When the Frame Is the Problem: Reading Coffee & Standards
Full disclosure: Used AI to summarize chapters of the 300+ page document and then personally reviewed sections to confirm accuracy / content.
TL;DR
David Rudin’s Coffee & Standards is a well-constructed field guide to standards governance and IP policy from a practitioner with more than two decades at Microsoft. It belongs on the shelf of anyone entering this practice.
The mechanical accuracy is not the issue. The issue I have with the publication is that its framing assumptions embed an implementer-side posture that an uninitiated reader will absorb as neutral description.
Rudin, writing from a standards governance and transactional practice rather than a patent prosecution background, opens his patent chapter with “The Monopoly Problem” and treats the patent itself as the source of the problem FRAND is designed to solve.
Rudin describes telecom standards development as companies independently developing technology, filing patents, and then contributing those patents to standards bodies. That is not how Standards Development Organizations (”SDOs”) like ETSI or 3GPP work. The technology and the standard emerge together, developed in concert by parties with opposing economic interests.
Treating royalty-free (”RF”) open source and Fair, Reasonable and Non-Discriminatory (”FRAND”)-bearing standards as interchangeable delivery mechanisms for interoperability is not a neutral governance observation. The Alliance for Open Media (”AOM”) and AV1 show what happens when that framing travels unchecked into real markets.
Open source and standards differ not only in IP framework but in governance structure, development process, and patent coverage. The Open Source Initiative (”OSI”) Open Source Definition (”OSD”) criterion 7 leaves the patent coverage question genuinely unresolved. A Benevolent Dictator for Life (”BDFL”)-governed project meets none of the due process criteria that formal standards bodies are expected to satisfy.
The governance frameworks Rudin analyzes (ANSI, WTO Technical Barriers to Trade (”TBT”), EU Regulation 1025/2012) are silent on what happens after a standard achieves widespread adoption. That is where the litigation lives.
Coffee & Standards grew out of internal training sessions David Rudin ran for attorneys at Microsoft. It shows. The book is organized, plainly written, and grounded in genuine institutional experience. Rudin’s treatments of necessary claims, patent pool antitrust questions, and multi-party negotiation dynamics are among the better short-form practitioner accounts available.
The concern here is not accuracy on mechanics. It is what the book teaches implicitly, through the assumptions embedded in its framing, that a reader without prior exposure to the licensor side of these disputes will carry forward without examining.
What the Book Gets Right
Start with what Rudin gets right in Chapter 1, because it is genuinely useful and often missed. A standard sets the floor for interoperability, not the ceiling for innovation. The standard tells you the minimum required to interoperate. It says nothing about what you are permitted to build on top. Rudin flags this early and correctly identifies it as the source of a persistent regulatory misunderstanding: procurement policies and government mandates that treat the standard as the boundary of permitted behavior rather than the baseline for competition. That error stifles the very innovation that standards are supposed to enable. It is a point worth making clearly to any practitioner advising on public procurement or regulatory compliance.
His Chapter 4 treatment of harmonized European standards deserves particular attention for practitioners advising on EU matters. When a directive sets essential requirements and the Commission issues a standardization request, a resulting harmonized standard cited in the Official Journal gives implementers a legal presumption of compliance with the underlying requirement. Compliance with the standard equals compliance with the law. That is a substantially higher legal consequence than anything attached to a U.S. voluntary consensus standard, and it bears directly on how ETSI’s standard-setting function should be understood in the context of EU regulatory enforcement.
Chapter 7 is more balanced than most practitioner guides in this space. Rudin names the implementer holdout problem directly: an implementer who delays, ignores licensing overtures, and forces the patent holder to litigate just to recover what was already owed is not behaving as a willing licensee. Courts have addressed this unevenly across jurisdictions, and the law is not settled, but Rudin presents both hold-up and holdout as real phenomena rather than treating one as established fact and the other as industry talking point. That even-handedness is the right starting position for a practitioner who needs to advise clients on either side of a licensing dispute.
