Practice makes perfect? An empirical study of claim construction reversal rates in patent cases

Michigan Law Review

Volumn 107, Issue 2, 2008, Pages 223-284

  Schwartz, David L ^a  

^a John Marshall Law School ^*

Author keywords

[No Author keywords available]

Indexed keywords

EID: 56249144537     PISSN: 00262234     EISSN: None     Source Type: Journal    
DOI: None     Document Type: Article

References (324)

1. See Tonja Jacobi & Emerson H. Tiller, Legal Doctrine and Political Control, 23 J.L. ECON. & ORG. 326 (2007).
2. Presumably, the Federal Circuit has significantly greater expertise at claim construction than district courts. Jeffrey A. Lefstin, Claim Construction, Appeal, and the Predictability of Interpretive Regimes, 61 U. MIAMI L. REV. 1033, 1046 (2007).
3. See, e.g., Gretchen Ann Bender, Uncertainty and Unpredictability in Patent Litigation: The Time is Ripe for a Consistent Claim Construction Methodology, 8 J. INTELL. PROP. L. 175 (2001);
4. Christian A. Chu, Empirical Analysis of the Federal Circuit's Claim Construction Trends, 16 BERKELEY TECH. L.J. 1075 (2001);
5. Kimberly A. Moore, Are District Court Judges Equipped to Resolve Patent Cases?, 15 HARV. J.L. & TECH. 1 (2001) [hereinafter Moore, Judges Equipped];
6. Kimberly A. Moore, Markman Eight Years Later: Is Claim Construction More Predictable?, 9 LEWIS & CLARK L. REV. 231 (2005) [hereinafter Moore, Eight Years Later];
7. Michael Saunders, A Survey of Post-Phillips Claim Construction Cases, 22 BERKELEY TECH. LJ. 2.15 (2007);
8. Andrew T. Zidel, Patent Claim Construction in the Trial Courts: A Study Showing the Need for Clear Guidance From the Federal Circuit, 33 SETON HALL L. REV. 711 (2003).
9. See, e.g., Russell B. Hill & Frank P. Cote, Ending the Federal Circuit Crapshoot: Emphasizing Plain Meaning in Patent Claim Construction, 42 IDEA 1 (2002);
10. Timothy R. Holbrook, Substantive versus Process-Based Formalism in Claim Construction, 9 LEWIS & CLARK L. REV. 123 (2005);
11. Kelly Casey Mullally, Patent Hermeneutics: Form and Substance in Claim Construction, 59 FLA. L. REV. 333 (2007);
12. Craig Allen Nard, A Theory of Claim Interpretation, 14 HARV. J.L. & TECH. 1 (2000);
13. Kristen Osenga, Linguistics and Patent Claim Construction, 38 RUTGERS L.J. 61 (2006);
14. R. Polk Wagner & Lee Petherbridge, Is the Federal Circuit Succeeding? An Empirical Assessment of Judicial Performance, 152 U. PA. L. REV. 1105 (2004).
15. H.R. 34, 110th Cong. (2007) (as retened to S. Comm. on the Judiciary, Feb. 13, 2007).
16. Amgen Inc. v. Hoechst Marion Roussel, Inc., 469 F.3d 1039, 1040 (Fed. Cir. 2006) (Michel, C.J., dissenting);
17. see also Hanis Corp. v. Ericsson Inc., 417 F.3d 124.1, 1266 (Fed. Cir. 2005) (Gajarsa, J., dissenting) ("For various reasons this court already has a high reversal rate on claim construction issues, which tends to encourage appeals and, perhaps, discourage trial courts from, heavily investing in claim constructions below.").
18. District Court Judge Patti Saris has stated that some district court judges are "demoralized" by the claim construction reversal rate. Kathleen M. O'Malley et al., A Panel Discussion: Claim Construction from the Perspective of the District Judge, 54 CASE W. RES. L. REV. 671, 682 (2004). District Court Judge Marsha J. Pechman stated that the high reversal rate on claim construction makes her believe that "you might as well throw darts."
19. Anandashankar Mazumdar, Federal District Courts Need Experts That Are Good 'Teachers,' Judges Tell Bar, 70 PAT. TRADEMARK & COPYRIGHT J. (SNA) 536, 537 (2005);
20. see also Merck & Co. v. Teva Pharm. USA, Inc., 395 F.3d 1364, 1381 (Fed. Cir. 2005) (Rader J., dissenting) (noting that the Federal Circuit "often hears criticism, from district court judges that its reversal rate on claim construction far exceeds that of other circuit courts");
21. Ultratech, Inc. v. Tamarak Scientific Co., No. C 03-03235 CRB, 2005 WL 2562623, at *7 (N.D. Cal. Oct. 12, 2005) ("Nor can the Court say that Ultratech's claim construction position is so frivolous as to warrant sanctions; to be candid, this Court is reluctant to hold that any claim construction is frivolous, given the well-known reversal rate in the Federal Circuit.").
22. For example, one New York trial court stayed a permanent injunction against an adjudicated infringer until the Federal Circuit had a chance to review the claim, construction. The trial court noted its concern about the high reversal rate of lower court's claim constructions as a partial basis for the stay. Cargill, Inc. v. Sears Petroleum & Transp. Corp., No. 5:03 CV 0530(DEP), 2004 WL 3507329, at *12-13 (N.D.N.Y. Aug. 27, 2004) (finding that on balance a permanent injunction against continued infringement should be stayed pending appeal and noting cognizance of the Federal Circuit's "unusually high rate of reversal" in claim construction cases).
23. The Supreme Court has acknowledged that "patent litigation can present issues so complex that legal minds, without appropriate grounding in science and technology, may have difficulty in reaching decision." Blonder-Tongue Labs., Inc. v. Univ. of Ill. Found., 402 U.S. 313, 33.1 (197.1).
24. Federal Circuit Judge S. Jay Plager has noted that district court judges typically do not handle a large volume of patent cases, and "the obscurities and peculiarities of patent law and the complexities of new technology are difficult for many district judges to handle on a one-time basis." S. Jay Plager, Abolish the Court of Federal Claims? A Question of Democratic Principle, 71 GEO. WASH. L. REV. 791, 796-97 (2003).
25. Id. at 797 ("Most district court judges do not have scientific training, and most have not chosen law clerks with technical or patent backgrounds.").
26. The court, not a jury, is required to perform this task. See infra note 17 and accompanying text.
27. Because technology and claim construction are somewhat foreign to most judges, judicial decision-making in this area is probably more deliberative than intuitive. See Chris Guthrie et al., Blinking on the Bench: How Judges Decide Cases, 93 CORNELL L. REV. 1, 8 (2007).
28. A patent claim is a single sentence at the end of an issued patent that "particularly" and "distinctly" points out what the invention is. 35 U.S.C. § 112 (2006).
29. State Contracting & Eng'g Corp. v. Condotte Am., Inc., 346 F.3d 1057, 1067 (Fed. Cir. 2003) ("[W]e have held that a claim 'must be construed before determining its validity, just as it is first construed before deciding infringement.' " (quoting Markman v. Westview Instruments, Inc., 52 F.3d 967, 997 n.7 (Fed. Cir. 1995) (en banc),
30. aff'g, 517 U.S. 370 (1996)));
31. Amazon.com, Inc. v. Barnesandnoble.com, Inc., 239 F.3d 1343, 1351 (Fed. Cir. 2001) ("Because the claims of a patent measure the invention at issue, the claims must be interpreted and given the same meaning for purposes of both validity and infringement analyses.").
