US 8676742

I'll verify the patent's identifying details and check for any 2026 litigation or appellate activity before summarizing.

Let me pull the complete claim set and verify whether any Federal Circuit (CAFC) activity exists.

US Patent 8,676,742 — verification and summary

Identifier discipline note: all numbers below are reproduced literally as they appear in the sources. Where a source is truncated or silent, I say so rather than filling the gap.

Bibliographic data

Abstract

Uncertainty flag: the authoritative Google Patents text supplied to me does not contain an "Abstract" section — it ends mid‑description ("…topic object saves are counted in addition to content object saves"). I therefore do not have the verbatim abstract of US 8,676,742 and will not reconstruct one. The closest authoritative statement in my source, the patent's own Summary of the Invention, reads:

"a method and system for transforming non-adaptive systems or inadequately adaptive systems into adaptive systems is disclosed. The transformation may include generating adaptive contextualizations by converting one or more originating computer-implemented structures into an integrated fuzzy network-based structure. Computer-implemented functions may perform beneficial knowledge and expertise discovery, as well as other functions, against the adaptive contextualization. Methods and systems for generating personalized recommendations with an enhanced capacity for beneficial serendipity may be applied to enhance personalization functions."

Background framing: prior recommender systems "inhibit beneficial serendipity" by biasing recommendations toward a narrow set of inferred interests; the invention aims to retain personalization while promoting serendipity.

Independent claims (plain language)

The patent has 20 claims; three are independent — claims 1, 8, and 15. (Claim text below is drawn from the Justia claim listing; I have not re-derived it.)

Claim 1 — method (broadest)

1. Invoke a "contextual scope function" running on a processor-based computing device that establishes a contextual neighborhood inside a computer-implemented fuzzy network-based structure.
2. Receive a recommendation produced by a computer-implemented discovery function, where the recommendation is generated according to (a) that contextual neighborhood and (b) an inference from a plurality of usage behaviors.
3. Receive the recommendation where it is generated according to the contextual neighborhood, the neighborhood being based on a selected object.

Note: claim 1 is drafted with two successive "receiving a recommendation" elements; the third element is what ties the neighborhood to a selected object. Claim 1 does not itself recite user access, scope-setting, or transformation — those appear in dependents.

Claim 8 — system

• A contextual scope function (processor-based) establishing a contextual neighborhood within a fuzzy network-based structure;
• a contextual neighborhood scoping function that sets the scope of that neighborhood at a level set by a user;
• a discovery function generating a recommendation according to the contextual neighborhood and an inference from a plurality of usage behaviors.

Claim 15 — system, with affinity vector and expertise inference

• A contextual scope function establishing a contextual neighborhood comprising a plurality of objects within a fuzzy network-based structure;
• a computer-implemented affinity vector of affinity values between a user and those objects, generated from inferences from a plurality of behaviors — at least one inference being an inference of a level of expertise;
• a discovery function generating a recommendation according to the contextual neighborhood and the affinity vector.

Selected dependents of note: claim 2 (scope established by the recommendation recipient); claim 6 (inference is of a level of expertise); claim 7 (recommendation generated in response to a search request); claim 11 (selected object chosen based on access by the recipient); claim 16 (expertise inference over a plurality of usage behavior categories); claim 20 (a transforming function that generates the fuzzy network-based structure from one or more originating structures — the structural-transformation feature that is central to the specification but absent from claim 1).

Key disclosed concepts (specification)

Fuzzy network = nodes with relationships that have affinity by degree (normalized 0–1), versus binary hierarchies. Objects are topic objects or content objects; relationships carry relationship indicators and can be asymmetric and multi‑typed. The specification describes affinity vector types (TTAV, CCAV, CTAV, MTAV, MMAV, MTEV), a transformational protocol mapping views/folders/subject tags from originating systems (e.g., SharePoint‑like folder hierarchies, tag taxonomies) into a fuzzy content network, metrics including AECR, SFR, CSFR and FWSO, standardized/portable MTAVs and MTEVs, a candidate recommendation score harmonization function across multiple source systems, and a "learning layer" providing knowledge discovery and expertise discovery.

