Is the Federal Circuit Succeeding by yungtyriq

VIEWS: 13 PAGES: 27

									Is the Federal Circuit Succeeding?
An Empirical Assessment of Judicial Performance

                R. Polk Wagner
           University of Pennsylvania
Considering the Patent Law as an Institutional Design Problem

•   The increasingly central role of the Federal Circuit
    o   an institutional mandate for jurisprudential management
    o   a robust (aggressive?) implementation of this mandate since 1990

•   Recent events call for such an inquiry:
    o   widespread dissatisfaction among the patent community
    o   and yet ... a growing awareness of the importance of decisions

•   Critical analytic need: the performance of the Federal Circuit
    against its mandate (clarity, stability, predictability)
    o   a look at the court “in action”
    Unit of Measurement: Federal Circuit Jurisprudence
        { First, the Caselaw on Claim Construction }
•   The Federal Circuit’s mandate is especially powerful here; the basis
    of Markman II is jurisprudential management.

•   The academic literature and commentary suggests opportunities:
    o   ‘high’ reversal rates (±30-50%) in literature (is this high?)
    o   practitioner and (district) court commentary
    o   universal agreement about importance (”dispositive”)

•   Claim Construction offers an excellent data profile:
    o   a ‘pure’ Federal Circuit question: question of law, no deference
    o   a large dataset: claim construction is raised in virtually all cases,
        typically dispositive
                        Overview of the Study

•   Dataset: all Federal Circuit claim construction analyses since
    Markman II (April 1996)
    o   rolling updates (typically within 2 to 3 months of current)
    o   as of May 10, 2005: 636 opinions (585 for the court, 51 alternative
        opinions) [all data released to public immediately]

•   Case-Coding (converting court opinions into useful data):
    o   each opinion is coded for 14 categories: usual categories +
        ‘methodological approach’
    o   all opinions independently coded by multiple coders (early dataset =
        2, later dataset = 5+ or more); any discrepancies among coders
        resolved, tracked, and tabulated.
     A Methodological Split: Procedural versus Holistic
                     A Taxonomy of Methodological Approach

                   procedural                                               holistic

    strong         intermediate          weak              weak          intermediate          strong
     (Ps)               (Pi)             (Pw)              (Hw)               (Hi)              (Hs)
 Expresses rigid   Framework of     Difficult to discern Difficult to discern Acknowledgment No discussion of
    process.       formal process      the form of         the form of      of process, but not process.
                    established.        analysis.           analysis.              used.

Ordinary meaning Departure along Some discussion of    Little or no   Specification/ Immediate use of
 controls, absent some dimension:    process.         discussion of prosecution history spec/prosecution
     express       e.g., no talk of                 process/ordinary     is clearly         history.
 definitions, etc.   presumption.                        meaning.        dispositive.
                  Why Methodology Matters:
           Which Approach is Used Affects the Results
•   Universal agreement that claim construction is most critical factor
    in patent disputes — and that different approaches lead to
    different results.
•   The Procedural/Holistic dichotomy is drawn from the jurisprudence
    itself — this is what people argue about (typically described as
    “two lines of cases”).
•   (Some) empirical testing of this point:
    o   Claim construction disputes among Federal Circuit judges tracked the
        detailed methodological categories 95% of the time.
    o   Federal Circuit reversals of D.Ct’s claim construction analysis tracked
        binomial categories 82% of the time.
OVERALL RESULTS
    Frequency of Methodological Approach (binomial)


   Holistic        9.28%          9.12%            15.57%




Procedural         11.16%                 16.82%                            38.05%




              0%            10%            20%              30%       40%        50%     60%   70%


                                          weak              Intermediate        Strong
                         45.0%
                                    Frequency Distribution of Methodologies
                         40.0%



                         35.0%



                         30.0%
% of total opinions




                         25.0%



                         20.0%



                         15.0%



                         10.0%



                          5.0%



                          0.0%
                                      Ps      Pi      Pw      Hw      Hi                Hs
                      court          37.9%   16.8%   12.0%   9.1%    8.5%             15.7%
                      alternative    39.2%   17.6%   2.0%    11.8%   15.7%            13.7%

