Patent Eligibility Under Section 101: The Alice/Mayo Framework and How Courts Apply It

Document type: Article Practice area: Intellectual Property — Patents Jurisdiction: United States (federal) Last reviewed: 5 September 2026

Here is the whole statute:

Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.

Fifty-two words. 35 U.S.C. § 101 contains no exceptions, no limitations, and no framework. It reads like a welcome mat.

And yet a substantial fraction of patent cases now end on a motion to dismiss, before an answer, before discovery, before the court has construed a single claim term — because the Supreme Court has read into those fifty-two words three judicial exceptions and a two-step test that a great many issued patents do not survive.

Understanding why requires holding two ideas at once. The first is that the exceptions are old, sensible, and not going anywhere: nobody should be able to patent the equation E=mc², the discovery that a mineral occurs naturally in a certain ore, or the abstract concept of hedging risk. The second is that translating that sensible principle into a workable test has proved genuinely difficult, and the resulting doctrine is the least predictable area of substantive patent law.

For practitioners, the practical consequences are enormous. A defendant with a strong eligibility argument can end a case for a fraction of what discovery would cost. A patentee who has not thought about eligibility before filing suit may lose a portfolio it spent a decade building. And a prosecutor drafting claims today is drafting into a doctrine whose contours shift with the panel.

The three judicial exceptions

Courts have long held that § 101 does not reach laws of nature, natural phenomena, and abstract ideas.

The rationale is preemption. These are, in the Court's phrase, "the basic tools of scientific and technological work." Granting exclusive rights over them would impede rather than promote innovation, because every subsequent inventor in the field would need a license to the building block rather than to any particular application of it.

The exceptions are old. They predate the 1952 Act and were applied consistently through the twentieth century:

Gottschalk v. Benson, 409 U.S. 63 (1972) held ineligible a claim to a method of converting binary-coded decimal numerals to pure binary numerals using a general-purpose computer. The claim, in substance, was to the algorithm itself.
Parker v. Flook, 437 U.S. 584 (1978) held ineligible a method of updating alarm limits in catalytic conversion using a mathematical formula, where the only novelty was the formula. Post-solution activity did not save it.
Diamond v. Chakrabarty, 447 U.S. 303 (1980) held eligible a genetically engineered bacterium capable of breaking down crude oil, in the phrase that launched a generation of patent applications: the statute reaches "anything under the sun that is made by man." The bacterium was a human-made composition with markedly different characteristics from anything in nature.
Diamond v. Diehr, 450 U.S. 175 (1981) held eligible a process for curing synthetic rubber that used the Arrhenius equation to determine cure time. The claim was not to the equation; it was to an industrial process that used it, and the Court emphasized that a claim must be considered as a whole.

The tension between Flook and Diehr — one saying you cannot claim a formula by adding conventional steps, the other saying you must consider the claim as a whole — is the tension the modern framework attempts to resolve, and it is why the modern framework is difficult.

The modern framework

Bilski: the machine-or-transformation test is a clue, not a test

Bilski v. Kappos, 561 U.S. 593 (2010) addressed a claimed method of hedging risk in commodities trading. The Federal Circuit had held that the sole test for a process claim was whether it is tied to a particular machine or transforms an article into a different state or thing — the machine-or-transformation test.

The Supreme Court affirmed the result and rejected the exclusivity of the test. Machine-or-transformation is "a useful and important clue, an investigative tool," but not the sole test. Nor is there a categorical business-method exclusion. The claims were ineligible because hedging is an abstract idea, and limiting it to the energy market and adding well-known mathematics did not change that.

What Bilski left: no workable test, and a strong signal that claiming a fundamental economic practice does not work.

Mayo: the two-step framework arrives

Mayo Collaborative Services v. Prometheus Laboratories, Inc., 566 U.S. 66 (2012) is the origin of the framework everyone now applies, even though the case is about diagnostics rather than software.