Litigation and CAFC status

• No Federal Circuit (CAFC) 2026 docket for 8,676,742 was found. My searches returned no appeal, no IPR/PTAB proceeding, and no CAFC briefing for this patent number. I flag this as a negative result from limited searching, not as proof of absence.
• District court litigation exists. Google Patents flags "Family has litigation" and cites Texas Eastern District Court case 2:26‑cv‑00303, Many Worlds 2T Innovations LLC v. Spotify Technology S.A. et al, filed 04/16/2026 (Judge Rodney Gilstrap; Markman set 11/16/2027; jury selection 5/8/2028). Reported assertions map the '742 patent's claim 1 elements onto Spotify's "Two Tower (2T)" embedding/nearest-neighbor recommendation engine. One complaint-analysis source references "Claims 1, 15, and 17" in proximity to the '742 discussion, but the snippet was truncated; treat the specific asserted-claim set as unconfirmed.
• A second suit, Many Worlds 2T Innovations LLC v. OpenAI OpCo, LLC, EDTX 2:26‑cv‑00774, filed Aug 31, 2026, asserts five patents including 8,676,742 against ChatGPT Memory, Search, Deep Research, Projects, Shopping Research, OpenAI Vector Stores and File Search.

Two consistency issues you should be aware of

1. Your stated current date (April 26, 2026) conflicts with my operating date (Sept 29, 2026) and with the retrieved evidence. The OpenAI complaint (filed Aug 31, 2026), the Sept 2, 2026 IPWatchdog article, and a case-tracker page "Last reviewed: Sep 27, 2026" all post-date April 26, 2026. If April 26, 2026 is the intended "as of" date, the OpenAI action and the later Spotify scheduling orders fall outside it.
2. Assignee vs. plaintiff naming. The patent record shows the current assignee as "Many Worlds 2T Innovations LLC" (assignment recorded 2026‑04‑15), while the pre‑2026 owner was ManyWorlds, Inc. Both names appear in the sources; they are distinct entities in the chain of title.

I'll search for litigation involving US patent 8676742 across litigation-specific sources.

Let me dig deeper into the specific cases and any other defendants.

Let me check for other cases, earlier suits by the original assignee, and any PTAB/IPR proceedings.

Litigation Involving U.S. Patent No. 8,676,742

Note on interpretation: I have taken the patent number literally as 8,676,742 (not 8,655,829, 8,676,742-family cousins, or other similar numbers), and report only assertions of that exact patent.

Patent reference: US 8,676,742 B2, "Contextual Scope-Based Discovery Systems," inventors Steven Dennis Flinn and Naomi Felina Moneypenny; original assignee ManyWorlds, Inc.; current assignee Many Worlds 2T Innovations LLC (assignment recorded 2026-04-15); filed 2011-10-07; issued 2014-03-18; priority 2011-03-29. The Google Patents page for this patent itself carries a litigation flag linking to EDTX case 2:26-cv-00303.

I identified two known litigations asserting the '742 patent, both filed in 2026 by the current owner, Many Worlds 2T Innovations LLC, in the Eastern District of Texas, Marshall Division, before District Judge Rodney Gilstrap. Neither has reached an outcome.

Case 1 — Many Worlds 2T Innovations LLC v. Spotify Technology S.A., Spotify AB, and Spotify USA Inc.

Current status detail: Defendants filed a Motion to Dismiss Plaintiff's Claims or Transfer Plaintiff's Claims Against Spotify USA Inc. and Spotify AB to the Southern District of New York (Dkt. 26, filed 07/16/2026), supported by a sealed declaration and exhibits (Dkts. 30–31, filed 07/29/2026); an unopposed motion for extension of time to respond followed. A Docket Control Order (signed 7/23/2026) sets Markman/claim construction for 11/16/2027, pretrial conference for 4/3/2028, and jury selection for 5/8/2028. No claim construction, summary judgment, trial, or appeal rulings yet.