                                                                             data through May 10, 2005
                                                                                   n = 636 (all opinions)
          Key Trends
1. A Clear Move Towards Proceduralism
2. Increasing Disagreement, Polarization
                                                       A Clear Move to Proceduralism:
                                            Claim Construction Methodology Over Time
                                 100%




                                 90%




                                 80%




                                 70%
% of opinions coded procedural




                                 60%




                                 50%




                                 40%




                                                                                        20-case moving average
                                 30%
                                                                                        100-case moving average
                                                                                        Overall Trend

                                 20%




                                 10%




                                  0%
                                        1   51   101     151   201   251   301   351   401              451          501             551

                                                                                                                   data through May 10, 2005
                                                                                                                 n=584 (opinions for the court)
                                                   Increasing Disagreement at the Federal Circuit:
                                                         The Rate of Alternative Opinions on Claim Construction
                                         40%


                                                                   20-case moving average
                                         35%
                                                                   100-case moving average


                                         30%
% of cases with an alternative opinion




                                         25%



                                         20%



                                         15%



                                         10%



                                         5%



                                         0%
                                               1    51       101         151       201       251   301   351   401   451   501   551            601

                                                                                                                                  data through May 10, 2005
                                                                                                                                       n = 636 (all opinions)
JUDICIAL METHODOLOGY
( PROCEDURALISTS, HOLISTICS & SWINGS )
                                                              The Methodology of Federal Circuit Judges:
                                                        of Factions of the Federal Circuit
                                             ProportionThe holistic opinions, as author and panelist
                                    80%

                                              Proceduralists                                          Swing Judges                                           Holistics
                                    70%



                                    60%
% of opinions coded holistic




                                    50%



                                    40%



                                    30%



                                    20%



                                    10%



                                     0%
                                               RL      TD      RC      AS      AG      SP      JP       PM      RR      [pc]    GA      GR      RM      AL       WB        PN
                               as Author      10.3%   11.5%   11.7%   26.2%   28.9%   28.6%   30.8%    31.9%   37.0%   38.2%   46.7%   38.9%   40.0%   50.6%    63.0%     70.0%
                               as Panelist    17.3%   24.1%   23.8%   27.9%   27.6%   25.4%   40.3%    38.2%   33.3%           31.4%   38.0%   33.3%   44.1%    41.8%     45.6%

                                                                                                                                                       data through May 10, 2005
                                                                                                                                                             n=637 (all opinions)
                  Factions on the Federal Circuit
Judges grouped by M-index (lower M-index = more procedural)
                       Judge         M-index (authors)   std. score
                        Linn              0.145            (1.41)
  Procedualists         Dyk               0.154           (1.35)
                     Clevenger            0.190            (1.10)
                        Prost             0.319           (0.22)
                       Schall             0.295           (0.38)
                      Gajarsa             0.318           (0.22)
                       Plager             0.327            (0.16)
  Swing Judges         Mayer              0.330            (0.14)
                       Michel             0.354            0.03
                       Rader              0.343           (0.05)
                    [ per curiam]         0.350           (0.00)
                       Archer             0.387            0.25
                       Bryson             0.503            1.05
    Holistics          Lourie             0.536            1.27
                      Newman              0.706            2.44
                       Judicial (In)consistency:
Federal Circuit Judges Ranked by Consistency in Claim Construction Methodology
                      Rank             Judge        Variance
                       1                 Linn        0.304
                       2              Clevenger      0.319
                       3                 Dyk         0.320
                       4                Schall       0.435
                       5                Prost        0.437
                              mean                   0.445
                       6               Gajarsa       0.446
                       7              Newman         0.464
                       8                Michel       0.471
                       9                Rader        0.474
                       10               Plager       0.480
                       11            [per curiam]    0.490
                       12              Bryson        0.502
                       13               Lourie       0.503
                       14               Archer       0.507
                       15               Mayer        0.516
                           The Changing Federal Circuit?
    Judge        seniority rank   procedural rank   consistency rank participation rank authorship rank   activity rank
    Linn              14                1                 2                 10                 9               3
     Dyk              15                2                 3                 11                11               8
  Clevenger            9                3                 1                 3                  3               4
    Schall            11                4                 5                 2                  7               10
   Gajarsa            13                5                 5                 7                  6               9
    Prost             16                6                 4                 13                13               12