The claims covered a method of optimizing thiopurine drug dosage: administer the drug, determine the level of a metabolite, and if the level falls outside a stated range, the need to adjust the dose is indicated.

The Court held the claims ineligible. The relationship between metabolite level and efficacy or toxicity is a law of nature. The remaining steps — administer, determine, and a "wherein" clause telling a doctor to think about the result — were, individually and as an ordered combination, "well-understood, routine, conventional activity previously engaged in by researchers in the field."

The framework the Court articulated: a claim directed to a judicial exception must contain "other elements or a combination of elements ... sufficient to ensure that the patent in practice amounts to significantly more than a patent upon the natural law itself" — an "inventive concept."

Why Mayo is more consequential than Alice. It imported a novelty-flavored inquiry into § 101, asking whether the additional elements are conventional. That question sounds like §§ 102 and 103, and the resulting overlap has been criticized by judges, practitioners, and the Federal Circuit itself. It is nonetheless the law.

Myriad: isolating a natural product is not enough

Association for Molecular Pathology v. Myriad Genetics, Inc., 569 U.S. 576 (2013) held that isolated naturally occurring DNA segments are not eligible — separating a gene from its surrounding genetic material does not create anything new — while cDNA is eligible, because it is not naturally occurring: the lab creates a molecule with the introns removed.

The line is not about effort or ingenuity. It is about whether the claimed thing is markedly different from what nature provides.

Alice: the framework applies to everything

Alice Corp. v. CLS Bank International, 573 U.S. 208 (2014) applied Mayo's framework to computer-implemented claims and produced the two-step test now universally cited:

Step one. Determine whether the claims are directed to a patent-ineligible concept — a law of nature, natural phenomenon, or abstract idea.

Step two. If so, consider the elements of each claim, individually and as an ordered combination, to determine whether the additional elements transform the nature of the claim into a patent-eligible application — the search for an inventive concept.

The claims covered mitigating settlement risk using a computer as an intermediary. Step one: intermediated settlement is a fundamental economic practice, an abstract idea. Step two: implementing it on a generic computer is not enough. "Simply appending conventional steps, specified at a high level of generality," does not supply an inventive concept, and neither does "mere recitation of a generic computer."

The sentence that decides most software cases: the claims must do "more than simply state the abstract idea while adding the words 'apply it with a computer.'"

Step one in practice: what is the claim "directed to"?

Step one sounds simple and is where most of the disagreement lives, because every claim can be described at multiple levels of abstraction.

The level-of-abstraction problem. A claim to a specific method of encoding video using a particular quantization scheme can be described as "directed to" compressing data, which is abstract — or as directed to a specific technical improvement in video encoding, which is not. Both descriptions are accurate. The framework does not tell you which to use.

The Federal Circuit's working answer is to ask whether the claim's focus is a specific improvement in the functioning of a computer or other technology, or whether the computer is merely a tool used to carry out an abstract process.

Claims that have tended to survive step one:

Improvements to computer functionality itself — a self-referential database structure that improves how a computer stores and retrieves data; a memory system with configurable operation; a method that improves the way a display renders.
Improvements to a technological process — a specific method of animating lip synchronization using rules applied to phoneme sequences; a particular technique for filtering network content at a specific network location.
Claims with specific technical detail in the claim itself, not merely in the specification.

Claims that have not:

Collecting, analyzing, and displaying information. This formulation has become almost a per se ineligibility finding, and it describes a great many claims.
Organizing human activity — managing a game, mitigating risk, matching consumers with providers, verifying identity.
Fundamental economic and business practices — hedging, escrow, advertising, loyalty programs.
Mathematical relationships and formulas untethered to a technical application.

The practical lesson for step one: the claim must recite the improvement, not merely benefit from it. A specification describing a genuine technical breakthrough does not save a claim that recites the result in functional terms. Courts read the claim, then look to the specification to understand what the claim means — not to import limitations the claim does not contain.

