedures. According to Patent Freedom, approximately 2% of active (unexpired) patents are owned by patent assertion entities. My review indicated that 44% of patentees in IPR proceedings (those owning the patents undergoing challenge by way of IPR) are PAEs. I defined a PAE as an entity with the primary business purpose of generating revenue from patents. Even with my conservative definition of PAE, and allowing for differences between my definition and that used by Patent Freedom, the difference between 44% and 2% can only be described as startling. The incidence of patents owned by PAEs that undergo IPR challenge appears to be disproportionately more than the incidence of all patents that are owned by PAEs – and substantially so. Those that argue PAE patents are of low patent quality should welcome this finding. Note later reference to the correlation between patents that are both challenged and litigated – PAE patents may be challenged at a higher because PAEs disproportionately participate in patent litigation.
More recently, some have argued patents relating to software are of low quality. Here again definitions vary. Methods can be implemented in hardware or software and patent claims often make no such distinction. The USPTO has reported statistics breaking down patents into the simple categories of electrical/computer, mechanical, chemical, bio/pharma, and design. This breakdown is not useful for the purpose of consistently identifying software patents. A slightly better USPTO technology breakdown is in the form of the patent examination Technology Centers. I provide two pie charts that deploy this breakdown. Chart 1 shows the breakdown of all active patents by Tech Center. Chart 2 shows the breakdown by Tech Center of the patents that have undergone IPR proceedings. I also provide a list of the Tech Centers and their respective technologies. Unfortunately, software patents are not exclusive to any single Tech Center, nor do they account for the entirety of any single Tech Center. Nevertheless, the portion of patents in computer related technologies – and certain others – in IPR proceedings is slightly elevated relative to the same portion of all patents. Whether the same elevation exists for software patents is unclear.
Looking to other sources of information about software patents, the report of the Government Accountability Office estimated (generously in my opinion) that about half of all patents are software patents. Unfortunately, the GAO’s methodology for determining which patents are software patents is not transparent and cannot be described here. However, IBM has developed a search query for identifying software patents. The query is based on keywords, USPTO classes, and other factors. Using the query, I determined that 37% of all patents are software patents. And using that same query, I determined that only 30% of patents that have undergone IPR challenge are software patents. As best as I can determine, software patents are under-represented among those patents in IPR proceedings.
I also reviewed patents in IPR proceedings against certain indicia of high or low value. Of most interest were the results for patents with indicia of high value. I found that 76% of patents that have undergone IPR proceedings are the subject of litigation (not necessarily involving the challenger in the IPR proceeding). I also found that 23% of patents that have undergone IPR proceedings had been transacted (sold/assigned) after patent issuance and outside of merger and acquisition activity, despite data from Patent Freedom indicating that fewer than 5% of all patents are transacted in this way. These findings indicate that challenged patents are of relatively high value.
Outcomes of Completed IPR Proceedings
In addition to characterizing the parties and patents in completed challenge proceedings, I reviewed the outcomes of those proceedings and attempted to correlate those outcomes to the parties and patents in the proceedings. The USPTO indicates 26% of claims for which IPR proceedings are actually instituted are eventually found invalid by the PTAB. In addition, the USPTO indicates that the PTAB finds at least one invalid claim in 90% of patents it reviews in completed IPR proceedings, which I confirmed. This latter metric is easier to apply in further characterizing PTAB findings of invalidity given the manual nature of much of the data collection, and it is used in the data that follows.
I reviewed the IPR outcomes for PAE patents, software patents, litigated patents, and transacted patents – all identified using the methodologies described earlier. In addition, I reviewed the IPR outcomes for patents which were issued without any prior art rejection during patent examination and patents which had been through at least one reexamination. These types of patents are sometimes identified as having received relatively decreased/increased examination scrutiny, which one might anticipate to correspond to the rate of findings of invalidity by the PTAB. I found very little variance in the rate of the PTAB finding at least one claim invalid in patents in completed IPR proceedings. In all but one of the foregoing characterizations, the PTAB found at least one claim invalid in 87-95% of the patents. This is a narrow range tightly coupled to the overall 90% rate cited in the preceding paragraph. It is hard to draw a statistical conclusion from such a narrow range. The one characterization that was outside of the 87-95% range was the percentage of PAE patents having at least one claim found to be invalid – the PTAB found at least one claim to be invalid in just 79% of PAE patents. Many will find this counterintuitive to expectations – as described earlier PAE patents are reviewed in completed IPR proceedings at a rate exceeding those for patents overall, but the validity rate of those patents is actually higher than that for patents overall. Interestingly, the PTAB found at least one claim invalid in 100% of patents not owned by PAEs!
Completed CBM Proceedings
The number of completed CBM proceedings is far less than that for IPR proceedings – too low to draw conclusions from. However, it is worth noting that the PTAB has found at least one claim invalid in 100% of completed CBM proceedings. Interestingly, although the USPTO indicates at least one claim in 60% of patents identified in petitions to institute CBM proceedings are challenged on 35 USC 101 grounds and at least one claim in nearly 40% of patents identified in petitions to institute CBM proceedings are challenged on 35 USC 112 grounds, I determined that the PTAB only found at least one claim invalid on 35 USC 101 grounds in 27% of patents in completed CBM proceedings and that the PTAB found no claims invalid on 35 USC 112 grounds in completed CBM proceedings.
Final Conclusions and Caveats
While it is not possible to draw firm conclusions from a limited data set such as that here, overall, it is clear that once a challenge proceeding is instituted the PTAB is finding invalidity of at least one patent claim at a very high rate – patentees should have little expectation that their patents will survive without any loss of claims. The data about PAEs and software patents bears watching going forward. Keep in mind that my findings may reflect more on the selection of patents and grounds for challenge by petitioners, and more on the decisions by the PTAB whether to institute proceedings, than on quality of the patents themselves. Note particularly that the earliest patents to be challenged may not be reflective of those that will be challenged in the future.
One final note of caution: I am not a statistician.
NOTE: This article was updated on 9/11/14 at 11:55 am (see comments 1 and 3)
 See http://www.uspto.gov/ip/boards/bpai/stats/index.jsp.
 No PGR outcomes have been achieved because only patents issued on patent applications with priority filing dates after March 15, 2013 are eligible for PGR.
 Thanks to IBM colleagues Michele Crecca, Jessie Maihos, Alison Mortinger, and Marian Underweiser for their assistance.
 Necessary image file wrappers were not always available and settlements are not necessarily open to public inspection.
 I take no position here critical or in defense of PAEs.
 There is no one universally accepted definition of PAE.
 PAEs initiated about two thirds of all US patent litigation according to recent patent litigation studies. See 2014 Patent Litigation Study, PwC, July, 2014 and 2013 NPE Litigation Report, RPX Corporation, 2014.
 Data courtesy of the USPTO.
 USPTO Technology Centers (some descriptions abbreviated): 1600: biotechnology and organic chemistry; 1700: chemical and materials engineering; 2100: computer architecture, software, and information security; 2400: computer networks, multiplex communications, video distribution, and security; 2600: communications; 2800: semiconductors, electrical and optical systems and components; 2900: designs; 3600: transportation, construction, e-commerce, agriculture, national security; 3700: mechanical engineering, manufacturing, and products.
 See GAO Report to Congressional Committees, Intellectual Property: Assessing Factors That Reflect Patent Infringement Litigation Could Help Improve Patent Quality, August, 2013. I am not aware of any other estimate in this range.
 Patents may not be challenged on 35 USC 101 and 35 USC 112 grounds in IPR proceedings.