Rochelle Dreyfuss, NYU
What the Federal Circuit Can Learn from the Supreme Court, and Vice Versa
In the early 1970s, the regional circuits couldn’t handle the appellate load. More judges would lead to more intracircuit inconsistency and more cases. New circuits would also breen more intercircuit splits for the Supreme Court to decide. One possibility: experimenting with specialist courts to reduce the dockets of the regional circuits, take pressure off the SCt, reduce pressure for forum-shopping, produce coherence in a field, and use expertise to decide cases more efficiently. A brain surgeon practicing every day may do a faster and better job than someone who does brain surgery once every few years. Patent law was the killer app.
Fears: tunnel vision. Judges might be so focused on patents they’d ignore nonpatent incentives to innovate like curiosity or prizes. They might leave the judicial mainstream. There might be difficult boundary issues between the specialized court and the regional courts.
Congress gave the Federal Circuit authority over things other than patent, meaning that other groups are involved in lobbying for appointments and judges must be aware of nonpatent developments in the law.
Result: success, on the whole. Many other countries are copying the model. No capture—if anything, concern is that not enough appointees have patent experience; repeat players don’t seem to dominate because patentees are often on both sides of cases. Supreme Court intervention cleared up early boundary problems: once a case is before the Federal Circuit, it decides all the issues, not just the patent issues; but the Circuit doesn’t have jurisdiction if the only patent issues are in defense.
Has eliminated forum shopping for patent issues, which is really important for industries looking to make big exceptions. Is some forum shopping at the district court level, but things are still vastly better. Markman: eliminated jury trials on claim construction, creating more predictability, another value for industry. There are still complaints about claim construction, but empirical study shows that the level of uncertainty is no greater than for contract construction, plus the Federal Circuit’s expertise has smoothed out uncertainty in a lot of areas, and the court has become knowledgeable about the tech industry so it’s now influential within the judiciary as a whole on licensing issues.
So now comes the Supreme Court, reversing its practice of reviewing Fed. Cir. decisions only intermittently and on procedural issues. Now it’s begun to intervene substantively and reversed or modified almost every time: Medimmune, Quanta v. LG, eBay v. Mercexchange, Microsoft v. AT&T, and several others. Bilski, pending, doesn’t look so good for the Fed. Cir. either. But SCt review generally is declining, and this trend is happening without any circuit splits. Is this an implicit criticism of the Fed. Cir.’s work?
She says no: every circuit comes into focus eventually, and it’s now the Fed. Cir.’s turn. Plus, SCt involvement is salutary.
The two courts have a lot in common. They’re both part of the experiment with a specialized court: how a judiciary largely committed to generalist jurisdiction deals with a specialized court. Both are largely courts of last result, and have responsibility for superintending courts below them.
Federal Circuit’s relationship to the generalist SCt, who reviews its work, and to the generalist district courts, whose work it reviews: Review of Fed. Cir. seems particularly intrusive. Fed. Cir. judges are specialists, but SCt justices have no experience at all with patent cases—except for Stevens, by the time they were appointed to the appellate bench, they wouldn’t ever have heard a patent case. The Fed. Cir. has little chance to see how patents fit into the economy as a whole, while the SCt does. At the micro level the SCt is making sure standards for injunctive relief and standing and antitrust treatment are uniform across all bodies of doctrine. At the macro level, the SCt has largely pushed the reset button: the Fed. Cir. took the commitment to patents to heart, but that turned into a mixed blessing with concerns that there are now too many patents with low quality and that the high cost of patent litigation is chilling innovation. eBay steps in to reduce incentives to litigate; KSR makes it harder to get patents, improving patent quality.
What about Fed. Cir.’s relationship to generalist district courts? Supposed to defer to trial court’s factual findings, given the trial court’s unique access to facts. But that comparative advantage is diminished in patent cases, where the judges may not be familiar with the tech. Fed. Cir. has the background and hires clerks for their technical backgrounds. Other countries that have specialized patent courts have mostly established the court at the trial level—real gains from specialization might well be in factfinding.
Fed. Cir. faces specialization issues more regularly, and has most to teach SCt. Fed. Cir. has been attentive to the factfinding review question. In earliest nonobviousness cases, the Fed. Cir. undertook a detailed examination of the facts. The SCt held that Rule 52 allowed reversal only for clearly erroneous rulings, even though the Fed. Cir.’s grasp of the facts was clearly better. But the Fed. Cir. didn’t give up. First, required trial courts to apply specific analytical techniques in factfinding, like “long-felt need.” Second, characterized issues as ones of law rather than fact to allow itself to bring its expertise to bear. True, these sacrificed flexibility for predictability, but patent industries liked that.
Markman approved recharacterizing facts as law—fit in with SCt agenda to limit jury factfinding—but the SCt is busily dismantling the Fed. Cir.’s analytical framework as overly rigid. May also happen in Bilski. But the SCt hasn’t faced the larger issue of expertise head-on. How to find the sweet spot between rules and standards? Need to help the Fed. Cir. deal with deficiencies in lower court factfinding. If the SCt looked at that, it could also help the Court figure out what to do about other complex factual issues like medical malpractice.
SCt should take Fed. Cir.’s advice on certworthiness—to a certain extent happening now, because of strong dissents in recent cases the Court has taken. Then, the judges of the Fed. Cir. have to be careful about what they’re teaching—should for example have let Bilski play out before assessing it. And other inputs should matter—for example when patent has diverged from other bodies of law, as with eBay.