Step two: the search for an inventive concept

If a claim is directed to an exception, step two asks whether the additional elements, individually and as an ordered combination, add "significantly more."

What does not supply an inventive concept:

Generic computer implementation. A processor, memory, a database, a network, a display, a server-client architecture.
Field-of-use limitations. Applying the abstract idea "in the field of energy markets" or "for medical records."
Insignificant pre- or post-solution activity. Gathering the data the algorithm operates on; displaying the result.
Well-understood, routine, conventional activity, which Mayo treats as adding nothing.

What can:

A non-conventional and non-generic arrangement of known components. The ordered combination matters, and a claim whose specific arrangement solves a technical problem can survive even where each component is individually known.
A specific technical solution to a technical problem rooted in the technology, particularly where the specification explains what was wrong with prior approaches and how the claim fixes it.
Unconventional steps confirmed as such by the specification or by evidence.

The uncomfortable overlap with novelty. Step two asks whether elements are "well-understood, routine, and conventional." That is close to asking whether they are new. Judges and commentators have criticized this as importing §§ 102 and 103 into § 101, and the Federal Circuit has acknowledged the tension while continuing to apply the framework. The practical consequence is that eligibility arguments and invalidity arguments draw on overlapping evidence, and a defendant should build both together.

Berkheimer and the factual question

For several years after Alice, district courts routinely resolved eligibility on the pleadings, treating the entire inquiry as a question of law.

The Federal Circuit's 2018 decisions in Berkheimer v. HP and Aatrix Software v. Green Shades changed the practice. They held that while eligibility is ultimately a question of law, it may involve underlying questions of fact — specifically, whether a claim element or combination is well-understood, routine, and conventional to a skilled artisan at the relevant time. Where the specification or the record raises a genuine factual dispute on that question, it cannot be resolved on a motion to dismiss or on summary judgment.

What this changed operationally:

Patentees learned to plead facts. A complaint that alleges, with specificity, that the claimed combination was unconventional — quoting the specification's description of the problem and the solution — is materially harder to dismiss.
The specification became more important. A specification that explains what was conventional and how the invention departs from it supplies the factual predicate; one that describes the invention in purely functional terms does not.
Defendants moved later, or moved with evidence. A Rule 12 motion is still available and still frequently granted, but a defendant facing a well-pleaded complaint may need a record — prior art, a declaration, admissions — and may be better served by summary judgment.

What it did not change. Where the claim is directed to a plainly abstract idea and the additional elements are, on the face of the claim, generic computer components, courts still dismiss at the pleadings. Berkheimer created a factual off-ramp; it did not close the highway.

The USPTO applies a different framework

This is the source of a great deal of client confusion, and it is worth stating plainly: the Patent Office and the courts do not apply the same test.

The USPTO's eligibility guidance, first issued in 2019 and revised since, structures the analysis differently:

Step 2A, Prong One: does the claim recite a judicial exception, categorized into enumerated groupings — mathematical concepts, certain methods of organizing human activity, and mental processes?
Step 2A, Prong Two: if so, does the claim integrate the exception into a practical application? This prong has no direct analogue in the judicial framework, and it is where most applications succeed. Integration is shown by an improvement to the functioning of a computer or another technology, application to a particular machine, transformation of an article, or a meaningful limitation beyond generally linking to a field of use.
Step 2B: only if the exception is not integrated does the examiner reach the inventive-concept inquiry.

Why this matters:

A patent can issue under the Office's framework and be held ineligible in court. This happens regularly, and it surprises clients who assume examination settled the question.
Prosecution arguments built on Prong Two integration do not automatically travel to litigation, where there is no Prong Two.
The Office's guidance is not binding on courts. The Federal Circuit has said so directly.

The practical instruction for prosecutors: use Prong Two to get the patent allowed, and separately ask whether the claim would survive the judicial framework. If the answer is no, the claim may be an asset for licensing and a liability for litigation.