32. E.g., Pall Corp. v. Micron Separations, Inc., 66 F.3d 1211, 1220 (Fed. Cir. 1995) ("A patent is infringed if any claim is infringed ... for each claim is a separate statement of the patented invention." (citation, omitted));
33. 5A DONALD S. CHISUM, CHISUM ON PATENTS § 18.03[1][c] (rev. 2007).
34. 517 U.S. 370 (1996).
35. Before 1996, the law was ambiguous as to whether claim construction was the responsibility of the judge or the jury Consequently, the jury would often be charged with both interpreting what the patent meant and determining whether an accused device infringed the patent. Markman, 52 F.3d at 967.
36. Cybor Corp. v. FAS Techs., Inc., 138 F.3d 1448 (Fed Cir. 1998) (en. banc). Recently, several judges on the Federal Circuit have expressed a willingness to revisit Cybor Corp. and the de novo review of district court judge's claim, construction.
37. Amgen v. Hoechst Marion Roussel, Inc., 469 F.3d 1039 (Fed. Cir. 2006). At the time of writing, there has been no change in the level of review.
38. 415 F.3d 1303 (Fed. Cir. 2005) (en banc). Among other things, Phillips resolved a shortlived dispute within the Federal Circuit as to the proper role of dictionaries in claim construction. See, e.g., Joseph Scott Miller & James A. Hilsenteger, The Proven Key: Roles and Rules For Dictionaries at the Patent Office and the Courts, 54 AM. U. L. REV. 829, 905-11 (2005). After Markman, there was an increase in the use of dictionaries in constraing claims by both district court judges and the Federal Circuit.
39. Id. at 847. According to the dictionary line of cases, dictionaries should guide the "plain meaning" of the claim language.
40. See e.g., Texas Digital Sys., Inc. v. Telegenix, Inc., 308 F.3d 1193, 1202-04 (Fed. Cir. 2002). Specifically, the dictionary definition controlled unless the patent specifically redefined the term at issue in the body of the patent (known as the specification).
41. A separate line of cases rejected the view that dictionaries were of primary significance in claim construction. For a fuller discussion of the discrete lines of cases, see Stephanie Ann Yonker, Post-Phillips Claim Construction: Questions Unresolved, 47 IDEA 301 (2007). Phillips ended the dispute, rejecting the line of cases favoring dictionaries.
42. Phillips, 415 F.3d at 1320-21.
43. Phillips, 415 F.3d at 1314, 1316.
44. The claims are the numbered sentences at the end of a patent that point out the scope of the patentee's right to exclude others. 35 U.S.C. § 112 (2006).
45. The specification is the body of the patent that includes drawings and a detailed description of how to make and use the invention. Id.
46. Phillips, 415 F.3d at 1314-15, 13.17. Presumably the claim language itself is the most important device to determine the meaning.
47. Id. at 1314. Claim "terms" refer to phrases within the claim. To "construe" the claim is to determine the meaning of the words or terms used in the claim.
48. Id. at 1315
49. (quoting Vitronics Corp. v. Conceptronic, Inc., 90 F.3d 1576, 1582 (Fed. Cir. 1996)).
50. 5A CHISUM, supra note 16, §1 8.03[2][c][i][A].
51. Id. § 18.03[2][b].
52. Phillips, 4.15 F.3d at 1323.
53. Id.
54. Judge Ronald Whyte, a district court judge with, considerable patent experience, has stated that these canons of construction "seem to be difficult to reconcile with one another." See O'Malley et al., supra note 7, at 675.
55. See, e.g., Andersen Corp. v. Fiber Composites, Inc., 474 F.3d 1361, 1369 (Fed. Cir. 2007).
56. For a more detailed discussion of claim construction, see, for example, EDWARD D. MANZO, CLAIM CONSTRUCTION IN THE FEDERAL CIRCUIT (2008 ed.);
57. Markman Subcomm. of the Patent Litig. Comm. of the Am. Intellectual Prop. Law Ass'n, The Interpretation of Patent Claims, 32 AIPLA Q.J. 1 (2004).
58. No. 2:01 cv 905, 2002 U.S. Dist. LEXIS 27501 (E.D. Va. Aug. 19, 2002).
59. Id. at *1.
60. Nystrom v. Trex Co., 374 F.3d 1105, 1107 (Fed. Cir. 2004), withdrawn and superseded, 424 F.3d 1136 (Fed. Cir. 2005).
61. See Nystrom v. Trex Co., 424 F.3d 1136 (Fed. Cir. 2005)
62. withdrawing and superseding, 374 F.3d 1105 (Fed. Cir. 2004).
63. Nystrom, 374 F.3d at 1110-11.
64. Id. at 1111.
65. Id. at 1109.
66. See id. at 1109-10.
67. Id. at 1110-11.
68. Id.
69. Id. at 1111.
70. Id. at 1112 (emphasis added by court).
71. In this context, "prosecution" describes the interaction between the applicant or its attorney and the U.S. Patent Office with regard to an application for a patent.
72. Nystrom, 374 F.3d at 1111.
73. Nystrom v. Trex Co., No. 2:01 cv 905. 2002 U.S. Dist. LEXIS 27501, at *5 (E.D. Va. Aug. 19, 2002).
74. Nystrom, 374 F.3d at 1109, 1110.
75. Id. at 1113.
76. See id. at 1112.
77. Id. (emphasis added by court).
78. See id.
79. Id.
80. Id. at 1.112-13 (finding that Trex Co.'s proposed narrower construction would improperly import limitations into the claim from the specification).
81. Phillips v. AWH Corp., 415 F.3d 1303 (Fed. Cir. 2005) (en banc).
82. Nystrom v. Trex Co., 424 F.3d 1136 (Fed. Cir. 2005).
83. Id. at 1143.
84. Id.
85. Id. at 1143-44.
86. See id. at 1144-45.
87. Id. at 1151.
88. See sources cited supra note 3.
89. The claim construction reversal rate has been tabulated to be between twenty-five and fifty percent. See Moore, Eight Years Later, supra note 3, at 233.
90. Moore, Eight Years Later, supra note 3.
91. Markman v. Westview Instruments, Inc., 517 U.S. 370 (1996).
92. Moore, Eight Years Later, supra note 3, at 239. This supplemented her previous study covering the period from April 23, 1996 to December 31, 2000.
93. Moore, Judges Equipped, supra note 3, at 8-9.
94. Moore, Eight Years Later, supra note 3, at 239.
95. Id. at 238.
96. Id. at 239.
97. Id.
98. Id. at 244 tbl.2.
99. Moore, Judges Equipped, supra note 3, at 38.
100. Moore, Eight Years Later, supra note 3, at 246.
101. Id.
102. Id. at 247.
103. Id.
104. Chu, supra note 3.
105. Id. at 1092, 1104.
106. Id. at 1094, 1100 n.121.
107. Federal Circuit Rule 36 permits the Federal Circuit to affirm, a decision of a lower court without any written opinion. According to the rule, summary affirmances are limited to situations when "an opinion would have no precedential value" and one of the following is present: (a) the judgment, decision, or order of the trial court appealed from, is based on findings that are not clearly erroneous; (b) the evidence supporting the jury's verdict is sufficient; (c) the record supports summary judgment, directed verdict, or judgment on the pleadings; (d) the decision of an administrative agency warrants affirmance under the standard of review in the statute authorizing the petition for review; or (e) a judgment or decision has been entered without an error of law. Fed. Cir. R. 36.
108. Moore, Eight Years Later, supra note 3, at 235 n.15, 236.
109. Without reviewing the Rule 36 cases themselves, Chu attempted to extrapolate their effect. See Chu, supra note 3, app. A at 1144-47.
