That Invention Is Not Obvious! Hindsight Is Worse Than You Think

December 5th, 2022

A recent newspaper article about the inventor of the traffic light observed: “It seems so obvious now. But then that’s the thing about inventions. They’re always plain to see in hindsight.”[1] As the saying goes, “hindsight is 20/20.” The U.S. Patent Office, however, is usually dismissive of this basic insight.

In my experience filing hundreds of patent applications, the U.S. Patent Office is usually dismissive of the effect hindsight has on its determination of obviousness for a claim of a patent or patent application.  This should be no surprise.  Examiners are simply following the guidance provided by the Manual of Patent Examining Procedure (MPEP)—the Patent Office’s official manual establishing ground rules for granting or denying an application. The MPEP is sparing in its guidance on avoiding hindsight, stating: “[h]owever, ‘[a]ny judgment on obviousness is in a sense necessarily a reconstruction based on hindsight reasoning, but so long as it takes into account only knowledge which was within the level of ordinary skill in the art at the time the claimed invention was made and does not include knowledge gleaned only from applicant’s disclosure, such a reconstruction is proper.’"[2] This terse statement dramatically downplays the effect hindsight has on the obviousness analysis that the Patent Office performs. The general belief seems to be that so long as an examiner is aware that he or she should avoid hindsight, that awareness of the principle suffices, without more, to avoid the use of hindsight in evaluating whether an invention would have been obvious.  In fact, hindsight arguments in response to an Office Action almost always fail.  I cannot recall a hindsight argument ever being successful.  The examiner is always sure that her or his rejection is not a product of impermissible hindsight.

This Patent Office view, however, is starkly inconsistent with the well documented understanding that people are generally incapable of avoiding hindsight, even when overtly attempting to do so.  For example, in an oft-cited study by Fischoff, subjects were given a scenario and asked to assign the probability to four different potential outcomes.  In one group, no actual outcome was provided to the subjects.  In another four groups, subjects were told that one of the four outcomes was the “true” outcome, but were told to respond “as they would have had they not known the outcome.”[3]   This study found that, in 13 of 16 cases, the mean probability of the “true” outcome was substantially higher for the group that was told to ignore the “true” outcome when responding.  On average, the probability of an event increased from an average 25% in the group that had no knowledge of the outcome, to an average of 34% in the group that was told to ignore what it knew.  This study demonstrates a principle directly at odds with the MPEP statement.[4]  This suggests that deliberately attempting to avoid hindsight is normally unsuccessful.

The process of examining a patent application is very similar to the method used in the study described above.  The examiner starts with the patent claims and searches the prior art for references that contain concepts found in the claims.  Thus, the examiner, like the subjects of the study above, is aware of the outcome before knowing the situation that gave rise to that outcome.  But, like the subjects in the study, the examiner is supposed to analyze the prior art and determine what is obvious as if the examiner does not know what the claims are. While this procedure is highly efficient, it is a terrible way to avoid hindsight bias, and by extension, a terrible way to render an Office Action that accurately reflects whether an invention is obvious.

If the Board or an examiner, applying the prevailing preponderance of the evidence standard for determining obviousness, believes that a claim is 51% likely to be obvious, the hindsight bias reflected by the study described above suggests that the probability that the claims are obvious is in fact about 18% lower, or 33%.  It follows that for most cases that seem to be close, the Patent Office is usually wrong when it determines that the claims are obvious.  Similarly, for a truly close case, the Board or examiner, failing to adjust for the bias of hindsight, would probably believe that the likelihood of obviousness is about 68%.

A paper by Mandel in the Ohio State Law Journal reported a study that was carried out to determine the role of hindsight in obviousness in patent cases.[5] In this study, subjects were presented with one of two hypothetical inventions and the accompanying prior art.  In the control group, the subjects were provided with the prior art and the problem that the hypothetical inventor was trying to solve. The subjects were then asked, inter alia, “whether, in light of the prior art and information provided in the scenario, a solution to the problem was obvious to a person with ordinary skill in the relevant field.”[6]

In another group in the Mandel study, the subjects were also provided with the same information that the control group received, but received additional information and instructions.  Specifically, the information provided to these subjects had one additional sentence at the end that said that the hypothetical inventor had come up with a solution, and stated what the solution was. These subjects were then provided with instructions based on Model Patent Jury Instructions that informed the participant of the hindsight problem, warned the subjects about it, and advised the subject not to use hindsight in answering the questions.  After receiving these instructions, these subjects were asked the same question as the control group.