Diagnostics and life sciences: the harder problem

Software eligibility is unpredictable. Diagnostic method eligibility is close to a dead letter, and the reason is structural.

The shape of a diagnostic claim is almost always: detect a marker, and correlate the marker with a condition. The correlation is the discovery, and the correlation is a natural relationship. Under Mayo, the detection step is conventional and the correlation is the exception, so the claim fails.

The Federal Circuit has said, repeatedly and often in concurrences, that it regards this outcome as compelled by Mayo and undesirable. In Athena Diagnostics v. Mayo Collaborative Services, the court denied rehearing en banc over eight separate opinions, several of which said explicitly that the panel decision was correct under binding precedent and that the precedent should change. The Supreme Court has declined repeated invitations to revisit the question.

What has worked in life sciences:

Method-of-treatment claims. A claim that administers a specific drug at a specific dose to a patient identified by a marker has been held eligible — the claim is to a treatment, not to the correlation. This is now the standard drafting response.
Novel detection techniques. Where the detection step itself is new and specific, rather than conventional assay technology, the claim can survive.
Compositions that are markedly different from natural products, following Myriad and Chakrabarty.

What has not:

Diagnostic correlations, however valuable the discovery.
"Detect and correlate" claims dressed with conventional laboratory steps.
Claims to naturally occurring correlations limited to a patient population.

The strategic consequence. Diagnostic innovators increasingly rely on trade secret protection for algorithms and reference databases, on regulatory exclusivity, and on method-of-treatment claims where a therapeutic intervention exists. Where the invention is purely informational, patent protection is unreliable.

Procedure: when and how eligibility gets decided

Rule 12(b)(6) or 12(c). The most common vehicle, and still frequently successful. The court takes the complaint's allegations as true and construes the claims, if necessary, in the light most favorable to the patentee.

Representative claims. Courts routinely allow a defendant to address a representative claim where the claims are substantially similar and linked to the same abstract idea. A patentee that wants claim-by-claim treatment must identify specific limitations in other claims that it contends change the analysis — a generic assertion that the claims differ is insufficient.

Claim construction. Eligibility can be decided before construction where the analysis does not turn on a disputed term. Where it does, the patentee must identify the term, propose a construction, and explain why that construction changes the eligibility outcome. Courts frequently proceed by adopting the patentee's proposed construction for purposes of the motion — which, if the patentee's construction still fails, ends the case efficiently.

Summary judgment. The vehicle when Berkheimer factual disputes exist and the record can resolve them, or when the defendant wants a record of conventionality — prior art, technical declarations, admissions in the specification or prosecution history.

Trial. Rare, and awkward: eligibility is a legal question with factual underpinnings, so the jury may be asked to resolve the conventionality question by special verdict while the court decides eligibility. Most cases do not get here.

The PTAB. Eligibility cannot be raised in an inter partes review, which is limited to §§ 102 and 103 grounds based on patents and printed publications under 35 U.S.C. § 311. It can be raised in a post-grant review under § 321, but PGR is available only for patents with an effective filing date on or after March 16, 2013 and only within nine months of issuance. For most patents in litigation, § 101 is a district court issue.

Appeal. Reviewed de novo as a question of law, with underlying factual findings reviewed for clear error following Anderson v. City of Bessemer City, 470 U.S. 564 (1985) principles.

Two worked examples

The software claim that survived

Larkspur Networks asserts a patent on a method of managing bandwidth in a mesh network. Claim 1 recites: at each node, measuring queue depth at defined intervals; computing a local congestion metric from the queue depth and a decay factor; broadcasting the metric only to nodes within two hops; adjusting the local transmission rate according to a specified function of the received metrics; and repeating.

The defendant, Cheswick Systems, moves to dismiss, arguing the claim is directed to the abstract idea of managing network traffic based on collected information.

Why the motion fails.