110. Id. at 1125-27.
111. Id. at 1122. Chu included the following courts in the "more active" category: the Central District of California, the District of Delaware, the District of Massachusetts, the District of Minnesota, the District of New Jersey, the Eastern District of Michigan, the Eastern District of Virginia, the Northern District of California, the Northern District of Illinois, the Southern District of Florida, and the Southern District of New York. Chu also included appeals from the Board of Patent Appeals and Interferences, the Court of Federal Claims, and the International Trade Commission in the "more active" tribunal group. AU other courts were in the "less active" tribunal group.
112. Id.
113. Id. at 1127.
114. Id. Chu based this statement on a comparison of the average claim, construction reversal rate to the average overall reversal rate for active tribunals.
115. Id.
116. Id.
117. Wagner & Petherbridge, supra note 4, at 1110-11. Wagner and Petherbridge's study included all appeals to the Federal Circuit dealing with claim construction, whether from the district courts or other fora.
118. Id. at 1145.
119. Id. at 1158-59.
120. Id. at 1112.
121. Id. at 1112, 1158-59.
122. R. Polk Wagner & Lee Petherbridge, Did Phillips Change Anything? Empirical Analysis of the Federal Circuit's Claim Construction Jurisprudence (Mar. 30, 2008) (unpublished manuscript, on file with author).
123. Id. at 26.
124. See, e.g., Bender, supra note 3, at 203, 206-07 (finding that between April 1996 and the middle of 2000, the Federal Circuit reversed forty percent of the 160 appealed claim constructions);
125. Saunders, supra note 3, at 235-37 (finding that between July 2005 and September 2006, the Federal Circuit reversed at least one claim in 53.5% of cases);
126. Zidel, supra note 3, at 74.1-42 (finding that in 200.1, the Federal Circuit reversed thirty-nine of the ninety-four claim constructions).
127. A recent student comment on specialized patent courts attempts to analyze "whether district court judges who currently hear the most patent cases are better at claim construction." Nancy Olson, Comment, Does Practice Make Perfect? An Examination of Congress's Proposed District Court Patent Pilot Program, 55 UCLA L. REV. 745, 745 (2008). The short empirical portion of the otherwise strong comment relies upon the number of district court opinions electronically available on Lexis as the measure of judicial experience.
128. Id. at 772-73. The use of electronically available district court opinions has been heavily criticized as possibly "unrepresentative."
129. See, e.g., David A. Hoffman et al., Docketology, Districts Courts, and Doctrine, 85 WASH. U. L. REV. 681, 686-87 (2007);
130. Pauline Kim et al., How Should We Study District Court Judge Decision-Making?, 29 WASH. U. J.L. & POL'Y (forthcoming 2008) (manuscript at 7, on file with author). The comment also excluded Rule 36 cases and included Lexis cases only from, a limited period.
131. Olson, supra, at 77.1-73. The similarity of this Article's title and that of the above comment is purely coincidental due to the overlapping times of submission and publication.
132. Mark A. Hall & Ronald F. Wright, Systematic Content Analysis of Judicial Opinions, 96 CAL. L. REV. 63, 105-06 (2008) (stating that it is crucial for a study using systematic content analysis to permit partial or full replication).
133. E.g., Chu, supra note 3, at 1093;
134. Moore, Eight Years Later, supra note 3, at 39;
135. Moore, Judges Equipped, supra note 3, at 9.
136. For example, in 1998 Beckson Marine, Inc. sued NFM, Inc. in the Western. District of Washington. The district court judge construed one claim term and granted summary judgment, holding there was non-infringement. On appeal, the Federal Circuit reversed and remanded. Beckson Marine, Inc. v. NFM, Inc., 292 F.3d 718 (Fed. Cir. 2002). The district court judge construed additional terms and found in favor the accused infringer. This resulted in another appeal to the Federal Circuit on claim terms construed after remand.
137. Beckson Marine, Inc. v. NFM, Inc., 144 F. App'x 862 (Fed. Cir. 2005).
138. In total, however, less than two percent of the cases in the appellate database resulted in more than one appeal on claim construction.
139. Jay P. Kesan & Gwendolyn G. Ball, How Are Patent Cases Resolved? An Empirical Examination of the Adjudication and Settlement of Patent Disputes, 84 WASH. U. L. REV. 237, 273-74 (2006) (finding that between sixty-five and sixty-eight percent of all patent cases filed in three particular years were resolved via settlement or a probable settlement).
140. See infra Appendix A. 1, A.2.
141. For the precise queries, see infra Appendix A.1, A.2.
142. Lee Epstein & Gary King, The Rules of Inference, 69 U. CHI. L. REV. 1, 81-97 (2002).
143. Id. at 83.
144. Id. at 87.
145. Hall & Wright, supra note 97, at 113-14 (stating that the best practice for relaying reliability information is to report a coefficient such, as "Cohen's Kappa");
146. Lee Petherbridge & R. Polk Wagner, The Federal Circuit and Patentability: An Empirical Assessment of the Law of Obviousness, 85 TEX. L. REV. 2051, 2074-75 (2007) (reporting Cohen's Kappa for intercoder reliability).
147. See infra Appendix A.3.
148. Moore, Eight Years Later, supra note 3, at 239-45.
149. The data in the present study could not be compared with Chu's results because it appears that Chu included non-district court decisions (i.e., appeals from, the Board of Patent Appeals and Interferences and from the International Trade Commission) in his analysis. Chu, supra note 3, at 1092. The data in this study are likewise not comparable to those set forth in Michael Saunders's recent work analyzing post-Phillips claim construction cases.
150. Saunders, supra note 3. Saunders excluded Rule 36 results, preventing their use in this Article for comparison purposes.
151. Id. at 235.
152. See Epstein & King, supra note 104, at 89-90.
153. For more information on how these data were collected, see Appendix A.4.
154. It was not feasible to review the PACER dockets in the 16,000-plus patent lawsuits to determine in which cases the claims had been construed.
155. Because the district court judge database consists only of objective information (the judge's birth year, size of docket provided by CourtLink, etc.), no formal measure of reliability is necessary. See Hall & Wright, supra note 97, at 112.
156. For a discussion of the frequency of patent litigation involving public companies across various industries, see JAMES BESSEN & MICHAEL J. MEURER, PATENT FAILURE 106-09 (2008).
157. John R. Allison & Mark A. Lemley, Empirical Evidence On The Validity Of Litigated Patents, 26 AIPLA Q.J. 185, 202-05, 250-51 (1998).
158. Kimberly A. Moore, Forum Shopping in Patent Cases: Does Geographic Choice Affect Innovation?, 79 N.C. L. REV. 889, 924-31 (2001);
159. Yan Leychkis, Note, Of Fire Ants and Claim Construction: An Empirical Study of the Meteoric Rise of the Eastern District of Texas as a Preeminent Forum for Patent Litigation, 9 YALE J.L. & TECH. 193 (2007).
160. Some have referred to non-practicing entities pejoratively as trolls. E.g., Raymond P. Niro, Who is Really Undermining the Patent System - 'Patent Trolls'or Congress?, 6 J. MARSHALL REV. INTELL. PROP. L. 185 (2007).
161. Eric Herman, Charting the yays and nays in federal court, CHI. LAW., Mar. 1996, at 1, 10 (" '[I]f [judges] have a really tough case, they can put tremendous pressure on the parties to settle so there won't be an appealable order.' "(quoting Judge Richard A. Posner)).
162. Id. (" 'One way not to get reversed is - don't do anything. If you don't do much, the parties will eventually settle the cases and you're not reviewed.' " (quoting an anonymous district court judge)).