Infringement theory (as alleged): The complaint maps the '742 claim limitations (e.g., "invoking a contextual scope function… establishes a contextual neighborhood within a computer-implemented fuzzy network-based structure"; receiving a recommendation generated by a "discovery function… in accordance with the contextual neighborhood and an inference from a plurality of usage behaviors") onto Spotify's use of contextual scope functions/contextual neighborhoods in fuzzy-network-based graph structures via node2vec algorithms, combined with behavior-based inferences.

Caveat: The Ex Parte AI Lab complaint analysis for 2:26-cv-00303 contains a section titled "U.S. Patent No. 8,655,829 Infringement Allegations" alongside its "U.S. Patent No. 8,676,742 Infringement Allegations" section. I could not fully reconcile whether 8,655,829 is separately asserted in this case or whether that label is an error in the secondary source. I am reporting the '742 assertion, which is directly corroborated by the Google Patents litigation flag for 8,676,742 pointing to this docket number.

Case 2 — Many Worlds 2T Innovations LLC v. OpenAI OpCo, LLC

Current status detail: Complaint filed 8/31/2026 with jury demand, AO 120 form, corporate disclosure statement, and summons issued the same day. Notices of attorney appearance for plaintiff were filed 8/31/2026 and 9/3/2026 (counsel from the same firm appearing in both cases/Cadwell, Kevin et al.). No answer, motion, claim construction, or outcome has been reported as of the last review dates in the sources consulted. Relief sought: past and continuing royalties, enhanced damages for alleged willful infringement, a jury trial, and an injunction.

Infringement theory for the '742 patent (as alleged): The complaint focuses on OpenAI's vector-store retrieval indexes that organize embedded content according to degrees of relatedness, and points to relevance thresholds, maximum-result settings, reranking, and hybrid semantic-keyword search as corresponding to the claimed mechanisms for setting the scope of a contextual neighborhood, combined with inferences drawn from users' prior behavior.

Negative findings and limitations

• No earlier litigation asserting the '742 patent appeared in my searches. The patent's own prosecution family (ManyWorlds, Inc.) shows the asserted-against products in the 2026 suits only; my searches did not surface any earlier complaint by ManyWorlds, Inc. asserting 8,676,742.
• No PTAB proceedings (IPR/PGR/CBM) challenging 8,676,742 were surfaced. I reached my search-step limit before I could exhaustively query the Unified Patents litigation caselist, PACER, the CAFC docket, and PTAB APIs, so I cannot state with high confidence that none exist — treat this as "none found," not "none exists."
• No outcomes: both cases are pending as of the most recent source dates (the AI Lawsuit Tracker page was last reviewed Sept 27, 2026). Neither has produced a judgment, settlement, dismissal with prejudice, or appeal.
• Both suits are asserted by a non-practicing-entity-style owner; commentators (IPWatchdog) note that a Section 101 eligibility challenge to the patents is "virtually guaranteed" in the OpenAI case.

Primary sources:

• Google Patents, US8676742B2 (litigation flag → EDTX 2:26-cv-00303): https://patents.google.com/patent/US8676742/en
• Unified Patents litigation link for the family: https://portal.unifiedpatents.com/litigation/Texas%20Eastern%20District%20Court/case/2%3A26-cv-00303
• PacerMonitor docket, 2:26-cv-00303 (Spotify): https://cdn.pacermonitor.com/public/case/64166196/Many_Worlds_2T_Innovations_LLC_v_Spotify_Technology_SA_et_al
• Ex Parte AI Lab docket + complaint analysis, 2:26-cv-00303: https://ai-lab.exparte.com/case/dct/txed/2:26-cv-00303/many-worlds-2t-innovations-llc-v-spotify-technology-sa
• PacerMonitor docket, 2:26-cv-00774 (OpenAI): https://cdn.pacermonitor.com/public/case/66548709/Many_Worlds_2T_Innovations_LLC_v_OpenAI_OpCo,_LLC
• AI Lawsuit Tracker case page, 2:26-cv-00774: https://ailawsuittracker.com/cases/many-worlds-2t-innovations-llc-v-openai-opco-llc-2-26-cv-00774/
• IPWatchdog, "OpenAI Sued Over AI Patents Covering Retrieval, Personalization and Content Generation," Sept 2, 2026: https://ipwatchdog.com/2026/09/02/openai-sued-over-ai-patents-covering-retrieval-personalization-content-generation/