The results in this study were astonishing.  For the subjects who were not told about the invention beforehand, only 23% of the subjects believed that a solution was obvious, as compared to 57% of the subjects who were told about the invention and instructed not to use hindsight.  Although not directly comparable to the Fischoff results, Mandel’s results are consistent with Fischoff and clearly show that a significant number of non-obvious inventions are rejected as obvious by the Patent Office.

The Federal Circuit has recognized the problem with “the trap of hindsight” and normally asserts that “[e]vidence of objective indicia” (or “secondary considerations”) such as unexpected results, can help to avoid the hindsight problem.  However, Mandel found that this is not the case.  Specifically, Mandel conducted a comprehensive survey of all reported Federal Circuit and district court non-obviousness decisions for an eighteen month period (from July 1, 2004 to December 31, 2002).[7]  Mandel found that “secondary consideration appears to rebut what would otherwise have been a holding that an invention was obvious in only one to two percent of reported cases over this eighteen-month period.”[8]  Thus, secondary considerations fall far short of overcoming hindsight bias.

The U.S. Supreme Court set forth the standard for obviousness in Graham v. John Deere (Graham) in 1965.  In Graham, the Court stated that “[u]nder § 103, the scope and content of the prior art are to be determined; differences between the prior art and the claims at issue are to be ascertained; and the level of ordinary skill in the pertinent art resolved.  Against this background, the obviousness or nonobviousness of the subject matter is determined.”[9]  It is interesting that in Graham, analysis of the prior art, and not the invention, is the first step in determining obviousness. However, in actual practice, analysis seems to always start with the claimed invention.

Following Graham, hindsight reasoning could be significantly reduced by starting with, and focusing on, the prior art rather than the claims. This really should not be difficult for the Patent Office. For example, an obviousness analysis that started with the prior art could be accomplished if the Patent Office followed the following procedure. A first examiner would search the prior art for the subject matter of the claims as is the current practice. The first examiner would then turn the most relevant references over to a second examiner without disclosing the claims or any other information about the invention other than the references that were found. The second examiner would then spend a few hours evaluating the references and listing all of the inventions that the second examiner believed are obvious based upon the disclosure of the prior art references. This list of obvious inventions would then be compared to the claims. If the list of obvious inventions overlapped with the claims, the claims would then be rejected as obvious. If they did not, then no obviousness rejection would be made. This would be a significant improvement over the current method of examination.

In summary, there is plenty of empirical evidence to show that hindsight bias results in frequent obviousness rejections by the Patent Office for inventions that are not obvious. This could be significantly reduced if the Patent Office divided searching and examination between two examiners, and by having examination carried out by focusing on the prior art without knowledge of the claimed invention.

[1] Lee Benson, The rather unremarkable tale of Lester Wire, and the rather remarkable invention that changed the world, Deseret News (January 3, 2021).

[2] MPEP § 2145 (X)(A), quoting In re McLaughlin, 443 F.2d 1392, 1395, 170 USPQ 209, 212 (CCPA 1971).

[3] Fischoff, Hindsight≠Foresight: The Effect of Outcome Knowledge on Judgment Under Uncertainty, Journal of Experimental Psychology: Human Perception and Performance, Vol. 2, No. 3, 288-299, at p. 293 (1975).

[4] A Google search of “Fischoff and hindsight” returned 3,180,000 hits.

[5] Gregory N. Mandel, Patently Non-Obvious: Empirical Demonstration that the Hindsight Bias Renders Patent Decisions Irrational, 67 Ohio St. L.J. 1391 (2006).

[6] Mandel, at p. 1408.

[7] Mandel, at p. 1421-1423.

[8] Mandel, at p. 1423.

[9] Graham v. John Deere Co., 383 U.S. 1, 18 (1966).