Policy 10: Eliminate software patents. Reform the patents system.

Researched: 2026-07-22. Status: draft.

The plank

Policy 10. Eliminate software patents. Reform the patents system.

That is the entire plank: the README gives no body text, and app/templates/index.html renders the identical title, also with no body text. There is no README/index.html drift for this plank (note: Policy 4's title does drift between the two files — "Disintermediate" vs. "Separate" — but that belongs to the Policy 4 doc).

Because there is no body text, this doc constructs the plank's meaning from the title: (1) end the patenting of software as such — inventions whose only claimed contribution is a program, algorithm, or business method running on ordinary computers; (2) reform the surrounding system — litigation venue, post-grant review, fee-shifting, and transparency — so that the patents that remain are cheap to test and expensive to abuse. A first drafting decision for the platform: whether "eliminate" means statutory exclusion of new software patents (prospective) or also extinguishing in-force ones (retroactive). This doc recommends the prospective reading; see Open questions.

The problem

Software patents are the raw material of a litigation industry that taxes companies that build things and pays companies that sue.

• Direct cost of troll litigation. Bessen & Meurer (Boston University) estimated that defendants accrued $29 billion in direct costs from non-practicing entity (NPE) patent assertions in 2011 — legal fees plus settlements/licenses, excluding indirect costs like diverted engineering time and delayed products (Bessen & Meurer, "The Direct Costs from NPE Disputes," Cornell Law Review, 2012/2014). This is the most-quoted number in the debate and it is contested: Schwartz & Kesan criticized its broad NPE definition (which sweeps in universities and individual inventors), its reliance on a non-public survey run by RPX (a defensive patent aggregator with a commercial stake in the troll narrative), and its mixing of transfers (settlements) with social costs (Schwartz & Kesan, "Analyzing the Role of Non-Practicing Entities in the Patent System," 2014). RPX's own later accounting differed from Bessen–Meurer by $16.8 billion (IAM, "The cost of NPE disputes in the US," reporting the gap). Honest bottom line: the true annual direct cost is somewhere in the billions to tens of billions; even the critics do not get it below "billions." Legal fees alone are pure deadweight loss regardless of definition.
• Software is where the litigation is. GAO found that over 2007–2011 the number of patent defendants grew ~129%, that companies sued over software-related patents accounted for about 89% of the 2011 increase in defendants, and that roughly two-thirds of defendants over the period were sued over software-related patents (GAO-13-465, "Assessing Factors That Affect Patent Infringement Litigation Could Help Improve Patent Quality," August 2013).
• The troll share keeps growing. In 2025, NPEs filed 55.4% of all US district-court patent cases (up from 51.8% in 2024), and 90.3% of high-tech patent litigation was NPE-driven (Unified Patents, "Patent Dispute Report: 2025 in Review," January 2026). RPX counted 1,889 defendants added to NPE campaigns in 2024 (+21.6% over 2023) and 1,189 in H1 2025 (+30% year-over-year) (RPX quarterly reviews, 2025).
• Defense costs force settlement regardless of merit. AIPLA's 2023 Economic Survey put the median cost of defending a patent case through trial at $600K when under $1M is at stake, rising to $3.6M when over $25M is at stake (AIPLA, Report of the Economic Survey, 2023; a 2025 edition was published February 2026 — update figures on second pass). The FTC's compulsory-process study of 22 PAEs (2009–2014 data) found "Litigation PAEs" filed 96% of PAE suits but earned only ~20% of PAE revenue, settled 66% of cases within a year, and typically settled for under $300,000 — below the cost of a defense through discovery ($300K–$2.5M). The FTC concluded discovery burden, not patent merit, drives settlements, and called this "nuisance litigation" (FTC, "Patent Assertion Entity Activity," October 2016).
• How it reaches ordinary people. These costs are a tax on making software, passed into the price of everything that contains software — which is now most things. They fall hardest on small firms: Bessen & Meurer found most NPE defendants are small or medium-sized companies (2012, above). App developers, podcasters (the Personal Audio "podcasting patent" campaign, 2013–2017), and retailers who merely use off-the-shelf technology (scanner/Wi-Fi demand-letter campaigns, ~2011–2013) have all been assertion targets. Wider estimates — Bessen, Ford & Meurer's finding that NPE suits were associated with ~$500B in lost defendant share value over 1990–2010, and Kiebzak/Rafert/Tucker's estimate that VC investment would have been ~$21.8B higher over five years absent troll litigation — exist but are event-study/model-based (estimate, weak evidence; both unverified this session).

