When a Network Trick Isn't an Abstract Idea
Two federal appeals courts, ruling the same day in June, each had to decide what a dispute was really about beneath the way the parties dressed it up. The Federal Circuit preserved a jury's conclusion that several wireless building-automation patents claim real engineering rather than an "abstract idea," while ordering a new trial on one patent because the jury was never told which abstract idea it was supposed to be measuring against.[1] The Ninth Circuit revived a smoke-shop owner's racketeering suit against three tribal officials, holding that a demand for money from their own pockets is a case against them personally, not against the tribe whose immunity they tried to borrow. The sharper reasoning is the Federal Circuit's, which refused to let a defendant win by restating a specific network mechanism at a level of generality high enough to make anything sound abstract.[1]
The setup
Ollnova, a patent-holding company, accused ecobee, a maker of internet-connected thermostats, of infringing a group of patents on wireless building-automation systems, the networked controls that run heating, lighting, and sensors inside a building.[4] A federal jury in the Eastern District of Texas agreed, found infringement, and awarded a single lump sum of $11.5 million.[5] ecobee appealed to the Federal Circuit, the court that hears every U.S. patent appeal, challenging both whether the patents claim anything the law will protect and whether the trial was run correctly.[1]
The question
Patent law refuses to protect an "abstract idea," a category the courts use to keep basic ways of organizing information in the public domain even when a computer carries them out.[2] The test comes from the Supreme Court's 2014 decision in Alice Corp. v. CLS Bank, which asks two questions in order: first, is a claim directed to an abstract idea; and if so, does it add an "inventive concept," something beyond well-understood, routine, and conventional steps, that turns the idea into a genuine application.[3] Everything here turned on how much a court may compress a claim before judging it, because almost any invention sounds abstract if it is described loosely enough.[1]
The reasoning, walked
Writing for a unanimous panel, Judge Chen took the patents in groups.[1] Two of them cleared the first Alice step outright, meaning the court held they are not directed to an abstract idea at all and never had to reach the second question.[1] One claims a controller that polls a sensor only during set intervals, transmits only during set intervals, and sends an update only when a reading drifts outside a preset range; another claims bundling separate change-of-value reports from several devices into one message and resending it until the network acknowledges receipt.[4] Read as a whole, the court said, these are not generic instructions to collect and send data but specific ways to cut bandwidth, save power, and survive dropped connections on a constrained wireless network, in other words a "technological solution to a technological problem."[1]
The load-bearing move was the court's refusal to accept ecobee's higher-altitude description of the claims.[1] ecobee cast them as nothing more than collecting, analyzing, and selectively communicating information, a phrasing that would have pulled them toward the abstract-idea category.[6] The court rejected that wording as untethered from the actual claim language, contrasting these patents with an earlier case whose claims merely gathered and displayed power-grid data without changing how anything operated.[1] The distinction the court drew is between describing a result and reciting a particular mechanism that achieves it, and here the claims specify the mechanism, the timing rules, the message aggregation, the resend-until-acknowledged loop.[4] That classification did the decisive work, because a claim that survives step one is eligible with no further inquiry.[3]
The third patent was harder.[1] The trial court had already found it directed to an abstract idea, so it could survive only if a jury found the missing inventive concept at step two, and the disputed point was factual: whether its design, two wireless networks arranged so that one keeps controlling equipment locally even after it loses contact with the other, was conventional.[5] The jury found it was not, and the Federal Circuit held that enough evidence supported that finding to deny judgment as a matter of law, the procedure by which a court overrides a verdict no reasonable jury could have reached.[1] Yet the panel still vacated that part of the verdict, because the jury instructions never named the abstract idea and never told jurors the rule that the abstract idea itself cannot be the inventive concept.[6] Without the idea identified, the court reasoned, the jury could credit the very thing the law excludes, and the patent owner's own expert had described the inventive concept in terms that tracked the abstract idea almost exactly, so the error was not harmless.[1]
A separate flaw reached the entire infringement verdict.[1] The jury had answered one combined question, whether ecobee infringed any of the asserted claims, across all the patents at once.[5] That form, the court held, let the verdict rest on jurors who never agreed on which patent ecobee infringed, some perhaps convinced of one and some of another, in tension with the rule that a civil jury must reach a unanimous verdict on each distinct claim.[1] The court vacated the infringement judgment and, with it, the $11.5 million award, and returned the case for a new trial on separate questions.[5]
What turns on it
For anyone who builds in software or connected hardware, the opinion is a working map of which claims survive the abstract-idea filter that kills so many computing patents.[4] A claim tied to a concrete mechanism, specific timing, message aggregation, a fallback mode that keeps a device running when the network fails, reads as an improvement to how the machine works, while the same invention described as handling information in the abstract invites invalidation.[1] The practical lesson is that the level of generality is the battleground, and that drafting which anchors claims to a named technical problem in the specification pays off years later in court.[6]
As reasoning, the step-one analysis is careful and closely tied to the claim text, and its comparisons to prior cases are apt.[1] Its soft spot is the test itself, because calling something a "technological solution to a technological problem" is a judgment rather than a deduction, and a differently inclined panel could re-describe the same claims one notch more abstractly and land the opposite way.[3] The jury-instruction holding is the cleaner logic, since an instruction that asks jurors to find an inventive concept while hiding the abstract idea from them is close to incoherent on its own terms.[6] This is an assessment of how the court reasoned, not legal advice.