On the FRAND commitment itself, his Chapter 7 formulation is correct and worth preserving: the RAND commitment is a commitment to license, not a license itself. The standards organization facilitates the commitment but does not set rates, administer licenses, or broker negotiations. Practitioners who conflate the two, treating the FRAND commitment as a rate determination or a cap on what a licensor can recover, are making a foundational error that Rudin’s framing correctly resists.
The Development Process Problem
Chapter 7 explains why RAND works for the telecom industry this way: “Companies spend years developing wireless technology, file patents on it, and then contribute it to standards bodies with the expectation that they’ll license those patents to implementers.”
That sentence describes something. It does not describe how technology development actually works inside 3GPP or ETSI.
Before turning to how telecom technology is actually developed, it is worth pausing on Rudin’s foundational premise. Writing from a standards governance and transactional practice rather than a patent prosecution background, he opens the patent chapter with “The Monopoly Problem” and treats the patent itself as the source of the problem FRAND is designed to solve. That framing has it backwards. A patent is a negative right. It is the right to exclude others from practicing the claimed invention for a limited term. It does not grant the holder the right to make, use, or sell anything. A patent on a technical approach that nobody is required to use creates no leverage over anyone. The question of what leverage, if any, a patent holder possesses is determined by whether implementers have commercially viable alternatives, not by the existence of the patent itself.
This is not an abstract reframing. The DOJ has made it an enforcement principle. The Statement of Interest filed in Disney Enterprises, Inc. v. Interdigital, Inc. (October 6, 2025) (link) established that market power cannot be presumed simply because a patent has been incorporated into a technical standard. The Statement of Interest in Samsung Electronics Co., Ltd. v. Netlist, Inc. (April 7, 2026) (link) reinforced this, holding that a complaint cannot sufficiently allege market power based on Standard Essential Patents (”SEPs”) without affirmative assessment of alternatives to the standard and the contractual obligations the patent holder has assumed under the SDO’s patent policy. Under the rule of reason framework confirmed in Texaco Inc. v. Dagher, 547 U.S. 1 (2006) (link), market power is a threshold element requiring affirmative proof, not a presumption that follows from standardization. A rate-setting methodology that begins from the hold-up assumption embeds an unproven premise at the foundational step of the analysis.
FRAND exists not to discipline a patent holder but to preserve the innovation incentive of contributors who took real technical risk under conditions of genuine uncertainty, with no guarantee their approach would be selected. The bilateral framework of mutual obligations it creates is the mechanism through which that incentive is honored, not a ceiling on what a licensor can recover. Bowman Heiden and Justus Baron, writing in the Harvard Journal of Law and Technology, Vol. 38, No. 3 (2024), document a royalty gap of $7 to $28 billion annually as of 2021 in which SEP holders are systematically undercompensated relative to FRAND-compliant rates. (link) The baseline those figures use, ex ante contribution value rather than hold-up value, is the same baseline the Disney and Netlist Statements of Interest identify as the legally correct reference point under the rule of reason. The gap figure is contested at the margins, as all empirical measurements in active litigation fields are, but it is probative of a directional question: whether Rudin’s foundational framing creates a systematic methodological bias that runs against licensors, not whether any particular asserted rate is FRAND-compliant. On that directional question, a documented multi-billion-dollar annual gap is meaningful evidence that the framing is not analytically neutral.
In a genuine SDO, the sequence Rudin describes does not exist. There is no pre-existing patented technology waiting to be contributed. Engineers from Qualcomm, Nokia, Ericsson, Huawei, InterDigital, and dozens of other companies enter working groups together and develop technical solutions to defined problems collaboratively. They propose competing approaches. They debate tradeoffs. They vote, document objections, and arrive at consensus under conditions of genuine technical uncertainty. Nobody in the room knows at the outset which approach will be selected. The patents that ultimately read on the resulting standard are frequently filed during or after that process, on inventions conceived and refined inside the standards work itself, though patent applications on related technology filed prior to adoption are not uncommon. The technology and the standard come into existence through the same process. They are not separable artifacts.