Who profits from the status quo

• Patent-assertion investors. Fortress Investment Group (owned by Mubadala/Abu Dhabi since 2023) has ~$2.9 billion committed to intellectual property and calls itself the world's largest institutional investor in patents, within ~$6.6B committed to legal assets overall (Bloomberg Law, "Fortress' Billions Quietly Power America's Biggest Legal Fights"). Its vehicle VLSI Technology LLC — a company with no products, asserting ex-NXP chip patents — won a $2.18 billion jury verdict against Intel in the Western District of Texas in March 2021 (reversed by the Federal Circuit in 2023) and a separate $948.8 million verdict in 2022; in 2025 a Texas jury found VLSI and Finjan were both under Fortress's control, which triggered Intel's license defense. Intellectual Ventures, the archetype portfolio PAE, raised on the order of $5–6B from investors (including operating tech companies and university endowments) and amassed tens of thousands of patents (order-of-magnitude figures widely reported; unverified this session).
• The litigation venues' local ecosystems. Marshall, in the Eastern District of Texas, built a boutique economy around patent plaintiffs — the emblem being Samsung's sponsored ice rink in front of the courthouse while it was a serial defendant there (widely reported, e.g., The Guardian 2015; unverified this session). After TC Heartland (2017) curtailed EDTX venue, the business moved: Waco's Judge Alan Albright openly recruited patent cases and within three years was drawing ~25% of all new US patent suits — W.D. Tex. filings grew from under 100/year to ~1,000 in 2021 — until the district's July 25, 2022 order randomly spread Waco patent filings across 12 judges (Bloomberg Law; Goodwin/Akin client analyses, 2022–2023).
• Large defensive-portfolio licensors. IBM cleared $1 billion+ per year in IP income most years since 1996 — roughly $27 billion cumulative — peaking at $1.1–1.2B/year in 2008–2012, much of it from "godfather" portfolio licenses backed by the implicit threat of thousands of software patents; income fell to $626M by 2020 as Alice weakened the threat (Forbes 2016; Bloomberg 2021). Microsoft's Android patent-licensing program was reported to earn billions per year at peak (~2013–2014; unverified this session). These firms profit from the system's opacity even when they never sue.
• The patent bar. Patent litigation and prosecution is a multi-billion-dollar legal services market — the AIPLA cost survey exists because its members bill those medians. Bar associations (AIPLA, IPO) and the licensing industry consistently oppose strong eligibility limits; AIPLA and IPO have proposed §101 rewrites that would expand eligibility, and the pharmaceutical industry — which spends roughly $300–400M/year lobbying across all issues (OpenSecrets; figure approximate, unverified this session) — backs the same direction via PERA. The Innovation Act (H.R. 9), the last serious anti-troll bill, passed the House 325–91 in December 2013 and died in the Senate under opposition from pharma, universities, licensing firms, and the trial bar (unverified this session).
• Who profits is not "inventors." The FTC found Litigation PAEs — 96% of suits — generated only ~20% of PAE revenue, and its report gives no evidence that meaningful shares of assertion revenue reach original inventors (FTC 2016).