Step one. The court finds the claim directed to a specific improvement in network functioning rather than to an abstract concept. The claim does not recite "managing congestion"; it recites a particular distributed mechanism — two-hop broadcast, a specific decay function, a specific rate adjustment — that solves a problem the specification identifies with centralized approaches. The claim's focus is the technical mechanism.

What made the difference. Three things. The specification's background section described the prior centralized approach and its latency problem in technical detail. The claim recited the mechanism rather than the result. And the complaint alleged, with citations to the specification, that the two-hop limited broadcast was unconventional at the priority date.

Note what did not matter: that the claim is implemented in software, that it involves collecting and using information, and that each individual step uses known techniques.

The business-method claim that did not

Fennimore Holdings asserts a patent on a method of processing insurance claims. Claim 1 recites: receiving claim data at a server; comparing the claim data to stored policy parameters; assigning a risk score using a weighted formula; routing the claim to an adjuster queue based on the risk score; and generating a report.

Step one. The court finds the claim directed to the abstract idea of evaluating and routing insurance claims — a method of organizing human activity that insurers performed with people and paper for a century.

Step two. A server, stored data, a weighted formula, a queue, and a report. Each is generic. The ordered combination is the ordinary sequence of doing the task. The specification describes the invention in functional terms and does not identify a technical problem or a technical solution.

The patentee's arguments and why they failed:

"The weighted formula is novel." Novelty is not an inventive concept, and a formula is itself an exception.
"The claim improves efficiency." Doing an abstract process faster with a computer is the paradigm of what does not qualify.
"There are factual disputes under Berkheimer." The court finds no genuine dispute: the specification itself describes the components as conventional, which is the fact pattern Berkheimer distinguishes.
"Claim construction is required." The patentee did not identify a term whose construction would change the outcome.

Dismissed with prejudice, four months after the complaint, for roughly $190,000 in defense costs against a case that would have cost $3 million through trial.

What this means for claim drafting

Eligibility is decided years later, on the claim as written. Drafting is where it is won or lost.

Recite the mechanism, not the result. "Determining a congestion metric" is a result. "Computing a congestion metric as the product of measured queue depth and a decay factor applied over a sliding window of N intervals" is a mechanism. The second survives motions the first does not.

Write a background section that identifies a technical problem. Not a business problem — a technical one. What did prior systems do, what was wrong with it technically, and what does the invention change? This section supplies the Berkheimer factual predicate and the step-one framing.

Say what is unconventional, in the specification. A specification that describes the combination as departing from conventional practice, and explains how, creates a factual question. A specification that says "as is well known in the art, servers store data in databases" hands the defendant an admission.

Avoid the words that invite trouble. "Determining," "analyzing," "processing," "generating a report," and "displaying" describe information handling. They are not fatal, but a claim built entirely from them will be characterized as collecting, analyzing, and displaying information.

Claim at multiple levels. File claims of varying scope. The broadest claim may be the most valuable and the least eligible; a narrower claim reciting specific technical implementation may be the one that survives.

In life sciences, claim the treatment. Where a therapeutic intervention exists, a method-of-treatment claim reciting administration of a specific agent at a specific dose to a patient identified by the marker is the reliable path. A pure diagnostic correlation claim is not.

Consider whether to file at all. For a purely informational innovation with no technical implementation, trade secret protection may be worth more than an unenforceable patent — and it does not require disclosure.

Draft the complaint at the same time as the claims. A useful discipline: before filing the application, write the paragraph you would need in a complaint to survive a Rule 12 motion. If you cannot write it from the specification you have, revise the specification.

What this means in litigation

For a defendant:

Screen every asserted patent for eligibility in the first two weeks. It is the cheapest available defense and it is the only one that can end a case before discovery.
Read the specification before the claims. A specification that describes the components as conventional is the best evidence you will get, and it is an admission.
Move early, but move with awareness of Berkheimer. If the complaint pleads unconventionality with specificity and the specification supports it, a Rule 12 motion may fail and a better vehicle is summary judgment on a developed record.
Address representative claims properly. Identify why the claims are substantially similar; do not simply pick claim 1 and hope.
Adopt the patentee's construction for the motion where it does not change the outcome. It removes the "claim construction is required" answer.
Build the § 101 and § 103 records together. The conventionality evidence overlaps substantially.

