Patent litigation is not merely a retrospective proceeding that determines winners and losers after infringement has already occurred. Patent decisions shape the R&D trajectory of companies that have yet to launch a product, define the permissible scope of a competitor's design-around, inform an investor's enterprise valuation, set the price of a license negotiation, determine collateral value in technology financing, and can even tilt the selection of technical standards across an entire industry. For this reason, the legal doctrine articulated through patent precedent must do more than resolve the particular dispute at hand. It must be specific enough for market participants to structure their future conduct and price their legal risks.

Patent litigation must not become a "panel roulette" game. The outcome of a patent dispute should not be something knowable only after litigation has commenced. Companies need a meaningful degree of predictability at the very moment they first decide to implement a patented technology — or to design around it. Businesses and markets do not treat unquantifiable uncertainty as a manageable risk. No company should stake its capital and R&D resources on a game whose outcome it cannot project. That is not investment in innovation; it is speculation on an unknowable outcome.

The phrase "panel roulette" is not an accusation of arbitrariness directed at any particular court. Rather, it identifies a structural risk: the risk that, given identical or closely similar claim language and prior art, the starting point and ultimate conclusion of claim construction, infringement under the doctrine of equivalents, non-obviousness, or damages analysis may vary depending on which panel happens to draw the case. As that structural risk grows, patent law is reduced from a forward-looking norm that guides commercial conduct to a backward-looking instrument that merely declares winners and losers after the fact. The law may resolve disputes, but it fails to prevent them.

1. The Boundaries of a Patent and the "Manifestly Unreasonable Literal Construction" Standard

Unlike real property, a patent has no physical boundary one can see and walk. It is a government-granted intangible exclusive right of limited duration, and its metes and bounds are understood, in principle, through the language of the claims. Article 97 of the Korean Patent Act provides that the scope of protection of a patented invention is determined by the matters stated in the patent claims. This is not a simple rule of textual interpretation; it is a public notice principle — the mechanism through which third parties learn how far a patent owner's exclusive rights extend. Article 42 of the same Act reinforces this notice function by governing the manner in which claims must be drafted and disclosed.

The Korean Supreme Court has repeatedly reaffirmed that the scope of protection of a patented invention is determined by the claims as written, and that claims may not be arbitrarily narrowed or broadened by resort to the specification or drawings. The landmark decision — Supreme Court Decision 2011Hu3230 (Dec. 27, 2012) — established the controlling standard: while claims must be construed in light of the specification and drawings to ascertain their full technical meaning, it is impermissible to narrow or expand claim scope based solely on the description or drawings. That ruling has become the baseline citation for all subsequent decisions in this line of authority. Supreme Court Decision 2021Da217011 (Jun. 30, 2021) reaffirmed the same principle and further held that a claim construction settled in patent invalidity trial proceedings (무효심판) — the Korean administrative equivalent of an inter partes validity challenge — must be applied consistently in a subsequent infringement action, reinforcing the requirement of analytical coherence across different types of proceedings.

That said, the plain-meaning rule is not without exception. In Supreme Court Decision 2023Hu11357 (Jul. 17, 2025), the Court held that where a literal reading of the claim would be "manifestly unreasonable" in light of other disclosures in the specification, the scope of the patent right may be construed more narrowly. The case arose from a negative scope confirmation trial (소극적 권리범위 확인심판) — a Korean patent administrative proceeding before the Intellectual Property Trial and Appeal Board (IPTAB) in which the accused party seeks a declaration that its product or process falls outside the scope of the registered patent. This proceeding is functionally analogous to a declaratory judgment action for non-infringement under U.S. law. What makes this decision significant is not merely the abstract recognition that narrowing construction is sometimes permissible. The value of the ruling lies in the relatively concrete showing it provides of what counts as a "manifestly unreasonable literal construction" — and under precisely what circumstances the exception is triggered.

In the case before the Court, the claim term "cleaning means" was, standing alone, facially broad enough to cover a means of sterilizing a storage tank by generating bactericidal substances through electrolysis. Nothing in the claim language expressly confined the cleaning means to a particular substance or a particular method of generation. Yet the Supreme Court did not treat the plain meaning of the claim term as dispositive.