163. Kesan & Ball, supra note 101, at 271-74; Arti K. Rai, Engaging Facts and Policy: A Multi-Institutional Approach to Patent System Reform, 103 COLUM. L. REV. 1035, 1059 (2003) (finding that twenty-nine percent of patent cases settle as a result of the district court's claim construction);
164. Patricia A. Martone, Before the Actual Markman Hearing - Timing, Discovery and Alternatives, 753 PLI/Pat 91, at 100 (2003) (noting that claim construction does not always affect settlement).
165. The author is only aware of a single case involving claim construction reviewed by the Federal Circuit in the past thirteen years before entry of an appealable judgment. Regents of Univ. of Cal. v. Dako N. Am., 477 F.3d 1335 (Fed. Cir. 2007).
166. Moore, Judges Equipped, supra note 3, at 9-10;
167. George L. Priest & Benjamin. Klein, The Selection of Disputes for Litigation, 13 J. LEGAL STUD. 1, 4, 16 (1984);
168. cf. Theodore Eisenberg, Testing the Selection Effect: A New Theoretical Framework with Empirical Tests, 19 J. LEGAL STUD. 337 (1990) (testing and rejecting the hypothesis that plaintiffs should prevail in fifty percent of tried cases). At least some empirical evidence of appellate reversal rates refutes the application of Priest and Klein's economic theory to appeals.
169. See, e.g., Kevin M. Clermont & Theodore Eisenberg, Litigation Realities, 88 CORNELL L. REV. 119, 151 (2002). Further, other empirical studies of patent litigation have shown plaintiff win rates in jury trials at almost seventy percent, contrary to what one would expect using the economic theory.
170. See, e.g., Kimberly A. Moore, Judges, Juries, and Patent Cases - An Empirical Peek Inside the Black Box, 99 MICH. L. REV. 365, 385-86 (2000).
171. See, e.g., Moore, Judges Equipped, supra note 3, at 9-10.
172. For a discussion of the potential for a selection bias in the evaluation of only appellate decisions, see Petherbridge & Wagner, supra note 107, at 2071.
173. AM. INTELLECTUAL PROP. LAW ASS'N, REPORT OF THE ECONOMY SURVEY 2007, at 25 (2007). This figure does not include the substantial non-monetary costs of patent litigation.
174. See Matthew Sag & Kurt Rohde, Patent Reform and Differential Impact, 8 MINN. J. L. SCI. & TECH. 1, 29-30 (2007).
175. Patent cases are roughly five times more likely to be appealed than other civil lawsuits. Paul R. Michel, The Court of Appeals for the Federal Circuit Must Evolve to Meet the Challenges Ahead, 48 AM. U. L. REV. 1177, 1193 (1999). Once the appeal has been filed, very few patent cases are settled.
176. See U.S. Court of Appeals for the Fed. Circuit, Merit and Non-Merit Dispositions for Appeals in Patent Infringement Cases: April 2007 through March 2008, http://www.cafc.uscourts. gov/pdf/Patent. DispositionsChartApr07-Mar08.pdf (last visited Oct. 11, 2008) (documenting a mere thirteen percent settlement rate in patent infringement appeals from April 2007-March 2008).
177. For an interesting model of how firms decide whether to litigate or settle a patent dispute, see Sag & Rohde, supra note 126, at 73-91.
178. It is still disputed whether the Federal Circuit decisions are inflicted by a panel effect. Compare Wagner & Petherbridge, supra note 4, at 1158-61 (asserting that Federal Circuit judges use different methodologies in approaching claim, construction and that which judges sit on the panel affects the results)
179. with John R. Allison & Mark A. Lemley, How Federal Circuit Judges Vote in Patent Validity Cases, 27 FLA. ST. U. L. REV. 745 (2000) (noting that the voting patterns of the Federal Circuit judges on validity were quite similar).
180. The plaintiff to a lawsuit can mark only one box to describe the nature of the lawsuit, regardless of how many different causes of action are asserted. In addition to the problems of over inclusion and under inclusion, this also may slightly affect the data. See Admin. Office of the U.S. Courts, Form JS 44: Civil Cover Sheet (Mar. 1999), available at http://www.uscourts.gov/ forms/ JS044.pdf.
181. Federal Judicial Center, How the Federal Courts Are Organized: Federal Judges and How They Get Appointed, http://www.fjc.gov/federal/courts.nsf (follow "How The Federal Courts Are Organized" hyperlink; then follow "Federal Judges and How They Get Appointed" hyperlink) (last visited Aug. 14, 2008).
182. See JAMES C. DUFF, JUDICIAL BUSINESS OF THE UNITED STATES COURTS: 2007 ANNUAL REPORT OF THE DIRECTOR 42 tbl.11 (2007), available at http://www.uscourts.gov/judbus2007/ JudicialBusinespdfversion.pdf.
183. Moore, supra note 117, at 903-04. Much more recently, the Eastern District of Texas has become a favorite venue for patent litigation.
184. Leychkis, supra note 117.
185. Although judge assignment methods vary district by district, all districts are required to make assignments that assure an equitable distribution of caseloads and avoid judge shopping. The majority of courts use some form of a random drawing. The Federal Judiciary: Frequently Asked Questions, http://www.uscourts.gov/faq.html (last visited Aug. 14, 2008).
186. Data obtained through a search of Judicial Strategic Profile in LexisNexis CourtLink, limited to Patent in the Nature of Suit field.
187. The Eastern District of Texas has more recently become a common venue for patent litigation. See Roderick R. McKelvie, Forum Selection in Patent Litigation: A. Traffic Report, 19 INTELL. PROP. & TECH. L.J., Aug. 2007, at 1, 2 (noting that the Eastern District of Texas was the second-most-popular venue for patent litigation in 2006). Because this trend has begun fairly recently, many of the cases from the Eastern District of Texas have not had time to proceed through appeal.
188. See id. (presenting evidence that the number of lawsuits filed in the Eastern District of Texas increased by eight-fold from. 2001 to 2006). Consequently, the Eastern District of Texas is not one of the top fifteen districts in. terms of appellate claim construction experience during the period from 1996 to 2007.
189. The Federal Circuit reversed, vacated, and/or remanded 43.5% of appeals of cases from the Central District of California.
190. According to the U.S. Courts: A U.S. magistrate judge is a judicial officer of the district court and is appointed by majority vote of the active district judges of the court to exercise jurisdiction over matters assigned by statute as well as those delegated by the district judges. . . . A full-time magistrate judge serves a term of eight years. The Federal Judiciary: Frequently Asked Questions, supra note 134.
191. Less than nineteen percent of decisions were ones in which the same district court judge was reviewed twice in a single calendar year.
192. This metric of evaluating the reversal rate is difficult to compare across cases. The Federal Circuit is inconsistent in how many construed terms it will review on appeal because some panels of the Federal Circuit will stop all claim construction analysis after finding a single claim element properly construed and missing from the accused device. Other panels will analyze and review all claim constructions on appeal. The result of the differences between Federal Circuit panels is that the same effort and construction by two district courts may result in different reversal counts if the focus is on claim terms. Take as an example a case in which the district court construes three disputed claim, terms and finds all three missing from the accused infringer's products. The patentee appeals a grant of summary judgment of non-infringement to the Federal Circuit. The Federal Circuit may respond on appeal in several ways. First, the Federal Circuit may analyze only one element, find the district court was correct, affirm, and decline to review the other two terms. Alternatively, the Federal Circuit may review all three elements, finding one to be correctly construed and the other two inconectly construed. In this case it will still affirm the district court because at least one term was missing from the accused device. Yet another possible result on appeal is review of all of the terms and finding all three terms correctly construed. In all three events, the district court's overall evaluation of the case was correct. However, the difference between the second and third decision is profound when the reversal rate by claim term is analyzed. Because many panels use the first method of reviewing terms, which entails stopping after finding one term to affirm, one cannot tell which of the cases should be in the second or the third category.