The scale of that collaborative process in 3GPP is documented. For the period 2005 to 2014, approximately 350 firms made close to 400,000 unique contributions to 3GPP meetings. Roughly 30 percent of those contributions were accepted.² Implementer-aligned commentators have argued that large incumbent contributors arrive at working groups with pre-developed technology platforms and patent prosecution strategies coordinated to the standards calendar, making in-session invention a framing choice rather than a genuine description. That argument proves too much. A 30 percent acceptance rate across nearly 400,000 contributions from 350 firms, including proposals from the largest contributors in direct competition with each other, is not consistent with a process designed to ratify predetermined outcomes. Qualcomm, Nokia, and Ericsson are not aligned. They compete in the same working groups for the same slots in the same specifications. The record of contested proposals, documented objections, and non-selected technical approaches is precisely what a genuine competitive technical process produces. The continuation-portfolio timing practices that implementer advocates cite are a real and separate concern about how patents are prosecuted after a standard is adopted, and that question warrants its own analysis. It does not address whether the technical outcome of the working group process was determined before the meeting, which is the specific proposition the contribution data is cited to rebut.
That reality creates a serious problem for the rate-setting methodology Rudin describes. He defines “reasonable” as the rate a patent would command in a hypothetical negotiation where competitive alternatives exist, stripping out what he calls the “monopoly premium created by standardization.” That methodology requires that the patent’s value be analytically separable from the standard. Where technology and standard emerge from the same collaborative process, that separation is not available. For technology developed inside the working group, what exists at the moment of adoption is not a standalone inventin with a separable market value. What exists is a technical approach that competed against other approaches in the same process, any of which could have been selected instead. The appropriate ex ante question is what rate would have been agreed upon before adoption, when the alternative approaches were still viable options in the process. Rudin’s “monopoly premium” formulation assumes a premium exists to be stripped out. Where the contribution and the standard are the product of the same collaborative technical work, that assumption has no foundation.
The practical consequence of Rudin’s framing is that it imports the hold-up assumption as first principle without naming it as such. The two framings lead to different rate-setting methodologies, different default assumptions about who bears the burden of proving reasonableness, and different answers to what “fair” means in Fair, Reasonable and Non-Discriminatory.
Not the Same Destination
Chapter 5 is titled “Open Source and Standards: The Convergence.” Rudin’s argument is that over the past two decades the two worlds have converged “to the point where the line between them is often invisible,” with the same organizations hosting both, the same participants, the same governance challenges, and the same IP questions. Both are, in his framing, “fundamentally, structured collaboration among parties who don’t fully trust each other and don’t share interests in any deep way,” held together by the same creative tension between self-interest and collective benefit.
That framing collapses distinctions that matter enormously when a real dispute arises.
Rudin correctly identifies the first layer: an open source license covers contributed code, while a standards patent commitment covers the entire specification regardless of who contributed what. An independent implementer who reads open source code and builds their own version gets no patent coverage from the license. That is accurate as far as it goes.
It does not go far enough.
Governance structure. Formal standards development requires openness, balance across participant categories, lack of dominance by any single interest, consensus-based decision-making, documented resolution of objections, and a formal right of appeal. Those criteria define what separates genuine collaborative standard-setting from governance captured by a subset of participants. It is worth noting that even accrediting bodies have applied these criteria unevenly. ANSI’s position has historically been that governance — hence IPR policies that “govern” participation — do not warrant the same level of scrutiny as technical development processes, an approach that allowed the IEEE’s 2015 PatCom patent policy change to proceed under accreditation conditions that would not have survived rigorous application of the balance and consensus requirements to the policy process itself. That is a separate argument for a separate piece. The point here is narrower: whatever the floor of governance legitimacy requires in a formal standards body, a BDFL-governed open source project does not approach it. A significant proportion of major open source projects are governed by a single maintainer or a small committee operating on contribution-based authority with no due process obligations, no balance requirement, and no formal appeals mechanism. Linux, for decades governed by Linus Torvalds, is the canonical example. A BDFL-governed project meets none of the criteria that even minimally legitimate standards governance requires. The two governance structures are not alternative instruments for the same function.
Development process. In an SDO, a technical proposal enters a working group, competes against other proposals from parties with opposing economic interests, and survives or fails based on documented technical merit assessed through a consensus process with appeal rights. In an open source project, a pull request is accepted or rejected at maintainer discretion. The criteria, the accountability, and the institutional record are structurally different.