First, the written description in the specification disclosed only one method of cleaning the storage tank: diluting a cleaning or bactericidal substance in filtered water and supplying the mixture to the tank. There was no disclosure — not even a suggestion — of any configuration that generates bactericidal substances through electrolysis. A reading of "cleaning means" broad enough to encompass the electrolytic configuration was therefore flatly unsupported by the specification.

Second, during prosecution of the parent application, the patentee filed a response asserting that the "cleaning means" of the parent did not include a means of generating bactericidal substances through electrolysis — specifically to distinguish prior art and overcome a rejection. The patent at issue was later filed as a divisional application tracing through multiple generations back to that parent. The Supreme Court held that, because the applicant in the parent and in the downstream divisionals was the same, and because the divisional fell within the scope of what was originally disclosed in the parent's specification and drawings, the prosecution history of the parent was properly considered in construing the claims of the later divisional.

Third, the Court concluded that the patentee had consciously excluded the electrolytic configuration from the scope of the divisional patent as well. Strictly speaking, the ruling is less a direct application of prosecution history estoppel in the doctrine-of-equivalents context than it is a determination to import the parent's prosecution history into claim construction of a descendant divisional. Nevertheless, because the ruling denied the patentee the ability to reclaim through a broad literal reading of a divisional claim what it had expressly surrendered during parent prosecution, the decision functionally embodies the logic of prosecution history estoppel and conscious exclusion.

The Supreme Court's reasoning proceeded through the following analytical sequence:

  1. Read in isolation, the claim term "cleaning means" is facially broad enough to encompass the electrolytic configuration.
  2. However, the specification discloses only the dilution method; the electrolytic configuration appears nowhere and is not even suggested.
  3. During parent prosecution, the applicant expressly excluded the electrolytic configuration in a filed response to a rejection.
  4. Given the common identity of the applicant and the technical and procedural continuity between the parent and the divisional, the parent's prosecution history properly informs claim construction of the divisional.
  5. Accordingly, a literal reading broad enough to include the electrolytic configuration conflicts with both the specification's written support and the applicant's objectively manifested intent.
  6. Construing "cleaning means" at its facial breadth is therefore manifestly unreasonable; the term must be confined to the dilution-based method the specification actually discloses.

The decision illustrates how to prevent an exception from swallowing the rule. Rather than simply declaring that "construing the claim at face value would be unreasonable given all the circumstances," the Court stepped through the analysis — the literal scope of the claim term, the specification's written support, the parent prosecution history, conscious exclusion, the continuity of the divisional lineage, and the reliance interests of third parties — in sequence. Third parties can follow that analytical path and, at least in cases presenting similar facts, predict both when the plain-meaning rule holds and when a narrowing exception is warranted.

The Court's treatment of the relationship with dependent claims also merits attention. Narrowing "cleaning means" as the Court did rendered the scope of the independent claim, as a practical matter, coextensive with the scope of the dependent claim that explicitly limited the cleaning means to the dilution-based configuration. The Court nonetheless prioritized the specification's disclosure, the applicant's objectively expressed intent, and the protection of third-party reliance over the formal argument that an independent claim must always be read broader than its dependents. In short, the mere existence of a dependent claim does not guarantee a broad construction of the independent claim from which it depends.

One should not, however, read this decision as a blanket rule that prosecution statements made in connection with a parent application invariably bind all divisional descendants. Whether the parent's prosecution history properly carries over to a divisional must be assessed on a case-by-case basis, taking into account: which claims and which rejections the parent's prosecution statements actually addressed; the degree of technical and procedural continuity between the parent and the divisional; and whether the statements constituted a clear and deliberate surrender of the excluded subject matter. Only when those conditions — and their outer limits — are articulated with precision can the exception be confined to its proper domain.

2. A Concrete Legal Standard Is Not About Outcomes — It Is About Making the Analytical Process Transparent

Legal certainty does not demand that precedent never change. As technology and market structures evolve, the case law must be capable of revision. What it requires is that when change occurs, the reasons are explained, the scope is defined, and the transition does not needlessly destroy the settled expectations of parties who acted in reasonable reliance on the prior framework.

Different facts may justify different outcomes. The critical question is whether divergent results across cases stem from genuine differences in the underlying facts, or from the fact that different adjudicators have poured different substantive content into the same abstract legal concepts. Only when the distinction is transparent can the system maintain its legitimacy.