The proposal, concretely

1. Amend 35 U.S.C. §101 to exclude software as such. Codify the Alice/Mayo two-step test and add an explicit exclusion modeled on European Patent Convention Art. 52(2)–(3): "programs for computers... as such" are not inventions. Define the boundary the way the EPO does not: a claimed invention whose contribution over the prior art lies in information processing itself — rather than in a physical process or device it controls or measures — is ineligible, regardless of drafting. (The EPO's caveat is the cautionary tale: under its COMVIK approach, T 641/00, a single technical feature clears eligibility and the fight moves to inventive step, so Europe in practice grants large numbers of "computer-implemented invention" patents. See Evidence.) The statute must aim at the claim's substance, not its wrapper, or drafting will route around it as it did after State Street (1998).
2. Prospective application. New exclusion applies to applications filed after enactment; in-force software patents are not summarily voided but remain subject to Alice and to a strengthened IPR (below). This avoids a Fifth Amendment takings fight and a decade of transition litigation.
3. Defend and strengthen inter partes review (PTAB). IPR — created by the America Invents Act of 2011 — lets anyone challenge a granted patent's validity before administrative judges for a few hundred thousand dollars instead of millions in district court. Keep it, end discretionary "Fintiv" denials that let fast trial venues shield patents from review (USPTO loosened Fintiv limits again in 2025 per RPX reporting), and oppose the PREVAIL Act, which would restrict who can petition (unverified this session on PREVAIL details — confirm on second pass).
4. Codify and tighten venue. Lock in TC Heartland v. Kraft Foods (2017) by statute and add random judge assignment within districts for patent cases — nationalizing the fix W.D. Texas applied to Judge Albright's Waco docket in July 2022 — so no single judge can market a forum to plaintiffs.
5. Fee-shifting with teeth. Strengthen 35 U.S.C. §285 beyond Octane Fitness (2014): presumptive fee award to prevailing accused infringers when the asserted claims are invalidated under §101/§103 or when the plaintiff is a shell with no operations, plus joinder of the real parties in interest (the funder pays, not just the empty LLC).
6. Transparency. Mandatory disclosure of ultimate ownership and third-party litigation funding in every patent suit — generalizing Chief Judge Connolly's D. Del. standing orders that exposed funder-controlled shells (2022; unverified this session).

Evidence

• Alice v. CLS Bank, 573 U.S. 208 (2014) — the natural experiment. A unanimous Supreme Court held that abstract ideas implemented on a generic computer are ineligible under §101. Measured effects: new patent case filings fell ~40% year-over-year within months (329 in Sept 2014 vs. 549 in Sept 2013, Lex Machina); within six months, 32 district-court decisions and 4 Federal Circuit opinions had invalidated patents under Alice (Goodwin year-end review, 2014). At the USPTO, first-office-action §101 rejections in Alice-affected technologies rose 31% in the following 18 months and examiner-level uncertainty rose 26%, later reduced by the 2019 eligibility guidance (USPTO Office of the Chief Economist, "Adjusting to Alice," April 2020). Crucially for the innovation question, a 2024 empirical study found Alice did not reduce R&D activity or the value of software patents — it stimulated firms' innovation and raised patent values by clearing out broad, ambiguous claims ("A rabbit hole to innovation land: An empirical examination of the Alice decision," Computer Law & Security Review, 2024). Alice is a partial version of this plank already run in production: software patenting got harder, litigation fell, and software innovation did not slow. Maps directly to the US case because it is the US case.
• The software boom that patents didn't cause. Software was barely patentable before the Federal Circuit's mid-1990s decisions (In re Alappat 1994, State Street 1998) — yet Microsoft, Oracle, Adobe, the internet, and the open-source movement all predate broad software patentability. Bessen & Hunt found firms that patented software more intensively had lower R&D intensity, consistent with patents substituting for innovation rather than funding it ("An Empirical Look at Software Patents," Journal of Economics & Management Strategy, 2007; unverified this session). Bessen & Meurer's book-length accounting concluded that for US public firms outside chemicals and pharmaceuticals, litigation costs of the patent system exceeded the profits patents generated by the late 1990s — the system is a net tax on innovators everywhere except molecule-based industries (Bessen & Meurer, Patent Failure, Princeton University Press, 2008; unverified this session). Boldrin & Levine's survey went further: "there is no empirical evidence that [patents] serve to increase innovation and productivity" absent industry-specific caveats, and recommended phased abolition ("The Case Against Patents," Journal of Economic Perspectives 27(1), 2013; unverified this session). Heller's anticommons work supplies the mechanism: when thousands of overlapping rights cover one product — a smartphone implicates hundreds of thousands of patents — fragmentation itself blocks use (Heller & Eisenberg, Science, 1998; Heller, The Gridlock Economy, 2008; unverified this session).
• Europe's exclusion — supportive but cautionary. EPC Art. 52(2)(c) excludes "programs for computers," but Art. 52(3) limits this to programs "as such," and EPO case law (T 1173/97's "further technical effect"; T 641/00 COMVIK) lets any claim with one technical feature through to an inventive-step analysis where only technical features count. Result: Europe has less software-patent litigation and no US-scale troll industry, but it still grants many computer-implemented-invention patents — the statutory words alone did not eliminate them (EPO case law; practitioner analyses by Bardehle, Mewburn, NLO, 2020–2023). Lesson for drafting: the exclusion must bind the contribution, not the claim's costume. Europe's smaller troll problem also owes much to loser-pays fee rules and no juries — which is why planks 4–6 above travel with the exclusion.
• Post-2014 reform effects. TC Heartland (2017) collapsed the Eastern District of Texas's share of national patent filings (from roughly 40%+ in 2015–16 to the teens; precise Lex Machina figures unverified this session) — proof that procedural rules, not merits, were steering caseloads; the subsequent migration to Waco (under 100 to ~1,000 filings/year by 2021) proves the fix must be structural, not one-off. Overall NPE share nonetheless hit 55.4% by 2025 (Unified Patents), showing court-made doctrine without statutory backing erodes.