If you want, I can attempt a follow-up pass (subject to a fresh tool budget) specifically targeting the Unified Patents caselist, PACER for any additional 2026 Many Worlds 2T complaints, and PTAB for IPR petitions against 8,676,742.

I'll verify the canonical "no PTAB activity" signal against live sources before writing.

PTAB / AIA Trial Activity — US 8,676,742 ("Contextual Scope-Based Discovery Systems")

Bottom line up front: The canonical structured source (USPTO Open Data Portal, as furnished in the "PTAB proceedings on file" block) reports no AIA trial proceedings — no IPR, PGR, or CBM — against US 8,676,742. My independent web checks surfaced nothing contradicting that. No IPR on this patent has ever been instituted, denied, settled, or decided. There is therefore no claim-level PTAB outcome, no judge panel, and no Federal Circuit appeal from an AIA trial to report for this patent.

Proceedings overview

Total AIA trial proceedings on US 8,676,742: zero (0 active / 0 claims invalidated / 0 claims sustained / 0 settled / 0 institution denied). The defensive posture this gives a defendant is neither "hardened patent" nor "dead claims" — it is an unlitigated patent in front of the Board: every one of its claims remains untested at the PTAB, and the full IPR menu is still on the table with no § 315(e)(2) estoppel and no § 315(b) bar attaching to anyone except a party already served more than one year ago.

Per-proceeding detail

None. There are no proceeding numbers to enumerate, and I will not invent them. Consistent with the operating instruction, the default here is the correct answer: no PTAB activity on file.

For completeness, the checks I ran (all negative):

One trap worth naming explicitly, because it appears in the source data and is easy to misread: the Google Patents page for the '742 carries a Darts-ip "Global patent litigation dataset" link and a Unified Patents litigation-data link. That is a data-source attribution, not a party. Unified Patents is not a petitioner here, and there is no defensive aggregator anywhere in the chain. The current assignee is Many Worlds 2T Innovations LLC, recorded 2026-04-15 as an assignment of assignor's interest from ManyWorlds, Inc. — i.e., a monetization vehicle, not a defensive aggregator.

Adjacent activity a defendant must not confuse with AIA trials

These are not AIA trial proceedings and must not be cited as such, but they materially shape a validity defense:

1. Family ex parte appeal decisions under § 101 (BPAI/PTAB Appeal Board, not AIA trials). Several sibling ManyWorlds/Flinn–Moneypenny applications were affirmed as directed to abstract ideas with only generic computer hardware:

   • Appeal 2017-011693 (rejection of claims 9–17 under § 101 affirmed; the Board applied the 2019 PEG and found the claims directed to "generating recommended objects in accordance with inferences of preference based on usage behaviors" — a practice "similar to those found by the courts to be abstract"). Source PDF: https://e-foia.uspto.gov/Foia/RetrievePdf?system=BPAI&flNm=fd2017011693-03-15-2019-1
   • Appeal 2018-003233 (Application 13/270,019; claims 5, 14, and 20 — all independent — rejected under § 101 and AFFIRMED; panel: Fischetti, Wieder, Schopfer, with Wieder writing). Mirror: https://drive.google.com/file/d/1x_ygRp1mwSVK0ZeBTBqaQztqW0iagSU1/view
   • Appeal 2018-005988 (Application 13/269,958; claim 7 rejected under § 101 and AFFIRMED; Board held the generic recitation of a "mobile processor-based device," "processor-based device," and "computer-implemented neural network" did not supply an inventive concept). Mirror: https://casetext-staging-alt.com/admin-law/steven-dennis-flinn-et-al-2