At a minimum, the following questions should be answerable by reading the court's opinion:

  • What is the statutory text and governing principle from which the analysis begins?
  • What specific facts are outcome-determinative?
  • What evidence and technical rationale support each critical factual finding?
  • How are competing factors ranked, and how do they interact with one another?
  • Which factors are necessary conditions for the result, and which are merely corroborating?
  • What specific reason justifies recognizing an exception to the general rule?
  • Why does the conclusion hold even in the face of countervailing considerations?
  • If the result differs from a prior decision, what factual distinction accounts for the difference?
  • Where is the tipping point at which applying the same analytical framework to the next case would yield a different result?

Saying that an outcome was reached "by weighing all the relevant circumstances" is not sufficient. The more a case calls for a totality-of-the-circumstances analysis, the more essential it becomes to identify which factors must be examined first, which are dispositive, and what countervailing facts, if credited, would reverse a provisional conclusion. A long list of relevant considerations, without any account of how they relate to and constrain each other, is not the articulation of a legal standard — it is a catalog of discretion.

A concrete legal standard, in the end, is a publicly disclosed method of analysis that controls the level of abstraction in reasoning — an algorithm that any practitioner can trace through and apply to reach a predictable conclusion.

The word "algorithm" is not meant to suggest that judicial judgment can be replaced by mathematical formulas or mechanical computation. Factfinding, evidence evaluation, and technical judgment are inherent in any legal determination. The point, rather, is that the inputs, the sequence of analysis, the relationships among the relevant factors, and the conditions under which a provisional conclusion may be revisited must all be made visible to the outside world. The fundamental demand is this: the same factual record and the same evidentiary showing, processed through the same legal framework, should produce the same analytical structure — regardless of which panel draws the case.

3. Predictability Converts Risk into Something Manageable

Legal predictability does not mean the ability to guarantee litigation outcomes. Language has inherent limits. New technologies constantly test the edges of existing legal categories. Expert witnesses differ; the weight of evidence varies from case to case. Perfect certainty is unattainable, and no serious practitioner expects it.

But case law must at least map the reasonable range and direction of likely outcomes. Practitioners advising clients need to be able to review patent documents and prior decisions and reach conclusions like: "This design is likely to read on the claims under the plain meaning standard," "This modification creates doctrine-of-equivalents exposure, but the prosecution history provides a strong basis to invoke estoppel," or "There is insufficient motivation in the record to combine these prior art references in the manner the challenger proposes." Predictability does not mean certainty — it means the ability to identify legal risk, assign that risk a probability, and price it accordingly.

When the legal standards that case law supplies are too vague to support such assessments — when the honest conclusion is that no one can know the outcome without litigating to judgment — companies cannot quantify the probability or magnitude of their exposure. Only when infringement likelihood and expected damages can be estimated with reasonable confidence can a rational business choose among licensing, designing around, mounting an invalidity challenge, litigating on the merits, or withdrawing from the market.

Conversely, when practitioners perceive that materially identical facts may produce very different infringement findings depending on which panel draws the case, the exercise of calculating expected loss becomes meaningless. Companies may find themselves unable to select the most technically efficient solution — forced instead to abandon market entry to avoid uncertain patent risk, to pay royalties that exceed their actual exposure, or to adopt an inferior design-around when circumstances leave no better option. These additional costs represent what might be aptly described as an "innovation tax" — a levy imposed on the market by legal uncertainty.

4. Predictability Is a Core Asset in Patent Transactions and Monetization

Predictability matters not only to implementers and accused infringers. Patent owners have an equal — and sometimes greater — stake in it. When the scope, validity, and enforceability of a patent are opaque, the cash flows the patent might generate cannot be reliably projected. The result is instability across every dimension of patent economics: collateral value, royalty rates, assignment prices, technology transfer pricing, and enterprise valuation.

A patent's economic value is not born with the grant certificate. It is the product of several interacting variables: what products and technologies the claims actually cover, how readily a competitor can design around the patent, how likely the claims are to survive an invalidity challenge, and whether injunctive relief and meaningful damages are available when infringement is established. A patent's value, at bottom, is the present-discounted value of its future enforceability. When precedent lacks predictability, investors and counterparties apply an additional uncertainty discount — sometimes a very steep one.