Should the SCt defer on substantive law? SCt has assumed role of teacher, chastising Fed. Cir. for departing from precedent. But the Fed. Cir. can’t just apply SCt precedent. Tech changes rapidly. The internet exploded; the structure of the patent industries changed, with joint ventures and university involvement becoming common. And yet before KSR the SCt hadn’t addressed nonobviousness since the 1970s. On micro and macro issues, the SCt should be the teacher, but it shouldn’t get testy about Fed. Cir. departures from its rules when there are contemporary problems not contemplated by old precedent.
SCt tends to leave implementation questions to the Fed. Cir. Is the SCt the best institution to set midrange policy? In other technical areas, an administrative agency takes care of that. SCt works very hard on identifying the theory on which it’s making midrange rules—the policy arguments for and against the alternatives, and why it chose the rule it did. In part this comes from experience resolving circuit splits, but the Court did the same thing in KSR and Festo, etc. In contrast, though the Fed. Cir. recites policy justifications for the statutory requirements of patent law, it rarely provides policy reasons for its own decisions and some judges affirmatively say that doing so would be inappropriate.
Why does the Fed. Cir. deny policy motives? Maybe didn’t want to make waves while it was experimental, but the Fed. Cir. is now part of the fabric of the US judiciary. It must justify and explain policy, or it can’t play its role. The Fed. Cir. never attempted to engage Breyer’s dissent in Labcorp, just dismissing it as not controlling law, even though the claims were basically the same. If the Fed. Cir. had explained why it was ignoring Rule 52 on factfinding, or why it adopted analytical rules, the SCt may have taken it more seriously. Bilski at least moves in this direction, referencing Diamond v. Diehr multiple times.
How to deal with the special problems for being a supervisory/administrative court that is mostly a court of last resort? The SCt’s docket is littered with cases arising from its own failure to provide clear rules, and lowered caseload makes the problem worse; the Fed. Cir. could set a good example. On the other hand, being a court of last resort means that you need to articulate policy well to guide the trial courts: good explanations can substitute for rigid rules—the more the trial court understands why the Federal Circuit rules as it does, the better they’ll be able to apply the rules. Might help avoid appeals based only on linguistic variations in tests.
Damage control: where a court needs to reconsider an issue it’s already laid to rest. That’s not always easy. Repeat players don’t want to annoy the judge/jeopardize future cases. The number of PTO cert petitions plummeted after the Fed. Cir. was established—the PTO is the ultimate repeat player. Another example: Common-law experimental use privilege: consternation in the research community at the cutback in the privilege; the Fed. Cir. heard a case in which it could have done damage control, but the attorneys in that case decided to use a statutory argument, creating doubt about the scope of the common law for more than 5 years. Need to empower lawyers to find good cases to bring back issues for consideration—some Fed. Cir. dissents are along those lines. Fed. Cir. bar should learn like the SCt bar does to read the tea leaves.
Q: What about TM?
A: Hasn’t studied Fed. Cir. TM cases, but might suggest putting all IP in specialized courts to give those courts experience with all the different ways of incentivizing innovation. More antitrust jurisdiction would help for the same reason. Taiwan specialized court: its jurisdiction was innovation policy.
Josh Sarnoff: The SCt cases have often been weak on explaining the policy behind its rules—eBay, KSR—destroys the clear rules but doesn’t give a clear sense of what the policy should be. The SCt may be reluctant to speak clearly and change the nature of the patent system too broadly.
A: Part of the problem is that the policies aren’t neatly teed up for the SCt. A 2nd Circuit case disagreeing with the 9th Circuit will explain why the 9th Circuit is wrong and the 2nd is right. We don’t have that in patent law—the Fed. Cir. at most will articulate its own policy; would be useful to describe the different policy options in more detail. If they did, the SCt might be less interested in intervening in midrange policy in the first place!
Q: What’s the role of judge made law?
A: The Patent Act is really short now (if it ever got reformed it would be like copyright, 120 pages), so most of the law has to be judge made law, and the question is “which judges?” The Fed. Cir. doesn’t have enough cases to make macro policy, and at a micro level you want the rules to be the same across the entire judicial system, as with eBay. (Though as a civ pro professor, she’d never taught the rule articulated in eBay as the rule for granting injunctions! Where did they get this rule? The SCt thought it was harmonizing, but it used a rule that didn’t exist anywhere else!)
Q: Why doesn’t the IP bar suggest that, given IP’s importance to our economy, more judges with patent experience should be appointed, even to the SCt?
A: Stevens (also Breyer and Ginsburg) have antitrust experience; she thinks that suggestion would be a good idea. Appointments to the SCt are so politicized, though.
Q: There seem to be a group of judges at the Fed. Cir. who seem to want to en banc every big question. Is that provocative to the SCt, compared to allowing the cases to bounce around for a bit?
A: That’s possible. Every circuit has panel disagreements, but seems more salient for the Fed. Cir. because everyone thought it could produce coherence. Fewer judges initially; fewer judges produce more coherence, even without en bancs—if you eat lunch together every day, you don’t need en bancs. But now the Fed. Cir. is too big for that, and if you still want them to produce coherence, you need more en bancs.