- Primary: Ollnova Technologies Ltd. v. ecobee Technologies ULC, No. 25-1045 (Fed. Cir. June 4, 2026), precedential slip opinion. The opinion under close reading; source of the Alice step-one eligibility holdings, the step-two remand, the verdict-form unanimity holding, and the disposition.
- 35 U.S.C. § 101, the patent-eligibility statute. Supplies the baseline rule that abstract ideas are not patentable, the provision the whole Alice inquiry interprets.
- Alice Corp. v. CLS Bank Int'l, 573 U.S. 208 (2014). The controlling two-step framework (abstract idea, then inventive concept) that structures every step of the panel's analysis.
- Justia opinion page for Ollnova v. ecobee (Fed. Cir. June 4, 2026). Independent mirror confirming the panel, the June 4 date, the asserted patents, and the disposition; source for the claim-language descriptions.
- Patently-O, 'What the Verdict Might Have Said: Jury Black Boxes in Ollnova v. ecobee.' Details the single combined verdict question, the $11.5 million award, and the Eastern District of Texas posture; load-bearing for the unanimity walk.
- Fenwick, 'Ollnova v. ecobee: Federal Circuit Reaffirms the Requirements for Patent Jury Instructions and Verdict Forms.' Practitioner analysis of the step-two jury-instruction defect and its drafting lessons.
Suing the Officer, Not the Tribe
The setup
Kyle and Jill Welsh, members of the Colorado River Indian Tribes, ran a smoke shop through their company, WW Young Money, in a commercial plaza on the tribes' reservation near Parker, Arizona, under a lease first signed in 2015 and renewed in 2020.[1] In 2021 the tribes' chairwoman and attorney general terminated the lease, citing unpaid rent and code violations along with a reference to a disturbance and a high-speed chase involving Kyle.[1] The Welshes told a different story, that the officials, together with the plaza's manager, had engineered the eviction to seize the business and its inventory for themselves.[1] They sued the three officials in federal court under the civil half of the Racketeer Influenced and Corrupt Organizations Act, or RICO, a statute that lets a private plaintiff recover triple damages from an "enterprise" run through a pattern of crimes, and they demanded millions from the officials personally.[3]
The trial court never reached whether any of that happened.[1] It dismissed the case on two threshold grounds, that tribal sovereign immunity shielded the officials, and that the tribes themselves were a required party who could not be joined because of their own immunity, leaving no way for the suit to go forward.[1] The Welshes appealed to the Ninth Circuit.[1]
The question
Indian tribes are sovereigns, and like other sovereigns they cannot be sued without their consent, a protection that normally extends to tribal agencies and to officials sued for acts of the tribe.[1] But the Supreme Court held in Lewis v. Clarke, a 2017 case about a tribal casino's limousine driver who caused a highway crash, that a suit against an official in an individual capacity, seeking money from the official rather than the treasury, is a suit against the person and not the sovereign.[2] The question was which kind of suit this was, a genuine claim against three individuals, or a claim against the tribes wearing the officials as a disguise, given that the alleged wrong was the termination of a tribal lease.[1]
The reasoning, walked
Writing for a unanimous panel, Judge Hurwitz started from the rule that immunity follows the real party in interest, not the label on the complaint.[1] Under Lewis v. Clarke the controlling question is "whether the remedy sought is truly against the sovereign," which turns on two things, whether the judgment would bind the tribe and whether any money would come from the tribe's own funds.[2] The Welshes had sued the officials in their individual capacities and sought damages payable from the officials' personal assets, not from tribal accounts, and they did not ask the court to restore the lease or order the tribes to do anything.[1] On that record, the panel held, the officials could not borrow the tribes' immunity, because a judgment would run against them alone.[1]