Overall strength of evidence: moderate-to-strong that software patents impose large net costs and that restricting them (Alice) did not harm software innovation; thin-to-moderate on the precise dollar magnitudes (the $29B is genuinely contested) and on predicting effects of full statutory elimination beyond what Alice already did.

Who wins, who loses

Transition: prospective-only exclusion means in-force software patents (a large share of the ~3M+ in-force US patents; exact software share undefined — GAO noted there is no agreed definition) die off over their remaining ≤20-year terms while remaining subject to Alice and IPR. Transition pain is therefore concentrated on future rent streams, not existing adjudicated rights. No compensation mechanism is proposed for NPEs or the patent bar, and none is warranted for a business model the FTC characterized as nuisance litigation; the honest hard case is the individual inventor with a pending software application, who loses expectancy without compensation.

Fiscal impact

Direct federal budget effect: approximately zero, and this must be said plainly — the costs of the status quo are private-sector deadweight loss, not federal outlays.

• The USPTO is fee-funded (~$4B/yr in fees; approximate, unverified this session). Fewer software applications would cut both fee revenue and examination workload roughly in parallel; net budget effect small either way. Historical caveat: Congress has at times diverted USPTO fees, so a fee decline is not a taxpayer loss.
• Federal courts: patent cases are a small share of the civil docket; reduced filings free judicial resources but produce no scoreable savings.
• Indirect revenue: if Bessen–Meurer-scale costs (order $10–29B/yr, 2011-era, contested) are partly converted back into taxable corporate profit and wages, federal revenue rises modestly — a back-of-envelope 21% corporate rate on even $10B of recovered profit is ~$2B/yr, but this is a hand-wave, labeled as such (estimate, weak evidence).
• No CBO or JCT score exists for any software-patent-exclusion bill; none has been introduced in this form. An economist reviewing this plank should be asked whether any scoring literature exists for the Innovation Act generation of bills.

Within the platform's fiscal frame, Policy 10's contribution is growth and cost-of-living, not deficit reduction. Do not claim otherwise in talking points.