Patent coverage. The OSI’s OSD has ten criteria. They are licensing requirements, not governance or development standards. OSD criterion 7 states that the rights attached to the program must apply to all to whom the program is redistributed without the need for execution of an additional license by those parties. That criterion has been interpreted in OSI license-review discussions as potentially bearing on whether open source licenses must include a patent grant. The OSD is silent on patents, a product of its 1990s copyright-law origins. The patent coverage question under OSD criterion 7 remains genuinely unresolved, a point OSI itself has acknowledged and actively contested when FRAND-dependent companies have argued that open source licenses can be “copyright only.”¹ An SDO patent policy and an open source license are not covering the same ground through different mechanisms. In significant respects, the open source license covers no ground at all on the patent question.
The AOM/AV1 case. A practitioner trained on the convergence thesis will be least equipped to recognize a real-world problem when they encounter one.
AOM was not formed as a neutral technical governance vehicle for collaborative development. It was formed in 2015 by a founding membership that included Google, Mozilla, Microsoft (employer of David Rudin), Amazon, Netflix, Intel, and Cisco, large-scale downstream implementers who shared a direct financial interest in eliminating per-unit royalty costs for video codec technology. The AOM Patent License 1.0 grants each implementer a royalty-free license to each licensor’s necessary claims, on condition that the implementer grants its own necessary claims on the same royalty-free terms. The grant-back obligation is the mechanism. Any contributor who holds FRAND-encumbered patents and depends on licensing revenue faces a binary choice: accept royalty-free terms for all necessary claims or do not participate.
The predictable result of that structure is that FRAND-dependent innovators do not participate under the AOM framework. They retain their licensing positions and operate through independent patent pools and bilateral agreements outside the consortium entirely. The existence of AV1 patent pools operating outside the AOM RF license confirms the structural point: there are essential patents in AV1 that the RF license does not cover. The “open” label attached to AV1 is real at the code layer. At the patent layer, the picture is materially more complicated, and practitioners who assume otherwise on the strength of AOM’s governance narrative will advise their clients incorrectly.
Dina Kallay, serving as Deputy Assistant Attorney General for International, Policy and Appellate at the DOJ Antitrust Division, addressed exactly this dynamic in her September 2025 Concurrences keynote, titled “’That’s What F/RANDs Are For’ and Antitrust Implications When They’re Gone.”³ Her concern was specific, not general. She identified the breakdown of the FRAND-assured standards development ecosystem as an enforcement priority, with particular attention to two scenarios: collaborative standards that incorporate technologies carrying negative or missing FRAND declarations, and proprietary consortia whose policies are not FRAND-based. On the first scenario, her analysis tracks the AOM structure precisely. When a standard that is supposed to be open incorporates technologies whose patent holders have declined FRAND terms, or where the governance structure of the consortium makes FRAND participation economically irrational, implementers deploy the standard without any assurance that essential patent access will be available on competitive terms. That is not a generic competition concern about open labels. It is a targeted description of what AOM’s grant-back obligation produces for any contributor who depends on FRAND licensing revenue. The convergence thesis Rudin advances in Chapter 5 has no framework for that analysis. A practitioner working from his book will see AOM as an innovative governance model. A practitioner who has read Kallay’s keynote will ask a different set of questions about who designed the structure, who benefits from it, and whose ability to participate on commercial terms it eliminates.
In Chapter 17, Rudin observes that patent litigation is concentrated in the RAND standards world and nearly absent from open source, and acknowledges that “the reality is almost exactly backwards” from what his framework would predict. He does not resolve why. One reading of that absence is that royalty-free cross-licensing among AOM members resolved patent exposure efficiently, which is what the AOM structure was designed to accomplish. The more consistent reading, given the structural evidence, is different. The AOM Patent License 1.0’s grant-back obligation makes participation economically irrational for any contributor whose business model depends on FRAND licensing revenue. The predicted behavior under that structure is non-participation, not voluntary acceptance of RF terms. And the existence of AV1 patent pools operating entirely outside the AOM framework confirms that essential patents remain unlicensed under the RF license, which is precisely the outcome Deputy Assistant Attorney General Kallay identified as an antitrust enforcement concern: implementers deploying a standard without assurance that essential patent access is available on competitive terms. The absence of litigation is not evidence of a clean IP landscape. It is evidence of a participation structure that made FRAND-dependent contribution uneconomic from the outset. When there are no licensors in the room, there is no one to litigate against.