The officials' best argument tried to collapse that distinction.[1] Because the Welshes measured their damages by the value of the lost lease, the officials said, any recovery would in substance reach into a tribal instrument and make the tribes the real defendant.[1] The court refused to let the yardstick for damages decide the identity of the defendant.[1] What matters, the panel reasoned, is whom the judgment legally binds and whose money satisfies it, not how the plaintiff calculates the number, so a personal judgment measured by reference to the lease is still a personal judgment.[2]
That left the second dismissal, under the rule governing absent parties.[4] Rule 19 asks whether someone not in the case is a "required" party, and if so whether the case can proceed in fairness without them, dismissing the suit when it cannot and they cannot be joined.[4] The trial court had treated the tribes as such a party, joinable only in theory because of their immunity.[1] The Ninth Circuit disagreed, distinguishing its earlier decisions in which an absent tribe was necessary because the plaintiff was trying to undo or reinstate a lease and thereby alter the tribe's own rights.[1] Here the Welshes sought neither, only damages from three people, so the tribes' contractual and property interests were not on the line and their presence was not required.[1]
The panel reversed both dismissals and sent the case back.[1] Crucially, it decided nothing about whether the officials actually ran a racketeering enterprise or whether the eviction was wrongful, and those questions remain open.[1] On remand the district court must decide whether the complaint states a valid RICO claim at all, a demanding standard the Welshes may yet fail, and whether to let them amend.[6]
What turns on it
The ruling marks a practical limit on tribal immunity that plaintiffs can use and officials must reckon with.[5] Frame a suit as a personal claim for money, avoid asking a court to undo a tribal act or reach tribal funds, and the immunity that would bar a suit against the tribe falls away from its officers.[2] The reach is narrow, since the case cleared only the courthouse door and the hard RICO questions are still ahead, but the door is now open where it had been shut.[7]
The panel's logic follows almost mechanically from Lewis v. Clarke once the suit is read as individual-capacity, and on that reading the result is hard to fault.[2] The vulnerability sits where the two cases quietly rhyme, since each turns on refusing to let a party recharacterize what sits in front of the court, ecobee enlarging a concrete mechanism into an abstract idea, the tribal officials deflating a personal suit into a claim against the sovereign, with the court each time insisting on the concrete characterization.[1] The weaker link is here rather than in the patent case, because when the very act complained of is an official termination of a tribal lease, a damages award reconstructing that lease's value does press on tribal self-governance in a way the "who pays" test does not fully capture, and a skeptic could reasonably want that effect weighed rather than assumed away.[1] That is a comment on the structure of the reasoning, not legal advice.[1]
- Primary: Welsh v. Loudbear, No. 25-5475 (9th Cir. June 4, 2026), published slip opinion. The opinion under close reading; source of the facts, the individual-capacity immunity holding, the Rule 19 analysis, and the reversal and remand.
- Lewis v. Clarke, 581 U.S. 155 (2017). The controlling Supreme Court test the panel applies: immunity follows the real party in interest, so an individual-capacity damages suit against a tribal official is not barred by the tribe's immunity.
- 18 U.S.C. § 1964(c), the civil RICO provision. Supplies the private treble-damages action the Welshes brought against the officials.
- Federal Rule of Civil Procedure 19. Governs when an absent party is 'required' and when a case must be dismissed for failure to join one; the basis for the second dismissal the panel reversed.
- Snell & Wilmer, 'Ninth Circuit Rules That Individual Tribal Defendants May Not Be Entitled to Sovereign Immunity Protection.' Practitioner analysis of Welsh v. Loudbear and the practical limit it places on tribal immunity.
- Justia opinion page for Welsh v. Loudbear (9th Cir. June 4, 2026). Independent mirror confirming the docket number, the date, the disposition, and the remand for a Rule 12(b)(6) test of the RICO claim.
- Law360, '9th Circ. Revives Tribal Smoke Shop Owner's RICO Suit.' News confirmation of the reversal and the posture of the revived claim.