   Caveat, stated precisely: these decisions concern different applications and different claims — they do not adjudicate any claim of the '742, and they carry no estoppel. Their value is evidentiary and rhetorical: they show this specification family repeatedly failed the Alice/Mayo step-two analysis on the same "infer preferences from usage behaviors and recommend" core.

2. District court assertion. The '742 is one of five patents asserted in Many Worlds 2T Innovations LLC v. OpenAI OpCo, LLC, E.D. Tex. No. 2:26-cv-00774, complaint filed 2026-08-31, assigned to Judge Gilstrap, jury demand, direct + induced + willful infringement, royalties, enhanced damages and injunction. Asserted patents: 8,676,742; 8,843,433; 10,699,202; 12,307,388; 12,299,603. The '742 theory is pleaded against ChatGPT / ChatGPT Memory / Search / Deep Research / Projects / Shopping Research and OpenAI Vector Stores + File Search, and is built on vector-store retrieval indexes that organize embedded content by degrees of relatedness, plus relevance thresholds, max-result settings, reranking and hybrid semantic-keyword search as the claimed "contextual neighborhood" scope controls. Sources: https://ipwatchdog.com/2026/09/02/openai-sued-over-ai-patents-covering-retrieval-personalization-content-generation/ and https://ailawsuittracker.com/cases/many-worlds-2t-innovations-llc-v-openai-opco-llc-2-26-cv-00774/.

3. Docket reference on the family page. Google Patents lists E.D. Tex. case 2:26-cv-00303 as family litigation, and a Darts-ip entry as "First worldwide family litigation filed." The fetched page does not identify the parties or the patents asserted in 2:26-cv-00303; I am not going to guess at the caption.

Strategic summary

Claim-by-claim status: ALL CLAIMS UNTESTED. Because no AIA trial was ever instituted against the '742, there are no canceled claims, no sustained claims, and no narrowed claim set to work with. Unlike the typical software-patent scenario — where an IPR leaves a defendant a clean list of dead independent claims and a thin set of surviving dependents — here the entire claim set is live in district court. Whatever the '742 claims, its full scope is intact and must be attacked on § 101, § 102/§ 103 in the district court, or in a new IPR.

Estoppel landscape: clean slate, and one clock running. With zero instituted IPRs, 35 U.S.C. § 315(e)(2) estoppel is nonexistent — no petitioner, privy, or real party in interest is barred from raising any § 102/§ 103 ground, in the district court or at the Board. Practically:

• § 315(b) one-year bar: a party served with a complaint alleging infringement of the '742 more than one year ago is time-barred from filing an IPR. For anyone served on/after roughly 2025-09-29, the window is open. For OpenAI (served on or about 2026-08-31 in 2:26-cv-00774), the § 315(b) deadline runs to approximately 2027-08-31 — but note that serving a complaint starts the clock only as to the party served; a supplier/customer relationship can create privity problems, so confirm the RPI chain before relying on a "we're not the served party" argument.
• § 315(e)(1) and § 325(e)(1) are equally inapplicable, so no risk of the patent owner extracting a litigation-conduct estoppel admission.

Statutory routes still available against this patent:

• IPR — available. The '742 issued 2014-03-18 from an application filed 2011-10-07, so it is pre-AIA and IPRs may be filed at any time on § 102/§ 103 grounds limited to patents and printed publications. Google Patents reports an adjusted expiration of 2032-07-27, so there is roughly six years of runway.
• PGR — unavailable. PGR applies only to patents subject to first-inventor-to-file; the '742 is a pre-AIA filing.
• CBM — unavailable forever. CBM review applies only to "covered business method" patents, and the Office stopped accepting CBM petitions filed on or after 2020-09-16. Nobody filed one against the '742 before the sunset.
• Ex parte reexamination — available to anyone, anonymously, on patents-and-printed-publications art, at a far lower cost, with no estoppel and no § 315(b) clock. This is the overlooked lever here: with no IPR on file and no estoppel risk, an anonymous reexam request can be used to test the art cheaply and to build a record before committing to an IPR.