   Mayer               5                8                 15                9                 16               15
    Rader             10                9                 8                 1                  2               2

[ per curiam ]         —                10                10                —                 10               —

   Michel              6                11                7                 8                  4               5

   Bryson             12                15                13                4                  5               7

   Lourie              8                16                12                6                  1               1

  Newman               3                17                6                 5                  8               13
PANEL-DEPENDENCY
Judges’ Impact on Methodological Approach
      Judge                    B                           S.E.                Holistic Probability

       Dyk                   -1.368                      0.441**                     20.3%

     Clevenger               -1.020                      0.250**                     26.5%

       Linn                  -0.894                      0.397**                     29.0%

       Prost                 -0.674                       0.865                      33.7%

      Rader                  -0.486                      0.234**                     38.1%

      Schall                 -0.446                      0.237*                      39.0%

     Gajarsa                 -0.294                       0.255                      42.7%

      Mayer                  -0.166                       0.244                      45.9%

      Michel                 -0.025                       0.255                      49.4%

     Newman                  0.157                        0.228                      53.9%

      Lourie                 0.365                        0.229                      59.0%

      Bryson                 0.589                       0.234**                     64.3%

                  * = significance @ .10 level ** = significance @ .05 level

      binary logistic regression, dependent variable = binomial category
Factional Impact on Methodological Approach
                                                Opinion Methodology

                                       procedural                     holistic

                                           1
                                   3                                     –
                                         100%
                                           30                            3
     Number of                     2
                                          91%                           9%
    Proceduralist
                                          179                           63
 judges impanelled                 1
                                          74%                          26%
                                          210                          149
                                   0      58%                          42%

                                                                         2
                                   3       –
                                                                       100%
                                           35                           34
     Number of                     2
                                          51%                          49%
      Holistic
                                          179                          120
 judges impanelled                 1
                                          60%                          40%
                                          206                           60
                                   0      77%                          23%

                                          420                          216
                 Overall Results          66%                          34%
                                      Panel Impact on Methodological Approach
                                                                                         PPH



                                                                                         PPS



                                                                                         PHH
panel composition (by class)




                                                                                         PSS



                                                                                         SSS



                                                                                         PHS



                                                                                         HHS



                                                                                         HSS



                                                                                         HHH


                               0.0%     12.5%   25.0%   37.5%             50.0%              62.5%   75.0%   87.5%   100.0%
                                                                % opinions coded procedual
Implications
                      Recap: The Basic Findings

•   A distinct, yet persistent split in methodological approach.
    o   65% procedural, 35% holistic.

•   A distinct trend towards proceduralism (statistically significant).

•   Increasing polarization (statistically significant).

•   The emergence of distinct factions (proceduralists, holistics).

•   Panel composition absolutely affects methodology.
    o   For some panels, results can be predicted with 90%+ reliability.
                  Is the Federal Circuit Succeeding?
•   Maybe: The performance of the Federal Circuit on claim construction has
    not been an unqualified success: a persistent split; increasing
    polarization; a division among Judges.
    o   note: pre-1982, we had variable results, based on circuit; now we have
        variable results, based on panel assignments!
•   But some of the trends might suggest a turnaround:
    o   a growing stabilization on the procedural approach
    o   polarization = ‘working it out’
    o   Phillips offers an opportunity ... though the court may not seize it
        - it might have been better to avoid en banc treatment
•   The results highlight the importance of personnel changes.
•   Pre-announcing panel membership is likely to have mixed effects at best.
predictor.claimconstruction.com
                        The data allows
                       some reasonable
                         predictability
                        concerning the
                      claim construction
                      results of a given
                            panel.
                      This tool provides
                         an interface.
search.claimconstruction.com
                          Search
                            and
                        download
                            the
                      entire dataset.

                        Build your
                           own
                         analysis.
Is the Federal Circuit Succeeding?
An Empirical Assessment of Judicial Performance

                 R. Polk Wagner
        University of Pennsylvania Law School
                 polk@law.upenn.edu
                    215.898.4356

								
To top