Amy Semet, Princeton University Patent Law Differences Among the Federal District Courts: A Look at the Patent Pilot Program 5 Years Later Amy Semet, Princeton University
Questions How do the district courts differ in deciding patent cases? How do the cases they hear vary among districts by issue and procedural posture? Do judges with more experience or expertise in patent law make decisions that are less likely to be overturned on appeal by the Federal Circuit?
Theoretical background Specialized judges make “better” decisions Foster uniformity and consistent precedent Reduce forum shopping, decreasing costs Increase efficiency by decreasing the learning curve Some scholars have found that specialization in patent law leads to more efficiency and less mistakes on appeal E.g., Kesan & Ball 2011; Shartzer 2009 But in deciding claim construction, scholars have not found that specialized judges differ from non-specialized E.g., Schwartz 2008, 2009; Olson 2009; Gutter 2009
Patent pilot program In 2011, Congress set up a patent law pilot program to enhance expertise among judges in 14 district courts 14 courts include popular districts like E.D.TX and N.D.CA but exclude D.DE as well as many regional patent judges Goals Enhance efficiency Reduce mistakes on appeal Reduce forum shopping
Dataset Patent law cases from late 2011 to 2016 heard at Federal Circuit heard in the district courts Exclude BPAI, ITC; jury cases; venue; discovery; patent law not primary issue; stays; remand Dependent variables Pr (Reverse in full = 1) Pr (Reverse in part=1) Pr (Reverse in part=1), Main Issue
Breakdown of cases in pilot program + Delaware
Appealed cases by main issue
Appealed cases by procedure
Types of mistakes overall
Mistakes made by trial courts
Measuring specialization Pilot program judge Pilot program district Pilot program judge + District of Delaware Pilot program judge + District of Delaware + regional judges with at least 3 cases in the dataset Adds judges from, among other districts, E.D.VA, E.D.WI, D.MA Only judges with at least 2 or 3 cases in database
Control variables Technology (11 categories) Issue Infringement, invalidity, claim construction, inequitable conduct, damages, preliminary injunction, other/jxn Procedure Summary judgment, bench, JMOL, preliminary injunction, dismissal/pleadings Summary motion v. after fact development Case complexity Proxied by dissent Alternatively, length of opinion, forward patent citations, number of claims Party of trial judge and CAFC panel Lower court decision/CAFC Year indicators or time trend District
Statistical analysis, full reversal Pilot Pilot District Pilot+DE Pilot+Experience Specialized Not Stat Not Stat. Stat.* Issue Claim, Validity + Sig.** Claim, Validity + Sig. ** Procedure Dismissal, Bench – Sig* Dismissal, Bench -Sig.* Dismissal< Bench - Sig.* Dismissal, Bench - Sig.* Techn. Included Complex. Other Controls Observ. 571 * p < 0.05, ** p < 0.01, *** p < 0.001
Statistical analysis, partial reversal Pilot Pilot District Pilot+DE Pilot+Experience Specialized Not Stat Not Stat. Not. Stat. Issue Claim, Validity, Damages+ Sig.** Claim, Validity , Damages + Sig. ** Claim, Validity, Damages+ Sig. ** Claim, Validity, Damages + Sig.** Procedure Dismissal -Sig** Dismissal -Sig.** Dismissal, Bench -JMOL + Sig.** Dismissal - Sig.** Techn. Included Complex. Other Controls Observ. 571 * p < 0.05, ** p < 0.01, *** p < 0.001
Limitations Coding sensitivities Claim construction v. infringement/invalidity, especially in Rule 36 cases How measure specialization/experience/expertise Alternative variables Social usefulness/more economically valuable inventions How measure case complexity Selection effects and strategic behavior Code more closely for party identity Use other models like propensity score matching
Results Specialization/experience does not appear to enhance correctness to any great extent But could have benefits for certain issues like claim construction Pilot program judges do not differ from non pilot program judges in terms of having lessened mistakes on appeal But adding in other judges who hear many patent cases results in the variable becoming statistically significant No solid evidence pilot program reduced forum shopping or led to more accurate results Could help achieve efficiency More analysis by legal issue May take time to work, i.e., District of Delaware took time to develop sustained expertise
Proposals for reform: pilot program Wrong judges: add more regional judges, Delaware Focus: focus should be both experience and expertise Resources: Allocate more technical experts Forum: instead of modeling the pilot program off the current system, perhaps a nationally-based trial court akin to Tax Court might be a better solution Model after similar systems in Europe and Asia
Proposals for reform: internal and administrative reform Problem may lie in CAFC not being an effective transmitter of clear and consistent precedent Bigger question of institutional capacity and whether courts v. administrative agency is a better choice Maybe we are relying too much on courts to decide patent cases Give PTO rulemaking authority/greater delegation to the PTO or another new administrative agency since they are the agency with the expertise to decide rules of patent law Administrative agency is in best position to design rules to set balance between innovation and protection
Future research Non CAFC cases: random selection of underlying district court cases including cases not appealed Other measures of success such as speed Look at more by issue Use of precedent by lower courts Perhaps pilot courts are seen as more authoritative source of patent law Differences with trademark and copyright law Do specialized patent cases make “better” decisions on other IP issues?