Pattern signals. (a) No serial petitioner — no entity has filed even one petition against this patent. (b) The patent owner is not a PTAB appellant in AIA trials — its Board appearances in this family are ex parte prosecution appeals, where it lost repeatedly on § 101 (§ 101 affirmances in Appeals 2017-011693, 2018-003233, 2018-005988). (c) No defensive aggregator — Unified Patents' name appears only as the licensor of the litigation dataset embedded on the Google Patents page; it is not a party. (d) Assertion cadence is escalating against AI, not against the prior targets of this portfolio — a single 2026 complaint against OpenAI covering five patents including the '742.

Recommended next steps

1. Do not wait for a PTAB record that does not exist. There is no FWD to cite, so a district court § 101 motion and/or a fresh IPR is the whole defense. If you are a defendant, say so in your invalidity contentions rather than assuming prior PTAB attack.
2. Lead with § 101. The strongest available leverage is the family's own prosecution record: the Board affirmed abstract-idea rejections in Appeal 2018-003233 (claims 5, 14, 20 — all independent — of Application 13/270,019, affirmed), Appeal 2017-011693 (claims 9–17, affirmed), and Appeal 2018-005988 (claim 7, affirmed). Quote the holdings that generic "processor-based device" and "computer-implemented neural network" recitations supply nothing inventive beyond the abstract idea. Be scrupulous: these are different claims of different applications — persuasive authority, not estoppel, and not an adjudication of the '742.
3. File the IPR inside the § 315(b) window if you were served on or after 2025-09-29, and identify every real party in interest and privy carefully. With no prior institution there is no § 315(e)(2) downside, and a stipulation not to pursue the same grounds/art in district court (Sand Revolution style) materially improves your Fintiv posture.
4. Consider an anonymous ex parte reexamination first. Zero estoppel, zero § 315(b) clock, no announcement of your invalidity theory — attractive where no one has yet run any art against this patent.
5. Calendar the OpenAI-side milestones. Complaint 2026-08-31; § 315(b) IPR bar approximately 2027-08-31 for the served party; any IPR instituted in that window faces the statutory one-year trial deadline from institution (§ 316(a)(11)), extendable up to six months for good cause. Given the AI-sector scrutiny of this suit, additional defendants and additional pretrial motion practice are likely.
6. Note the litigation reference you cannot yet name: E.D. Tex. 2:26-cv-00303 is listed as family litigation on the '742's Google Patents page with no caption in the fetched data. Pull the docket via PACER before making any coordination or privity assumption.

Source and confidence notes

• Canonical and authoritative for the count: the USPTO Open Data Portal "PTAB proceedings on file" block supplied in the prompt — zero proceedings. Patent bibliographic data taken from https://patents.google.com/patent/US8676742/en (filed 2011-10-07; priority 2011-03-29; issued 2014-03-18; adjusted expiration 2032-07-27; assignee Many Worlds 2T Innovations LLC per 2026-04-15 reassignment).
• Verified by search: the OpenAI complaint and asserted-patent list (IPWatchdog, AI Lawsuit Tracker); the three § 101 ex parte appeal affirmances (USPTO e-FOIA PDF and two non-authoritative mirrors — the mirror URLs are cited for convenience, and the e-FOIA PDF is the primary record for Appeal 2017-011693).
• Not verified / not asserted: any PTAB proceeding number, panel, institution decision, FWD, settlement, or Federal Circuit docket for the '742 — because none exists. If you need a negative confirmed from the primary source of record rather than the ODP block, the PTAB E2E / Patent Trial and Appeal Board End-to-End docket and the USPTO PatentCenter "Post-Grant Proceedings" tab for RE44,968-family members are the places to pull the certified record before filing anything that depends on the absence.