The opposition's best case

1. "Startups need patents to raise money." The strongest version comes from the Berkeley Patent Survey (Graham, Merges, Samuelson & Sichelman, 2008 survey, published 2009): among 1,332 startups, patents mattered for securing investment — VC-backed startups patent far more than others — and executives reported signaling and competitive-prevention value, including in software more than previously believed. Related work (Farre-Mensa, Hegde & Ljungqvist's patent-"lottery" study finding first patent grants causally boost startup growth; unverified this session) strengthens this. Response: the same survey found many software entrepreneurs forgo patents entirely because of cost, and software startups rank patents low among appropriation mechanisms relative to speed and secrecy; meanwhile startups are disproportionately troll targets (FTC 2016; Bessen–Meurer 2012 on SME defendants). The signal is real but is a signal because the underlying right is costly — copyright, trade secrets, and traction can carry the load, as most software startups already demonstrate. Concession in writing: for a minority of deep-tech software startups, losing patents removes a genuine financing asset, and the causal-grant literature says this cost is not zero.
2. "Section 101 chaos is the real problem — codify eligibility broadly instead" (PERA). Senators Tillis and Coons reintroduced the Patent Eligibility Restoration Act (S.1546, 119th Congress) on May 1, 2025, with House companions from Reps. Kiley and Peters; a Senate Judiciary IP Subcommittee hearing was held October 8, 2025, cosponsors Blackburn and Hirono joined in June 2025, and Tillis has said he wants a markup this Congress — as of this research date it has not passed committee (track before publication). PERA would statutorily overturn Alice, Mayo, and Myriad and re-expand eligibility. Its proponents (also including former USPTO directors Kappos and Iancu) argue Alice-era uncertainty — documented by USPTO's own report (31% jump in §101 rejections, +26% uncertainty) — chills investment in AI, diagnostics, and quantum, and pushes capital to China and Europe. Response: the uncertainty is real and conceded; but the empirical record shows software R&D and patent value did not fall after Alice (Computer Law & Security Review, 2024), and the cure for an unclear line is a clear line, which a statutory exclusion provides more cheaply than universal eligibility. PERA resolves uncertainty in the direction that re-arms the troll business model documented by the FTC. Note honestly: PERA's existence means this plank swims against the current legislative tide.
3. "You can't define 'software patent' — the exclusion will be arbitraged." The critics' best evidence is Europe: EPC Art. 52 excludes programs "as such," yet the EPO grants computer-implemented-invention patents at scale through the COMVIK inventive-step channel; GAO itself noted the absence of an agreed definition. Response: largely conceded as a drafting risk — which is why the proposal targets the claimed contribution and pairs the exclusion with fee-shifting, venue, and IPR reforms that reduce the payoff to gaming the line. Europe's experience simultaneously shows the downside (porous line) and the upside (no US-scale troll industry).
4. "The troll-cost numbers are inflated." Schwartz & Kesan (2014) showed the $29B figure rests on RPX's non-public survey, a definition of NPE that includes universities, and conflation of transfers with costs; IAM documented a $16.8B gap versus RPX's own estimate; a 2013 GAO reading emphasized most litigation growth was software-patent-driven rather than entity-driven. Response: conceded that $29B is an upper-bound advocacy number and this doc treats it as contested everywhere it appears. But the critique cuts oddly for the status quo: if the problem is the patents (GAO's reading), the case for eliminating software patents is stronger, not weaker, than the troll framing suggests.
5. "Property rights, not litigation reform, drive investment — weakening patents weakens the innovation contract" (the Epstein/Kappos property-rights school; also the "death squad" critique of PTAB from former Chief Judge Rader). Response: the theory is coherent but the evidence for software specifically is against it — the industry's formative decades predate software patentability; firms that patent software more do not do more R&D (Bessen & Hunt 2007); and litigation costs exceed patent profits outside chemistry/pharma (Bessen & Meurer 2008). Where the property-rights argument is strong — molecules, with high invention cost and cheap copying — this plank deliberately changes nothing. Concession: IPR does create repeat-challenge harassment potential against legitimate patentees; reform should include estoppel and serial-petition limits.

Talking points

• One line: Software patents don't fund inventors — they fund a litigation industry that taxes everyone who builds software; end them and make the rest of the patent system honest.
• Thirty seconds: More than half of all US patent lawsuits — and over 90% of high-tech patent suits — are now filed by companies that make nothing (2025). Defending even a small case costs hundreds of thousands of dollars, so the FTC found trolls price settlements just under the cost of defense — a shakedown by design. When the Supreme Court restricted software patents in 2014, lawsuits dropped about 40% within months and software innovation didn't slow at all. Write that into law: no patents on software as such, and reform venue, review, and fee rules so bad patents die cheaply.
• Two minutes: adds the villains and the honest costs. Fortress, a private-equity firm, has ~$2.9B invested in patents and won a $2.18B verdict against Intel through a shell company that makes nothing (later reversed). One judge in Waco marketed his courtroom so well he drew a quarter of all US patent cases by 2021. IBM extracted roughly $27B over two decades in portfolio licensing backed by software-patent threats. The costs land on small companies — most troll defendants are SMEs — and pass through to everyone who buys anything with software in it. Estimates of the direct cost ran as high as $29B/yr (2011); even the academics who attacked that number concede billions. The honest costs of our fix: a minority of startups genuinely use patents to signal investors and would lose that asset; Europe's version of the exclusion shows clever lawyers will probe any line we draw, so the fee-shifting and venue reforms are load-bearing, not decoration; and Congress is currently moving the other way (PERA, 2025), so this plank is a fight, not a formality. Pharma and biotech are deliberately untouched.