Consider two patents of equal technical contribution. If claim construction standards and the doctrine-of-equivalents framework are settled and stable in the relevant technology space, expected royalty income and enforceability can be assessed with reasonable reliability. But if claim construction or obviousness determinations in the relevant field are known to vary materially depending on which panel draws the case, even a patent with enormous upside potential becomes hard to price. Counterparties respond to this uncertainty by discounting royalty rates or assignment prices, deferring consideration to contingent milestones, or walking away from deals altogether.

This discount is particularly devastating for cash-constrained startups, universities, research institutions, and independent inventors. Large corporations can pool their uncertainty across a broad portfolio — the expected value of any single patent's uncertain scope is diversified away. A startup whose entire enterprise value rests on one or two key patents has no such cushion. When the scope of a core patent is unpredictable, sophisticated investors price in litigation risk before they price in the technology's technical merits. Strong patents end up trading at weak prices; patent owners are pressed to license or transfer their technology on terms that fall short of what the invention is genuinely worth.

Patent-backed financing illustrates the same dynamic. Before extending credit secured by a patent, a lender must estimate recovery value — what the lender could recoup through a sale or license of the patent in a default scenario. If the patent's validity and scope are as a practical matter unknowable before a court speaks, collateral valuation becomes unreliable. Lenders respond by slashing loan-to-value ratios or demanding additional security. Legal uncertainty in patent doctrine translates directly into higher borrowing costs for technology companies.

In licensing negotiations, predictability is the foundation of rational bargaining. When legal standards are clear, both sides can derive a sensible range for deal terms from estimates of win probability, expected damages, litigation cost, and the risk of business interruption. When no one can know the outcome without litigating to judgment, license prices become a function of the parties' relative staying power, risk tolerance, and appetite for uncertainty — not the objective technical contribution of the invention. At that point, the patent transaction market ceases to be a market for technology value and becomes a market for litigation capacity.

Predictability, therefore, does not undermine patent rights. It is precisely what allows valid, infringed patents to command fair economic value. Clear doctrine prevents weak patents from being overvalued, while also preventing strong patents from being discounted simply because their legal status is uncertain. Predictability is not a principle that favors patent owners over implementers, or implementers over owners. It is the market infrastructure that makes patent assets tradable and financeable in the first place.

Conclusion

The quality of patent jurisprudence cannot be assessed solely by the internal logical coherence of a decision on the day it is handed down. The analysis must reach back to the years before the decision was rendered — to ask what choices engineers and business planners could actually make when they first reviewed the patent documents. When decisions fairly resolve past disputes while providing specific guidance for future conduct, patent law functions as infrastructure for innovation. When the same decisions leave practitioners unable to assess the risk profile of the next case, patent law becomes a source of uncertainty imposed on commercial activity.

That is precisely the significance of Decision 2023Hu11357. The Supreme Court did not rest on the abstract proposition that narrowing construction is sometimes permissible. Instead, it worked through a sequential analysis: the absence of specification support for the broad construction; the prosecution statements made in the parent application; the technical and procedural continuity of the divisional lineage; the deliberate exclusion of specific subject matter; and the interests of third-party reliance. As a result, the decision stands as a worked example — a template illustrating how the narrowing-construction exception can be applied without allowing it to expand in an unprincipled or self-defeating way.

What the patent system owes its constituents is neither unconditional victories for patent owners nor absolute safe harbors for implementers. It owes them the assurance that a party who has diligently reviewed the published claims and the accumulated case law can predict, with reasonable confidence, the legal consequences of its choices. With that assurance in place, companies will disclose their inventions, competitors will design around legally, investors will assign fair value to technology, lenders will accept patents as collateral, and markets will allocate capital in the direction of genuine innovation.

The proposition that "patent litigation must not become a panel roulette game" is more than a call for consistency in outcomes. It is the foundational principle that holds together patent law's notice function, legal certainty, freedom to design around, patent transactions and finance, and the long-run sustainability of investment in innovation. The point is not that courts must predict the future more accurately than the businesses they serve. The point is that it is the law's obligation to provide a reproducible set of analytical coordinates — so that businesses can plan their own futures.