I'll research this patent's ownership chain. Let me start with parallel searches.

Let me dig for the actual assignment reel/frame records and the NPE-related directories.

US 8,676,742 B2 — "Contextual Scope-Based Discovery Systems"

Ownership-chain reconstruction and NPE analysis

Critical sourcing caveat up front: I could not complete a live, session-based query of USPTO Assignment Center (https://assignmentcenter.uspto.gov/ — the legacy indexed endpoint is https://assignment.uspto.gov/patent/index.html). That interface requires an interactive session, and no web-indexed mirror surfaced reel/frame numbers, execution dates, recording dates, or the recording correspondent for this patent. The chain below is reconstructed from the Google Patents legal-events record for this patent (https://patents.google.com/patent/US8676742/en) plus the litigation record. Where a field is not verifiable, I say so rather than filling it in. Do not treat the reel/frame slots below as populated.

Inventors

• The two are the only named inventors, and they are the same pair that appears across essentially the entire ManyWorlds portfolio (43 applications; Flinn and Moneypenny are the top two inventors by count on GoodIP's assignee page for ManyWorlds, Inc.).
• Employer attribution: the inventors executed an assignment of their rights to ManyWorlds, Inc. at the filing date (2011-10-07), and the patent's attorney/agent of record is listed as "HOLMES, MICHAEL B" / firm "MANYWORLDS, INC. (BRENHAM, TX, US)" — i.e., prosecution was handled through the assignee's own address rather than an outside firm of record. Both facts point to Flinn and Moneypenny being ManyWorlds personnel (Flinn being the well-documented founder/principal of ManyWorlds). I am stating this as an inference from the assignment and correspondence records, not from an employment document I retrieved.
• Unusual-pattern check: I found no evidence that either inventor departed before the 2026 transfer, and no employment records were retrievable. However, the portfolio did not sit with the inventors — see the "Nokia" loose end under Signal 2. I cannot confirm a "both inventors gone within 12 months" pattern. Not established.

Original assignee

ManyWorlds, Inc. — assignee of record on the issued patent. Front-page correspondence address: Brenham, TX.

• Primary line of business: enterprise software/consulting for "adaptive" knowledge and recommendation systems. The patent's own specification is oriented to commercial knowledge-management deployments (it likens its originating structures to those "available, for example, in a Microsoft® SharePoint® system," and describes transformations of SharePoint-style views, folders, and subject taxonomies). ManyWorlds' published portfolio is broad — adaptive recommendations, social/process networks, generative investment process, adaptive digital asset management, location-aware adaptive systems.
• Did they ship a product embodying the claims? Not determinable from what I retrieved. The specification describes a deployable "learning layer," and ManyWorlds marketed adaptive-recommendation technology, but I found no evidence of a commercial product reading on claim 1 (the contextual-scope-function / fuzzy-network-neighborhood claim asserted against Spotify). Treat product-shipping as unproven.
• Current status: Unclear / appears dormant. I found no current operating web presence, no SEC filings (ManyWorlds appears to have been privately held), and no bankruptcy docket. What is documented is that it divested rights to this patent in April 2026 to Many Worlds 2T Innovations LLC. I will not characterize it as dissolved or bankrupt without a record.

Assignment timeline

Only two assignment events are visible in the record I could retrieve. Neither exposes a reel/frame to me.

• 2011-10-07 (executed; date shown coincides with the filing date) — Reel not retrieved / Frame not retrieved

  • Conveyance: Assignment of assignors' interest (inventor→company assignment)
  • Assignor: FLINN, STEVEN DENNIS; MONEYPENNY, NAOMI FELINA
  • Assignee: MANYWORLDS, INC.
  • Correspondent: Not retrieved. (Note: the patent's prosecution correspondent of record is Michael B. Holmes, c/o ManyWorlds, Inc., Brenham, TX — that is a different field from the assignment-recording correspondent and I did not confirm Holmes filed this recording.)
  • Context: Standard inventor-to-employer assignment at filing; not a transfer between companies.

