Empirical Predictors of Patent Validity

Empirical Predictors of Patent Validity

Ralph D. Clifford

The problem of low patent quality continues. Too many patents are issued for claims that are anticipated or made obvious by the prior art. Few believe that the AIA will have any significant impact on this problem. Many commentators (including myself) have suggested that a significant source of the problem is a result of the lack of competent prior art searching as neither the applicant nor the PTO are doing an adequate job of screening applications. This paper is designed to continue my research on how the patent bar contributes to this problem.

In the first paper written to address this topic[1], a group of us established that members of the patent bar are not properly trained in the technologies underlying the applications they are submitting. This lack of competency results in more than nonoptimally drafted claims. Further, a patent practitioner without training is not likely to be familiar with the fundamental scholarship within the technology. Without knowledge, disclosure to the PTO is impossible.

In the next paper in the series[2], the compounding problem of a lack of a requirement on the patent attorney to search for prior art was explored. This second paper established that, rather than requiring searching and disclosure by patent applicants, the PTO’s rules provide a disincentive for the patent attorney to even attempt to find relevant prior art. All that need be disclosed is what is known. If the patent practitioner is willfully ignorant of existing prior art — or was ignorant because of training in a non-relevant discipline – the burden of discovery rests exclusively with the PTO’s examiners who have not proven to be particularly competent themselves. The purpose of this third paper will be to examine issued patents in an attempt to develop empirically-based measurements that confirm the failures defined in the first two papers. The fundamental question being posed is, “Are there empirical measurements that accurately predict whether the claims within a patent would be ultimately found to be invalid?” Consequently, measurements of the quantity and quality of the prior art searching done by the applicant and his or her representatives will be the highest priority.

To derive predictors of patent quality, a database will be developed comprised of the record of any patent that has had its validity litigated. To be included in the database, the standards required for res judicata will be used — the patent has had one or more claims declared by a court to be either valid or invalid.

Once this database is available, the individual patents involved will be examined as will their prosecution history. Various quantitative values about each patent will be captured. A primary measurement will be the number of prior art references contained in the application as well as those disclosed during the prosecution. Other values which may be probative include the type of prior art disclosed (prior patent, publication, etc.), the training of the patent attorney or agent who drafted the patent, the number of claims contained in it, the duration of the prosecution, as well as other measurable values that may prove predictive.

After the patent-specific data are derived, it will be possible to determine if the data contained within the patent application provide methods of predicting when a patent will be found wanting by a court. The primary hypothesis of the paper is that there will be a strong relationship between the quantity and quality of pre-submission prior art disclosed and the ultimate validity of the patent. As the data are examined, other secondary hypotheses will be articulated and tested. If the primary hypothesis is proven, the article will conclude with a call for the strengthening of the requirements for prior art searching.

[1] Ralph D. Clifford, Thomas G. Field, Jr., & Jon R. Cavicchi, A Statistical Analysis of the Patent Bar: Where Are The Software-savvy Patent Attorneys?, 11 N.C. J. L.. & TECH. 223 (2010).

[2] Ralph D. Clifford, Is It Time for a Rule 11 for the Patent Bar?, – IDEA – (2013).