Mystery Graph of the Day

I created the following graph after reading the Federal Circuit’s 3-2 claim construction reversal in Martek. Can you guess what the graph represents?

Mystery Solved by Joe Helmsen from Pepper Hamilton. Joe writes

“I think that the graphs represent the percent chance of having a majority decision in a particular direction given that each of x judges has a y% chance of deciding in that way.  For example, the top line represents 1 judge at 90%, 2 judges have to go 2-0 (.9*.9=81%), 3 judges have to go 3-0 or 2-1 (.9^^3 + 3*.9*.9*.1 =97.2%, etc.).” 

Right. One point of the graph is to illustrate the interesting phenomenon that the odds of convincing a “majority” are much higher when the panel has an odd number of judges. This is easy to understand when comparing a two-judge panel with a three-judge panel. A two-judge panel offers no room for error because requires that you convince both judges. On the other hand, a three-judge panel will side with you if you convince two of the three judges.

The graph also illustrates a second point – that in theory multiple judges tend to make marginal cases more predictable. Thus, an argument that will convince a judge 70% of the time would be predicted to carry the day in almost 90% of 12–member en banc panels.

Of course, this discussion relies on several false premises.  Most notably, the analysis assumes that each panel member decision is independent of the decision made by other panel members. That is is clearly not true.  Rather, the judges and clerks communicate and influence one another.  In addition to independence, the analysis presented here presumes that each judge has the same likelihood of deciding the case in a particular direction. 

19 thoughts on “Mystery Graph of the Day

  1. “When she becomes a registered patent attorney AND has prosecuted 50 – 100 apps before the USPTO, then she will be ready to assume a leadership role. ”

    On the contrary, I think only a patent attorney who has successfully overturned 50-100 patents in court would be a really excellent leader for the USPTO. But I can see why the patent bar, which mostly makes their money churning out patents of dubious quality, would largely disagree.

  2. it’s a sinc function that becomes inverted at 50% and follows a gaussian envelope…what this means is that your graph is most probably correct.

  3. Um folks,

    This isn’t real data. per DC “Of course, this discussion relies on several false premises.”

    It’s pretty though.

  4. Clearly it is the degree of conviction with which the 9 Dentist recommend Trident on either a per minute or per chew basis, the tenth dentist being off the chart in the negative direction somewhere.

  5. This graph shows another interesting point about the impact of a judge abstaining. With an appellate tribunal with an odd number of judges, when a judge abstains the number of voting judges is even. Moreover, if a deadlock supports the status quo (in that the decision on appeal stands), then one judge abstaining can heavily stack the odds in favor of affirming. An easy example of this is shown on the graph where the chance of finding in your favor is 50%. If all of the justices on the Supreme Court vote, then your chance of winning is 50%. If one justice abstains, however, your chance of winning on appeal drops to below 40%.

  6. “We’ve gotten several and can’t figure out why it’s only coming up now.”
    - recent SPE training.
    - flavor of the month.
    - “let’s see, what haven’t I used lately?”
    - “compact prosecution” – yeah right.

  7. Chance that an Examiner will find a new basis for rejection after all previous bases have been addressed as a function of the number of claims, by art unit (1600 at the top).

    Seriously, has anyone else been getting rejections for indefiniteness due to the use of “wherein”? In cases where the wherein clause has been the same for many rounds of prosecution? We’ve gotten several and can’t figure out why it’s only coming up now.

  8. Odds that I will agree with the court’s opinion (broken down by color coded justice writing majority opinion) vs. how many beers I’ve consumed?

  9. Knee jerk thought (without deep reflection : I don’t want to diappoint NAL).

    Is the Y axis the likelihood of success and the x axis the number of judges who will decide, with the proviso that a unanimous verdict is forbidden?

  10. I think that the graphs represent the percent chance of having a majority decision in a particular direction given that each of x judges has a y% chance of deciding in that way.

    For example, the top line represents 1 judge at 90%, 2 judges have to go 2-0 (.9*.9=81%), 3 judges have to go 3-0 or 2-1 (.9^^3 + 3*.9*.9*.1 =97.2%, etc.).

  11. The vertical axis is percentage “sunner” of the Patently-O commenter (e.g. Actual Inventor = 90% sunner, Mooney = 10% sunner), horizontal axis represents “beers drank”, and the curves are confidence curves in Judge Newman’s decisions.

Comments are closed.