Open questions

• Retroactivity and takings. Does the plank mean prospective exclusion (this doc's reading) or extinguishing in-force patents? Retroactive elimination invites Fifth Amendment takings litigation; get a constitutional-law read before hardening the language.
• The definition problem. Can a statutory "contribution-based" software exclusion actually out-draft the patent bar where EPC Art. 52 has not? Commission a comparative drafting memo (EPO COMVIK line vs. proposed US text) before the ebook chapter.
• Startup financing effect size. How large is the VC-signaling loss really? The patent-lottery causal literature (Farre-Mensa/Hegde/Ljungqvist) needs to be read and quantified against troll-defense savings for the same startup population. This is the number most likely to change our mind about scope (e.g., a small-entity carve-out).
• PERA trajectory. S.1546 status as of research date: hearing held October 2025, markup promised, not yet reported out. If PERA passes first, this plank becomes a repeal effort — different politics entirely. Track through 2026.
• What an economist should check first: (1) whether the 2024 Alice/R&D study's identification holds up and has been replicated; (2) current (2025 AIPLA, published Feb 2026) litigation cost medians; (3) whether any credible post-2014 estimate of aggregate NPE costs updates Bessen–Meurer; (4) IPR outcome statistics (institution and invalidation rates) from primary USPTO data, which this draft cites only from memory.
• Verification debt (session limitation). Outbound web access was cut mid-research; items marked "unverified this session" are established literature cited from knowledge and must be link-verified before this doc is marked reviewed.

Sources

Verified this session via web search (titles/URLs returned by search; page contents not always independently fetched):

• Bessen, J. & Meurer, M., "The Direct Costs from NPE Disputes," Cornell Law Review (2012 working paper; published 2014). SSRN abstract 2091210. $29B direct NPE costs in 2011. — primary (academic)
• Schwartz, D. & Kesan, J., critique of Bessen–Meurer methodology ("Analyzing the Role of Non-Practicing Entities in the Patent System," 2014). — primary (academic)
• IAM Media, "The cost of NPE disputes in the US and the $16.8 billion gap between RPX and Bessen & Meurer." — secondary
• Alice Corp. v. CLS Bank International, 573 U.S. 208 (2014), via Justia/Cornell LII. — primary (case law)
• USPTO Office of the Chief Economist, "Adjusting to Alice: USPTO patent examination outcomes after Alice Corp. v. CLS Bank" (April 2020). +31% §101 first-action rejections; +26% uncertainty; 2019 PEG effects. — primary (government)
• "A rabbit hole to innovation land: An empirical examination of the Alice decision," Computer Law & Security Review (ScienceDirect, 2024). No negative R&D effect from Alice. — primary (academic)
• Goodwin, "Impact of the Alice v. CLS Bank Decision – A Year-End Review" (Dec 2014). 32 district-court + 4 Fed. Cir. invalidations in ~6 months. — secondary
• Lex Machina data via BIOtechNOW/IPWatchdog (2014–2015): Sept 2014 filings 329 vs. 549 Sept 2013 (−40%). — secondary (reporting primary data)
• GAO-13-465, "Intellectual Property: Assessing Factors That Affect Patent Infringement Litigation Could Help Improve Patent Quality" (Aug 2013). Software patents ≈ 89% of 2011 defendant increase; defendants +129% 2007–2011. — primary (government)
• Unified Patents, "Patent Dispute Report: 2025 in Review" (Jan 2026). NPEs 55.4% of 2025 district-court cases; 90.3% of high-tech suits. — secondary (industry data; Unified sells anti-NPE services — interested party)
• RPX Corp. quarterly reviews (2024–2025). 1,889 NPE-campaign defendants added 2024 (+21.6%); H1 2025 +30% YoY. — secondary (industry data; RPX is a defensive aggregator — interested party)
• AIPLA, Report of the Economic Survey 2023 (and 2025 edition, published Feb 2026). Defense medians $600K–$3.6M through trial. — primary (survey)
• FTC, "Patent Assertion Entity Activity: An FTC Study" (Oct 2016). Litigation PAEs: 96% of suits, ~20% of revenue; settlements typically <$300K. — primary (government)
• Graham, S., Merges, R., Samuelson, P. & Sichelman, T., "High Technology Entrepreneurs and the Patent System: Results of the 2008 Berkeley Patent Survey," Berkeley Technology Law Journal 24:1255 (2009). — primary (academic)
• Bloomberg Law, "Fortress' Billions Quietly Power America's Biggest Legal Fights"; Bloomberg/Reuters coverage of VLSI v. Intel ($2.18B verdict 2021, reversed; $948.8M verdict 2022; 2025 Fortress-control jury finding). — secondary
• Bloomberg Law, "West Texas Spreads Patent Case Duties, Curbing Judge Albright"; Akin/Goodwin analyses of the July 25, 2022 W.D. Tex. order. Albright ≈25% of US patent suits; W.D. Tex. <100 → ~1,000 filings/yr by 2021. — secondary
• Forbes (Jones, 2016) and Bloomberg (2021) on IBM IP income: >$1B/yr most years since 1996, ~$27B cumulative, $626M by 2020. — secondary
• S.1546, Patent Eligibility Restoration Act of 2025, 119th Congress (Congress.gov); Tillis/Coons press releases (May–June 2025); Patently-O, "PERA 2025: Eligibility Reform Returns to Capitol Hill" (Oct 2025). — primary (legislative) / secondary
• EPO practice sources: EPC Art. 52; T 1173/97; T 641/00 (COMVIK); practitioner analyses (Bardehle, Mewburn, NLO, Bird & Bird). — primary (treaty/case law) / secondary