193. A case was categorized as "reversed" if, as a result of an improper claim, construction, the district court was reversed, vacated, and/or remanded as to any accused product or patent claim.. The "reversed, vacated, and/or remanded" category and the "at least one term wrongly construed" category have a significant difference. Specifically, the latter includes harmless errors. While these errors were harmless for the patentee vis-à-vis the accused infringer at issue, the corrected claim, constraction may be material for others in the industry. A better measure of the reversal rate is the number of decisions that had to be reversed, vacated, and/or remanded due to an erroneous claim construction. If the case is remanded, a judge must redo the case after utilizing the correct claim constraction. If the case is affirmed despite an erroneous claim constraction, the district court judge may pay less attention to the appellate decision. Other items, such as such as the technology at issue, may affect the reversal rate of the district court judge.
194. The Federal Circuit also has jurisdiction over non-patent cases, the most significant of which are cases that arise under the Little Tucker Act. 28 U.S.C. § 1295 (2006). A search of the LexisNexis "CAFC" database reveals that, from. 2004 to 2006, 769 of Federal Circuit cases dealt with patent law, while twenty dealt with the Little Tucker Act.
195. STATISTICS DIV., ADMIN. OFFICE OF THE U.S. COURTS, 2004 JUDICIAL BUSINESS OF THE UNITED STATES COURTS 117 tbl.B-8 (2004), available at http://www.uscourts.gov/judbus2004/ appendices/b8.pdf;
196. STATISTICS DIV., ADMIN. OFFICE OF THE U.S. COURTS, 2005 JUDICIAL BUSINESS OF THE UNITED STATES COURTS 143 tbl.B-8 (2005), available at http://www.uscourts.gov/ judbus2005/appendices/b8.pdf;
197. STATISTICS DIV., ADMIN. OFFICE OF THE U.S. COURTS, 2006 JUDICIAL BUSINESS OF THE UNITED STATES COURTS 147 tbl.B-8 (2006), available at http://www. uscourts.gov/judbus2006/appendices/b8.pdf.
198. Many of the metrics used to evaluate district court judges could not be used to evaluate magistrate judges. For example, the number of patent lawsuits assigned to a magistrate judge would be biased relative to the number for district court judges. It is common for magistrate judges to be assigned to a patent case and participate in discovery only, with no role in claim construction. Martone, supra note 121, at 155. Counting each, lawsuit a magistrate judge was assigned as equal to each one a district court judge was assigned would not yield useful results.
199. Joseph L. Smith, Patterns and Consequences of Judicial Reversals: Theoretical Considerations and Data from a District Court, 27 JUST. SYS. J. 28 (2006) (finding empirical support for the proposition that judges react to reversals by predictably changing their decision-making patterns).
200. Reversal rates are one possible way to measure judicial performance. RICHARD A. POSNER, HOW JUDGES THINK
201. cf. Jeffrey A. Lefstin, The Measure of the Doubt: Dissent, Indeterminacy, and Interpretation at the Federal Circuit, 58 HASTINGS LJ. 1025, 1032 (2007) (arguing that appellate dissents are a better indicator than reversals to measure "indeterminacy").
202. Appendix B provides a further dissection of the data based upon prior appeals. More specifically, the data shows that the Federal Circuit decides cases under Rule 36 without a strong regard to the experience of the district court judges. Thus, the reversal rate based upon experience does not appear to be caused by a selection bias in terms of whether a decision is resolved through Rule 36.
203. Summarizing the figure, 29.6%, 27.4%, 30.2%, 28.6%, 50.0%, and 27.8% of the cases were reversed, vacated, and/or remanded due to claim construction for district court judges with, zero, one, two, three, four, and five or more previous claim construction appeals, respectively. Figure 3 also shows that 38.4%, 36.5%, 37.7%, 39.3%, 56.7%, and 3.1.9% of cases had at least one enoneous claim construction for judges with zero, one, two, three, four, and five or more previous claim construction appeals, respectively. The n (number of cases) for each bar was 453, 197, 106, 56, 30, and 72, respectively.
204. The null hypothesis that the percentages in each group are the same cannot be rejected. A chi-square test was performed on the number of cases that were affirmed and that had to be reversed, vacated and/or remanded due to enoneous claim, constructions in the six categories (no previous appeals, one previous appeals, two previous appeals, three previous appeals, four previous appeals, and five or more previous appeals). The p-value measures the confidence level at which a hypothesis can be rejected. A p-value of 0.05 or less signifies that the hypothesis can be rejected with a 95% confidence level. Here, the p-value was 0.252, which is greater than the 0.05 expected pvalue for a 95% confidence level. Consequently, the hypothesis - that there are no differences between the groups-cannot be rejected. As previously mentioned, all inferences from the data are subject to the limitations of the study described in Section ILD.
205. More precisely, 31.4%, 25.5%, 30.8%, 44.4%, and 25.0% of cases were reversed, vacated, and/or remanded for zero, one, two, three, four, and five or more appeals after first reversal, respectively. Figure 4 also shows that the percentage of cases with any erroneous claim, constraction was 37.3%, 31.4%, 34.6%, 44.4%, and 33.3%, respectively. The n (number of cases) for each batwas 102, 51, 26, 18, and 36, respectively.
206. The null hypothesis that the percentages in each group are the same cannot be rejected. The chi-square test p-value was 0.59.1, greater than the expected p-value for a 95% confidence level.
207. A variable such as age may appear to be unimportant because another variable is concealing its affect on claim construction. Multivariate analysis would permit one to control for other variables.
208. The data illustrated in Figure 5 does not adjust over time the number of patent cases handled by a particular district court judge. For example, a district court judge that has handled 300 patent cases is grouped in the 300 category for all of the appeals of his or her decisions.
209. The percentage of cases reversed is 26.3%, 26.9%, 30.0%, 32.9%, 29.9%, and 27.1% for judges with 1-10 cases, 11-25 cases, 26-50 cases, 51-100 cases, 101-150 cases, and more than 151 cases, respectively. The number of cases in each group was 99, 167, 290, 210, 97, and 48, respectively. The number of judges in each group was 91, 114, 153, 69, 19, and 5, respectively. CourtLink inexplicably identified a few judges with zero patent cases. These judges were excluded from Figure 5.
210. The null hypothesis that the percentages in each group are the same cannot be rejected. The chi-square test p-value was 0.801, which is greater than the expected p-value for a 95% confidence level.
211. The three-year "trail" is the average of the three years that precede the given year. For example, the three-year trailing average of fifteen years' experience is the number of cases reversed that were handled by judges with thirteen, fourteen, and fifteen years' experience divided by the total number of cases appealed from judges with thirteen, fourteen, and fifteen years' experience. Using a trailing average reduces noise since there are so many discrete years of judicial experience.
212. The null hypothesis that the percentages in each group are the same cannot be rejected. The chi-square test p-value was 0.105, which is greater than the expected p-value for a 95% confidence level.
213. See, e.g., Patricia A. Reuter-Lorenz & Cindy Lustig, Brain Aging: Reorganizing Discoveries About the Aging Mind, 15 CURRENT OPINION IN NEUROBIOLOGY 245 (2005);
214. cf. Sara Reistad-Long, Older Brain Really May Be a Wiser Brain, N. Y. TIMES, May 20, 2008, at F5 (stating that research has found that while the aging brain may take longer to remember items, it "is simply taking in more data and trying to sift through a clutter of information, often, to its long-term benefit").
215. The number of decisions for judges in the age groups of forty to forty-nine years old, fifty to fifty-nine years old, sixty to sixty-nine years old, seventy to seventy-nine years old, eighty to eighty-nine years old, and ninety to ninety-eight years old were 72, 358, 298, 143, 40, and 4, respectively.