The Gap After Adoption
The three governance frameworks Rudin analyzes in Chapter 4 converge on process legitimacy. They govern the development of standards. They are silent on what happens after a standard achieves widespread commercial adoption.
Once a standard is broadly deployed and alternatives have receded from the market, the governance question shifts. The ANSI Essential Requirements no longer apply. The WTO TBT Decision has nothing further to say. EU Regulation 1025/2012 is silent. What governs from that point is the FRAND commitment made inside the standards body, and the bilateral framework of mutual obligations that commitment creates between the patent holder and every implementer who deploys the standard, including implementers who were never parties to the original standards process and who may have entered the market years or decades after the relevant technical work was complete.
That bilateral framework is not administered by the SDO. ETSI does not set rates. ETSI does not broker negotiations. ETSI does not resolve disputes. What ETSI’s IPR Policy does, and what Rudin’s warehouse metaphor cannot account for, is create a present-tense legal obligation at the moment of declaration. ETSI Clause 6.1’s “prepared to grant” language is not a procedural checkbox. It is a legal act with bilateral consequences: the patent holder commits to offer licenses on FRAND terms, and by accepting that commitment as the condition of standardization, the ecosystem creates a corresponding obligation on implementers to negotiate in good faith rather than to use the standards process itself as a mechanism for compressing or eliminating royalty obligations.
The Optis Wireless Technology LLC v. Apple Inc. proceedings now before the UK Supreme Court (UKSC/2025/0144 and UKSC/2025/0145), with a hearing scheduled for late June 2026, sit precisely in this space. The questions before the court go to the nature and enforceability of the obligations that arise from the FRAND commitment, who bears the burden of compliance, and what remedies attach when those obligations are not honored. Those are post-adoption questions. They are not answered by ANSI, TBT, or EU Reg 1025/2012. They are not addressed in Rudin’s Chapter 4. They are where the litigation lives, and they are what a practitioner trained solely on governance frameworks will be unprepared to analyze.
The Ecosystem Problem
The concern with a book like Coffee & Standards is not the book itself.
It is that well-written practitioner guides with embedded framing assumptions become baseline training. They travel into regulatory submissions, judicial briefing, and licensing negotiations. They shape which questions get asked and which frameworks get applied before a single argument is made on the merits.
The collaborative development error in Chapter 7 is not a footnote. It is the foundation of the FRAND rate-setting framework, and it imports the hold-up assumption as a first principle. The convergence thesis in Chapter 5 treats structurally distinct institutions as interchangeable, erasing differences in governance, development process, and patent coverage that define how disputes actually resolve. The post-adoption gap in Chapter 4 is the space where FRAND actually operates, and it is the space this kind of guide structurally cannot reach.
Sophisticated practitioners will read past these defaults. The audience most at risk is the one that will not, because no one told them to look.
Jim Harlan is an independent intellectual property practitioner and the author of Standards at Risk (standardsatrisk.com).
Footnote
Richard Fontana, “Is it time to revise the Open Source Definition?”, Opensource.com, September 30, 2020. Fontana notes that companies with FRAND-based SEP licensing models have advanced the view that open source licenses can be “copyright only,” and that the OSI has strongly opposed that position. The OSD’s silence on patents leaves OSD criterion 7 as the primary but contested textual hook for any patent coverage requirement. (link)
Kirti Gupta, “The Role of SMEs and Startups in Standard Development,” SSRN, July 2017. The contribution and acceptance data cited here are drawn from Gupta’s analysis of 3GPP participation for the period 2005 to 2014. (link)
Dina Kallay, “’That’s What F/RANDs Are For’ and Antitrust Implications When They’re Gone,” Remarks as Prepared for Delivery, Concurrences Dinner, New York, September 19, 2025. (link)