Cited from established literature, not re-verified this session (mark resolves before status: reviewed):

• Bessen, J. & Meurer, M., Patent Failure: How Judges, Bureaucrats, and Lawyers Put Innovators at Risk, Princeton University Press (2008). — primary; unverified this session
• Boldrin, M. & Levine, D., "The Case Against Patents," Journal of Economic Perspectives 27(1):3–22 (2013). — primary; unverified this session
• Bessen, J. & Hunt, R., "An Empirical Look at Software Patents," Journal of Economics & Management Strategy 16(1) (2007). — primary; unverified this session
• Heller, M. & Eisenberg, R., "Can Patents Deter Innovation? The Anticommons in Biomedical Research," Science 280:698 (1998); Heller, M., The Gridlock Economy (2008). — primary; unverified this session
• Bessen, J., Ford, J. & Meurer, M., "The Private and Social Costs of Patent Trolls," Regulation (2011–12) (~$500B lost share value 1990–2010). — primary; unverified this session, contested
• Kiebzak, S., Rafert, G. & Tucker, C., "The effect of patent litigation and patent assertion entities on entrepreneurial activity," Research Policy (2016) (~$21.8B forgone VC estimate). — primary; unverified this session, weak evidence
• Farre-Mensa, J., Hegde, D. & Ljungqvist, A., "What Is a Patent Worth? Evidence from the U.S. Patent 'Lottery'," Journal of Finance (2020). — primary; unverified this session
• TC Heartland LLC v. Kraft Foods Group Brands LLC, 581 U.S. 258 (2017); post-decision EDTX share decline. — primary (case law); aftermath statistics unverified this session
• Octane Fitness v. ICON Health & Fitness, 572 U.S. 545 (2014); eBay v. MercExchange, 547 U.S. 388 (2006); Bilski v. Kappos (2010); Mayo v. Prometheus (2012); State Street Bank v. Signature Financial (Fed. Cir. 1998). — primary (case law); unverified this session
• Innovation Act, H.R. 9 / H.R. 3309 (House passage 325–91, Dec 2013; died in Senate). — primary (legislative); unverified this session
• Samsung ice rink in Marshall, TX (The Guardian, 2015); Microsoft Android licensing revenue; Intellectual Ventures fund sizes; USPTO fee funding (~$4B/yr); PTAB/IPR cost and outcome statistics; PREVAIL Act (2025); D. Del. Connolly funding-disclosure standing orders (2022); pharma lobbying totals (OpenSecrets). — secondary; all unverified this session