216. The null hypothesis that the percentages in each group are the same cannot be rejected. The chi-square test p-value was 0.530, which is greater than the expected p-value for a 95% confidence level.
217. While a 30-40% reversal rate appears quite high, this Article does not show that the claim construction reversal rate is high relative to the reversal rates in other complex areas of law. See Lefstin, supra note 2, at 1038-39. Nor does this Article address the reason for the "high" reversal rate in claim construction. Rather, this Article addresses why district court judges do not appear to improve at claim construction as they hear more cases.
218. The Federal Circuit's decision in Phillips may have corrected this problem. However, Phillips was decided in 2005 - relatively recently. As more cases are resolved by the district courts post-Phillips and are subsequently appealed, the effect of Phillips can be more fully explored.
219. District court judges have a strong incentive to learn, because they do not enjoy being reversed. POSNER, supra note 146, at 141;
220. Evan H. Caminker, Precedent and Prediction: The Forward-Looking Aspects of Inferior Court Decisionmaking, 73 TEX. L. REV. 1, 77-78 (1994).
221. Yet another possible explanation is that the abilities of counsel for the litigants have improved over time. The improvement may be at either the trial-court level or the appellate-court level, or both. This possible explanation is not considered in this Article.
222. Lefstin, supra note 146, at 1030;
223. S. Jay Plager, Challenges for Intellectual Property Law in the Twenty-First Century: Indeterminacy and Other Problems, 2001 U. ILL. L. REV. 69, 71-72;
224. see also Improving Federal Court Adjudication of Patent Cases: Hearing Before the Subcomm. on Courts, the Internet, and Intellectual Property of the H. Comm. on the Judiciary, 109th Cong. 57 (2005) (statement of the Honorable T.S. Ellis, III) (noting that claim construction is difficult because of the "vagaries of language");
225. O'Malley et al., supra note 7, at 676 ("Often when I get to claim construction . . . I see a couple of reasonable interpretations.").
226. See, e.g., H.L.A. HART, THE CONCEPT OF LAW 131-34 (2d ed. 1994).
227. Id.
228. Even if the terms were expressly defined in the patent, the words used to define the claim terms may still need to be interpreted.
229. U.S. PATENT & TRADEMARK OFFICE, U.S. DEP'T OF COMMERCE, MANUAL OF PATENT EXAMINING PROCEDURE § 608.01(m) (8th ed. rev. 2007).
230. Phillips v. AWH Corp., 415 F.3d 1303, 1313 (Fed. Cir. 2005) (en. banc).
231. See e.g., Gregory N. Mandel, The Non-Obvious Problem: How the Indeterminate Non-Obvious Standard Produces Excessive Patent Grants, 42 U.C. DAVIS L. REV. (forthcoming 2008) (manuscript at 14, available at http://papers.ssrn.com/sol3/papers.cfm7abstract_id=1117618) (arguing that, for many of the same reasons that claim construction is indeterminate, the doctrine of patent non-obviousness is indeterminate).
232. Blonder-Tongue Labs., Inc. v. Univ. of Ill. Found., 402 U.S. 313, 331 (1971) ("We are also aware that some courts have frankly stated that patent litigation, can present issues so complex that legal minds, without appropriate grounding in science and technology, may have difficulty in reaching decision.").
233. This applies to judges at all levels, including both district court and appellate court judges.
234. E.g., Nyyssonen v. Bendix Corp., 342 F.2d 531, 532 (1st Cir. 1965) ("[T]his [patent] case presents great difficulties to judges like ourselves who have only the most elementary training in science and mathematics and little experience with, modern technological developments.");
235. Parke-Davis & Co. v. H.K. Mulford Co., 189 F. 95, 115 (S.D.N.Y. 1911),
236. aff'd in part, rev'd in part, 196 F. 496 (2d Cir. 1912) ("I cannot stop without calling attention to the extraordinary condition of the law which makes it possible for a man without any knowledge of even the rudiments of chemistry to pass upon such questions as these. The inordinate expense of time is the least of the resulting evils, for only a trained chemist is really capable of passing upon such facts . . . .").
237. Because patents are issued for all different technologies, judges could be assigned to disputes based upon technological experience. U.S. Patent Examiners are assigned to review patent applications based upon technical expertise. Moreover, claims must be construed as they would be understood by a person of skill in the art. The closer the fact finder is to that hypothetical person, the easier it will be to understand the underlying technology and properly to construe the claims.
238. The Federal Rules of Civil Procedure permit the court to appoint a master to construe the claims and require the parties to pay the expenses for the master. FED. R. CIV. P. 53.
239. E.g., John B. Pegram, Should There Be a U.S. Trial Court With a Specialization in Patent Litigation?, 82 J. PAT. & TRADEMARK OFF. SOC'Y 765 (2000) (advocating that all patent cases be assigned to the United States Court of International Trade);
240. James F. Holderman, Keynote Address, Judicial Patent Specialization: A View From the Trial Bench, 2002 U. ILL. J.L. TECH. & POL'Y 425, 431-33 (2002) (supporting Mr. Pegram's proposal to assign patent cases to the United States Court of International Trade).
241. Rochelle Cooper Dreyfuss, In Search of Institutional Identity: The Federal Circuit Comes of Age, 23 BERKELEY TECH. L.J. 787, 804-806 (2008);
242. Pegram, supra note 177, at 782-86;
243. Arti K. Rai, Specialized Trial Courts: Concentrating Expertise on Fact, 17 BERKELEY TECH. L.J. 877 (2002).
244. Representatives Darrell Issa of California, Adam Schiff of California, and Steve Cohen of Tennessee co-sponsored the bill. Govtrack.US, H.R. 34, http://www.govtrack.us/congress/bill.xpd?bill=h110-34 (last visited Sept. 29, 2008).
245. An Act to Establish a Pilot Program in Certain United States District Courts to Encourage Enhancement of Expertise in Patent Cases Among District Judges, H.R. 34, 110th. Cong. (2007) [hereinafter Patent Pilot Program Bill].
246. Id.
247. Id. § 1 (b). For Fiscal Year 2007, the judicial districts with the most patent litigation were the Central District of California, Eastern District of Texas, Northern District of California, District of Delaware, District of New Jersey, Northern District of Illinois, Southern District of New York, District of Massachusetts, Northern District of Georgia, Southern District of Florida, District of Minnesota, Eastern District of Pennsylvania, Southern District of California, Eastern District of Michigan, and Middle District of Florida.
248. STATISTICS DIV., ADMIN. OFFICE OF THE U.S. COURTS, 2007 JUDICIAL BUSINESS OF THE UNITED STATES COURT 194-207 tbl. C-11 (2007), available at http://www.uscourts.gov/judbus2007/ JudicialBusinespdfversion.pdf.
249. Patent Pilot Program Bill, supra note 180, § 1(b)(1). In order for a district to qualify, a minimum of three judges must opt into hearing patent cases.
250. Id. § 1(b)(2).
251. Id. § 1(b).
252. Id. § 1(a)(1)(A).
253. Id. § 1(a)(1)(B).
254. Id. § 1(a)(1)(C).
255. Id. § 1(a)(1)(D).
256. Patent experience may aid in areas of patent law other than claim construction.
257. It is possible that district court judges need a huge number of appeals - i.e., forty or fifty-before they learn. The present study cannot eliminate that possibility. It is also possible that providing education to judges on claim construction would help.
258. The Patent Pilot Program also appropriates $5,000,000 per year for education of judges in patent law and for compensation for law clerks with technical backgrounds. Patent Pilot Program Bill, supra note 180, § 1(f). Presently, few trial court judges have clerks with technical backgrounds.