For a patentee:

Assess eligibility before filing suit, honestly. Assert the claims most likely to survive, not the broadest.
Plead the facts. Allege what was conventional at the priority date, what the invention changed, and why the combination was unconventional — with citations to the specification.
Identify the technical problem and the technical solution in the complaint, in the patentee's own words, before the defendant characterizes the claim at the highest level of abstraction.
If claim construction matters, say which term and why, with a proposed construction and an explanation of how it changes the analysis.
Consider the portfolio consequence. A published decision holding your claims ineligible affects every related patent and every licensing conversation. Choosing the wrong case as the first case is expensive.

Where the doctrine may go

Three developments are worth watching, and clients ask about all of them.

Legislative reform. Bills to amend § 101 — typically by abrogating the judicial exceptions and replacing them with a statutory list of excluded subject matter — have been introduced repeatedly, with support from the life sciences sector and opposition from parts of the technology sector. None has passed. The disagreement is genuine: the sectors have opposite interests, and the exceptions do real work in preventing preemption.

Supreme Court revisiting. The Court has denied certiorari in a long line of eligibility cases, including several where the Solicitor General recommended review. The Federal Circuit has effectively asked for guidance in multiple opinions. Predicting when this changes is not a strategy.

Federal Circuit refinement. The most likely near-term movement. The court has developed a body of case law identifying the fact patterns that survive step one — improvements to computer functionality, specific technical solutions — and each new decision adds a data point. Practitioners follow this closely because it is where the doctrine actually moves.

What to tell a client. The framework is unlikely to change soon. Draft and litigate for the law as it is, build specifications that supply the factual predicate, and treat a broad functional claim as a licensing asset rather than a litigation asset until it has survived a motion.

The criticism, fairly stated

It is worth setting out the objections to the doctrine, because clients raise them and because the objections are serious.

It collapses distinct statutory requirements. Section 101 asks what subject matter is patentable. Sections 102 and 103 ask whether a particular claim is new and non-obvious. Section 112 asks whether the disclosure supports the claim. Step two's "well-understood, routine, and conventional" inquiry asks a novelty question under the eligibility heading — without the procedural protections that accompany a §§ 102 and 103 analysis, notably the requirement that the challenger identify prior art.

It is applied inconsistently. The same claim can be characterized at a level of abstraction that dooms it or one that saves it, and the framework provides no principle for choosing. Practitioners can name panels whose eligibility outcomes are predictable, which is a criticism of a legal test.

It falls hardest on the sectors that most need patents. Diagnostics is the clearest case: an enormously valuable discovery about a biological correlation is, under Mayo, unpatentable in its most natural claim form. Multiple Federal Circuit judges have said in print that this is bad policy compelled by binding precedent.

It resolves cases before the record exists. A doctrine applied on the pleadings decides validity without prior art, without expert testimony, and often without claim construction.

The defense of the doctrine, also fairly stated. The exceptions do real work. Before Alice, patents claiming ordinary business processes performed on computers issued in volume and were asserted broadly against defendants with no realistic path to invalidity short of trial. The framework gave district courts a tool to end those cases early, and it substantially reduced a category of litigation that most observers regarded as abusive. Whatever its analytical difficulties, it addressed a real problem.

What a practitioner does with this. Not much, directly — the doctrine is the doctrine. But understanding the criticism explains why the case law moves, why the Federal Circuit writes concurrences asking for review, and why a client whose valuable invention is held ineligible is not being unreasonable when they say the result makes no sense. Sometimes it does not.

Frequently asked questions

Our patent issued. Doesn't that mean it's eligible? No. The Patent Office applies its own guidance, including a practical-application prong that has no counterpart in the judicial framework, and the guidance does not bind courts. Patents issue and are then held ineligible with some regularity.