259. See Alan D. Lurie, U.S. Circuit Judge, Federal Circuit, Speech to PTC Section of D.C. Bar (June 12, 2000), in 60 PAT., TRADEMARK & COPYRIGHT (BNA) 147, 148 (2000). It is possible that more technical law clerks would aid in lowering the reversal rate.
260. Patent experience may also affect other aspects of litigation, including the duration a lawsuit is pending.
261. The judges for the regional court of appeals are more likely to know the local district court judges, both because of proximity and because of frequency of review. See FMC Corp. v. Glouster Eng'g Co., 830 F.2d 770, 771-72 (7th Cir. 1987).
262. See Plager, supra note 165, at 77-78 (identifying several district court judges with a "particular bent for patent cases");
263. Interview by Mark Smith with Ronald M. Whyte, U.S. Dist. Judge, N. Dist. of Cal. (Mar. 16, 2007), available at http://www.techlawforum.net/ transcript/whyte-9-representative-claims.txt.
264. From the data, it is not possible to determine the reason for the high reversal rate and the lack, of teaching or learning. It seems likely that the Federal Circuit is at least partially responsible for the high reversal rate. In addition, the rate of non-unanimous opinions by the Federal Circuit in claim construction decisions has increased since Phillips. Wagner & Petherbridge, supra note 93, at 24. Dissents on the appellate court hint that the reversal rate is not entirely the district court's fault because it shows that there was some disagreement on the correct outcome or reasoning at the appellate level.
265. Changing the standard of review could be done through overruling Cybor Corp. v. FAS Technologies, Inc., 138 F.3d 1448, 1454-56 (Fed. Cir. 1998) (en banc). Several judges of the Federal Circuit are inclined to do so.
266. See Amgen Inc. v. Hoechst Marion Roussel, Inc., 469 F.3d 1039 (Fed. Cir. 2006) (Michel, C.J., dissenting). Alternatively, Congress could change the standard through legislation.
267. See Tom Chen, Patent Claim Construction: An Appeal for Chevron Deference, 94 VA. L. REV. (forthcoming 2008);
268. John F. Duffy, On Improving the Legal Process of Claim Interpretation: Administrative Alternatives, 2 WASH. U. J.L. & POL'Y 109 (2000);
269. Rai, supra note 178.
270. See Paul M. Schoenhard, Reversing the Reversal Rate: Using Real Property Principles to Guide Federal Circuit Patent Jurisprudence, 17 FORDHAM INTELL. PROP. MEDIA & ENT. L.J. 299 (2007), for a proposal to provide deferential review of claim construction based upon real property law concepts.
271. Fed. Circuit Bar Ass'n Markman Project, Guidelines for Patent Claim Construction: The Basics of a Markman Hearing, 14 FED. CIR. B.J. 771, 779 (2005).
272. See O'Malley et al., supra note 7, at 679.
273. See James F. Holderman in collaboration with Halley Guren, The Patent Litigation Predicament in the United States, 2007 U. ILL. J.L., TECH. & POL'Y 1, 7 (2007).
274. An even more radical approach would be to overturn Markman. Not only does that seem unlikely, but it would also be unwise. Transparency would be decreased if juries construed claims in the "black box" of deliberations.
275. Patent Reform Act of 2007, H.R. 1908, 110th Cong. § 11(b) (2007);
276. Patent Reform Act of 2007, S. 1145, 110th Cong. § 11(b) (2007).
277. This increase would occur because at present many cases in which the claims have been construed settle before an appealable judgment is entered.
278. Chief Judge Michel of the Federal Circuit has written to Congress on behalf of the Federal Circuit expressing this concern. His letter states: "I would expect an interlocutory appeal in virtually every patent infringement case as soon as a claim construction order issues." Letter from Paul R. Michel, Chief Judge, U.S. Court of Appeals, Federal Circuit, to Senators Patrick Leahy and Arlen Specter, U.S. Senators, 2 (June 13, 2007), available at http://www.intellectualpropertylawblog.com/ Michel%20letter%20to%20Senators%206-13-07(1).pdf. He further predicts that this flood of appeals would not only overload the Federal Circuit, it would lead to "extended delays" for the litigants before the matter is resolved.
279. Id.
280. See Craig Allen Nard & John F. Duffy, Rethinking Patent Law's Uniformity Principle, 101 NW. U. L. REV. 1619 (2007).
281. Id. at 1651-55.
282. Holmes Group v. Vornado Air Circulation Sys., Inc., 535 U.S. 826 (2002).
283. Schinzing v. Mid-States Stainless, Inc., 415 F.3d 807, 811 (8th Cir. 2005).
284. Dreyfuss, supra note 178, at 810-11. Others have argued that even if the second appellate court adopted different claim construction rules, there is no evidence that increasing the number of judges would increase the quality of decision-making or decisional rules.
285. S. Jay Plager & Lynne E. Pettigrew, Rethinking Patent Law's Uniformity Principle: A Response to Nard and Duffy, 101 NW. U. L. REV. 1735, 1744 (2007).
286. It is possible that certain judges of the Federal Circuit are more likely to rule pro-patent or pro-accused infringer in claim construction cases. The disposition of the appeal could in turn be dependent upon the makeup of the panel.
287. Different panels of the Federal Circuit have occasionally construed the same patent terms differently. E.g., Moore, Judges Equipped, supra note 3, at 18-20 (discussing a case in which a patent was construed inconsistently by two different panels of the Federal Circuit).
288. See Holbrook, supra note 4, at 150;
289. Lefstin, supra note 2, at 1041.
290. Others have noted that the Supreme Court prefers a standards-based approach in patent cases and that the Federal Circuit prefers a rules-based approach. Claim construction appears to be an example of the Federal Circuit preferring a standards-based approach. See Holbrook, supra note 4, at 144-45. As the results of this Article appear to demonstrate, the standards-based approach is largely ineffective.
291. Since the standards approach, taken to date appears to be ineffective, one may argue that a rules-based approach is better. Under a rules-based approach, the particular canons of claim construction could remain as presently articulated. But the district court judges would no longer be required or permitted to select which canons to apply in a given situation. The canons of construction could be locked into a fixed hierarchy at the appellate court level. For example, the first rule could be that limitations are never imported from the specification into the claim. This hierarchy could be set with a practical goal in mind - the more objective and easily definable the canon, the more important the Court should deem the rule. However, selecting the hierarchy requires policy decisions because it will have policy implications. The precise order of application of the canons will affect the value of patents. Making rules that broaden patents favors patentees, while making rules that result in narrow constructions disfavor patentees. It would also be difficult to prescribe iron-clad rules for all claim construction issues. And even iron-clad rules like the "dictionary first" rule from Texas Digital Systems, Inc. v. Telegenix, Inc. still leave the indeterminacy of words problems. 308 F.3d 1193 (Fed. Cir. 2002). Thus, it is likely unfeasible to shift claim construction to a rules-based approach.
292. Plager & Pettigrew, supra note 210, at 1746.
293. Plager & Pettigrew, supra note 210, at 1746.
294. See Plager, supra note 165, at 72.
295. 2.18. See Lee Petherbridge, Positive Examination, 46 IDEA 173 (2006) (arguing that an important part of the problem with claim construction is that the U.S. Patent Office does not focus its examination on the scope of the claims and urging reform to make prosecution focus on establishing the boundaries of patent claims).
296. Further study of the cause of the high reversal rate is warranted. It is possible that a more complex study which models the relationship between the appellate and district courts with multiple control variables might yield a different result. A future study should focus on lawsuits at the time of filing to reduce potential selection-bias concerns. These lawsuits could be traced and evaluated at various points during the litigation process. It would also be advisable to control for various aspects that may affect the appeal rate of the cases. For example, non-practicing entities as patentee may be more likely to appeal an adverse claim construction. Similarly, a generic drug manufacturer as an accused infringer may be more likely than, average to appeal an adverse claim construction.
297. The exact search query was "claim w/10 (constru! or interp!) and date(geq (04/24/1996) and leq (6/30/2007)".
298. The LexisNexis CAFC database appeared to have minor instability issues. When identical queries were run one week apart, the number of hits went down slightly. A sample of the missing cases were manually checked, and it was determined they should not have been in the CAFC database in the first instance. These cases included non-patent appeals from the United States Court of Appeals for Veterans Claims and other irrelevant appeals. A call to a search consultant for LexisNexis confirmed that technical improvements were being made to the CAFC database during the time the searches were executed. Cf. Jason J. Czarnezki & William F. Ford, The Phantom Philosophy? An Empirical Investigation of Legal Interpretation, 65 MD. L. REV. 841, 862 n.95 (2006) (noting other unexplained inconsistencies in a LexisNexis database).
299. While a validity analysis technically requires claim construction, a majority of Federal Circuit decisions relating to validity do not explicitly address it.
300. In a small number of Federal Circuit decisions issued shortly after Markman, the district court trial had occurred prior to Markman, and a jury had construed the claims.
301. A very few cases were eliminated in which the Federal Circuit decided two cases on the same day from the same judge on the same claim terms of the same patent(s). The only difference in the cases was the identity of the defendant(s). In these instances, one of the companion cases was eliminated. It did not make sense to count these cases as two separate decisions by a particular district court judge.
302. Included in the lower court judges' decisions were nineteen cases decided entirely by U.S. magistrate judges. Apparently, the parties in those cases consented to trial before a federal magistrate judge, and the magistrate judge construed the claims and entered final judgment.
303. The exact search query was "(("federal circuit rule 36") (fed.cir.r.36) (fed.cir +2 r.36) (fed.cir.r +2 36)) & da(aft 4/23/1996 & bef 7/1/2007)".
304. The PACER Service Center is maintained by the Administrative Office of the U.S. Courts and is available at http://pacer.psc.uscourts.gov/.
305. There were five Rule 36 cases for which there was insufficient information to determine whether claim construction was at issue on appeal. These cases were not included in the study.
306. FED. CIR. R. APP. P. 32.1 (Practice Notes).
307. See Matthew Lombard et al., Practical Resources for Assessing and Reporting Intercoder Reliability in Content Analysis Research Projects, http://www.temple.edu/sct/mmc/reliability (last visited Aug. 16. 2008).
308. Federal Judicial Center, Judges of the United States Courts, http://www.fjc.gov/public/home.nsf/hisj (last visited Sept. 29, 2008).
309. LexisNexis Courtlink, http://www.lexisnexis.com/courtlink/online (last visited Sept. 29, 2008).
310. LexisNexis Courtlink Frequently Asked Questions, http://www.lexisnexis. com/courtlink/online/faqs.asp (last visited Aug. 16, 2008).
311. A small number of patent cases may be misclassified as trademark cases in PACER and thus may not appear in response to a search limited to patent cases. Deepak Somaya, Strategic Determinants of Decisions Not to Settle Patent Litigation, 24 STRATEGIC MGMT. J. 17 (2003). Others have found that some cases coded as patent cases in PACER are not patent cases in fact.
312. See, e.g., Kesan & Ball, supra note 101, at 261 (finding that the nature of suit category was overinclusive by about eight to ten percent for three years of patent filings).
313. It is believed the results include all cases in which the district court judge was the judge of record at the time the matter was closed - settlement, judgment, transfer, or otherwise - or was the judge who was presiding over a case pending at the time of this Article.
314. More precisely, 20.3%, 19.3%, 17.9%. 32.1%. 20.0%. and 31.9% of decisions in the claim construction appellate database were Rule 36 decisions for judges with zero, one, two, three, four, and five or more prior appeals, respectively.
315. Excluding the last group, the null hypothesis that the percentages in each group are the same cannot be rejected. The chi-square test p-value was 0.154, which is greater than the expected p-value for a 95% confidence level. Including all groups, the chi-square test p-value was 0.069, just slightly greater than the expected p-value for a 95% confidence level.
316. More precisely, 22.7%, 23.6%, 15.0%, 22.2%. and 22.2% of the cases were decided using Rule 36 or judges with a total of one, two. three, four, and five, or more total appeals, respectively.
317. The null hypothesis that the percentages in each group are the same cannot be rejected. The chi-square test p-value was 0.479. which is greater than the expected p-value for a 95% confidence level.
318. Here too, the null hypothesis that the percentages in each group are the same cannot be rejected. The chi-square test p-value was 0.696. which is greater than the expected p-value for a 95% confidence level.
319. The dataset obviously includes some district court judges who handle a large number of patent lawsuits and who are particularly efficient at encouraging settlement. These judges have a lower number of appeals than judges who are otherwise comparably busy. The dataset was not adjusted to take into account efficiency at encouraging settlements.
320. More precisely, 28.1%, 33.0%. 27.9% 25.9%. 35.6% and 30.7% of cases were reversed, vacated, and/or remanded due to a claim construction error for judges that had one, two, three, four, five, and six or more, respectively, total claim construction appeals. The number of observations for each bar was 256, 182 (91 judges). 147 (49 judges), 108 (27 judges), 45 (9 judges), and 176 (21 judges), respectively. The data also shows that 37.5%, 41.2%, 33.3%, 35.2%, 42.2%, and 39.8% of cases had at least one error in claim construction for judges that had one, two, three, four, five, and six or more, respectively, total claim construction appeals.
321. This number differs from the overall reversal rate because the cumulative percentage is an average of the raw, underlying data.
322. More particularly, 29.7%. 27.4%, 29.7%, 30.6%. 30.3%, 32.4%, 33.3%, 28.1%, 30.0%. 23.8%, and 37.5% of the cases were reversed, vacated, and/or remanded due to claim construction for district court judges with zero, one, two, three, four, five, six, seven, eight, nine, or ten or more years since their first Federal Circuit claim construction decision, respectively. The number of observations for each bar. beginning with year zero, was 512, 84, 74, 62, 66, 34, 39, 32, 20, 21, and 8.
323. The null hypothesis that the percentages in each group are the same cannot be rejected. The chi-square test p-value was 1.000, which is greater than the expected p-value for a 95% confidence level.
324. To compare district court judges with different numbers of appeals, a table was derived of the odds of all combinations of decisions affirmed and decisions reversed, vacated, and/or remanded. To determine the odds, it was assumed that the likelihood of being affirmed for each decision was 70.0% (which is approximately the actual average affirmance rate for this dataset) and that the odds for each decision were independent. For example, for judges with four appeals, there is a 0.8% chance of having four reversals ( 1 · (0.3) · (0.3) · (0.3) · (0.3)), a 7.6% chance of one affirmance and three reversal/vacated (4 · (0.7) · (0.3) · (0.3) · (.03)), a 26.5% chance of two affirmances and two reversals (6 · (0.7) · (0.7) · (0.3) · (0.3)), a 41.2% chance of three affirmances and one reversal (4 · (0.7) · (0.7) · (0.7) · (0.3)). and a 24.0% chance of four affirmances (1 · (0.7) · (0.7) · (0.7) · (0.7)). From the table, the ten judges were selected whose decisions were reversed, vacated, and/or remanded the least (the "good" judges), based upon the likelihood of randomly achieving those results. The thirteen judges were also selected whose decisions were reversed, vacated, and/or remanded the most (the "bad" judges), based upon the likelihood of randomly achieving those results. An equal number of "good" and "bad" judges could not be selected because the break of judges with the same record was uneven.