Can we fix an eligibility problem after issuance? Sometimes. A continuation application with claims reciting more technical specificity may produce a patent that survives. Reissue is available in limited circumstances. What you cannot do is add technical detail that the original specification does not support — 35 U.S.C. § 112 written description and enablement constrain what claims the specification can carry.

Is software patentable? Yes, and a great deal of software is patented and enforced. What is not patentable is an abstract process implemented on a generic computer. The question is whether the claim recites a specific technical mechanism or a result.

Does adding hardware save a claim? Generic hardware does not. A claim requiring a particular machine or transforming an article is more likely to survive — the machine-or-transformation test remains, in Bilski's words, "a useful and important clue" — but it is not dispositive either way.

Can eligibility be raised in an IPR? No. Section 311 limits inter partes review to §§ 102 and 103 grounds based on patents and printed publications. Post-grant review under § 321 reaches § 101, but only for patents with an effective filing date on or after March 16, 2013 and only within nine months of issuance.

When should a defendant move? As early as the record permits. Screen in the first two weeks; move on the pleadings if the specification describes the components as conventional and the complaint does not plead unconventionality with specificity. Otherwise develop a record and move for summary judgment.

Does the presumption of validity apply? Section 282 provides a presumption of validity, and courts have divided on whether the clear-and-convincing standard applies to eligibility, which is a legal question. The practical answer is that the standard rarely decides the outcome; the claim language does.

What is the single best predictor of survival? Whether the claim recites a specific technical mechanism that the specification connects to a specific technical problem. Everything else is commentary.

The short version

Section 101 has three judicial exceptions — laws of nature, natural phenomena, and abstract ideas — grounded in preemption concerns that are old and sensible.

The two-step framework of Mayo and Alice asks first what the claim is directed to, and second whether the additional elements supply an inventive concept.

Step one is a level-of-abstraction problem with no formal answer; the operative question is whether the claim's focus is a specific technical improvement or an abstract process implemented on generic technology.

Step two asks whether the additional elements are well-understood, routine, and conventional, which imports a novelty-flavored inquiry that the courts acknowledge is uncomfortable and apply anyway.

Berkheimer made conventionality a factual question in some cases, which raised the bar for pleading-stage dismissals without eliminating them.

The USPTO applies a different framework with a practical-application prong the courts do not have, which is why patents issue and then fail.

Diagnostics are close to categorically ineligible absent a treatment step or a novel detection technique.

And drafting is where this is decided. The claim you write today is the claim a district judge will read in six years, without the benefit of your explanation.

A note on other jurisdictions

Clients with international portfolios ask why a claim allowed in Europe fails in the United States, and the answer is that the two systems ask different questions.

The European Patent Convention excludes from patentability, among other things, programs for computers, mathematical methods, business methods, and presentations of information — but only "as such." The EPO's approach then asks whether the claim has technical character, and if so, assesses inventive step considering only those features that contribute to a technical effect. The result is that a claim solving a technical problem is generally patentable, while a claim implementing a business method on conventional hardware fails at the inventive-step stage rather than at eligibility.

The practical difference. The EPO framework is more predictable, because "technical character" and "technical effect" have accumulated substantial doctrinal content. A claim to a specific improvement in how a device operates tends to succeed in both systems. A claim to a business process implemented on a computer tends to fail in both, though for different stated reasons.

Where they diverge most: diagnostics. Europe excludes diagnostic methods practised on the human body from patentability but permits claims to in vitro methods and to substances for use in such methods. The result is that some diagnostic inventions unpatentable in the United States under Mayo are protectable in Europe.

Drafting for both. A specification written to support the U.S. framework — identifying a technical problem, describing the prior approach and its technical shortcoming, and explaining the technical mechanism of the solution — is also the specification the EPO wants. The claim sets will differ, but the disclosure work is shared, and doing it once properly serves both.