• 2026-04-15 (date shown is the legal-event/recording date; execution date not retrieved) — Reel not retrieved / Frame not retrieved

  • Conveyance: "Assignment of Assignor's Interest"
  • Assignor: MANYWORLDS, INC.
  • Assignee: MANY WORLDS 2T INNOVATIONS LLC (Google Patents renders the current assignee as "Many Worlds 2t Innovations LLC")
  • Correspondent: Not retrieved. This is the single most important missing field for the analysis you asked for — the recording attorney on this link is the tell for whether the 2T entity sits inside a known assertion shop. I could not obtain it.
  • Context: Transfer-to-asserter. All rights moved from the original operating assignee to a litigation vehicle; the assignee entity filed infringement suits within days/weeks.

If the Assignment Center returns no further records on your live query, that is consistent with what I found — the chain is just two links. I am not aware of any intervening recorded assignment (e.g., a security agreement, merger, or license) between 2011 and 2026.

Timeline diagram

timeline
    title Ownership of US 8676742
    2011 : Filed by ManyWorlds Inc
         : Inventors assign rights to ManyWorlds
    2014 : Patent issued
    2026 : Assigned to Many Worlds 2T Innovations LLC
         : Suit filed against Spotify
         : Suit filed against OpenAI

NPE / troll-pattern signals

1. Shell-entity transfer — PRESENT
The patent moved from an operating assignee (ManyWorlds, Inc., which had a going technology/consulting business and a 43-application portfolio) to Many Worlds 2T Innovations LLC on 2026-04-15. The strongest evidence is not the naming alone but the name-to-accused-product match: the LLC's "2T" corresponds directly to the accused instrumentality in its first suit — Spotify's "Two Tower (2T) recommendation engine" (Compl. ¶55 in EDTX 2:26-cv-00303). An entity named after the specific accused technology, formed to hold a single acquired patent, is a concrete purpose-formation indicator, not a name-suffix guess. Supporting: no evidence of products sold by the LLC; the LLC is the plaintiff of record in multiple suits. Caveat: I have no recorded registered-agent address and no state LLC formation record, so the "registered-agent service address / single-member Delaware or Texas LLC" prong is unverified.

2. Known asserter in the chain — NOT ON THE PUBLISHED LISTS (but high-frequency-plaintiff profile)
Neither ManyWorlds, Inc. nor Many Worlds 2T Innovations LLC matches the named lists you supplied (Acacia, Marathon, IV, IPNav, Wi-LAN, Mosaid/Conversant, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, Document Generation Corp, Spangenberg entities) based on the sources I retrieved. However, Many Worlds 2T Innovations LLC now meets the behavioral definition of a serial plaintiff: it asserted this patent plus four others (US 8,843,433; 10,699,202; 12,307,388; 12,299,603) against OpenAI OpCo, LLC in EDTX 2:26-cv-00774 (filed 2026-08-31) and against Spotify entities in EDTX 2:26-cv-00303. Unified Patents' litigation record carries the Spotify case (https://portal.unifiedpatents.com/litigation/Texas%20Eastern%20District%20Court/case/2%3A26-cv-00303). I could not complete a targeted check of the RPX and Unified Patents asserter directories (my remaining searches were cut off), so I am not asserting it appears on those lists — only that it behaves like a listed asserter.

Loose end worth verifying: Unified Patents' patent page for sibling patent US 8,515,900 B2 (Flinn/Moneypenny, original assignee Manyworlds Inc) lists "Parent Company: Nokia." If accurate, it implies the broader ManyWorlds portfolio was distributed to third parties — possibly including a Nokia-linked entity — which would matter for the whole family's assertion history. I could not corroborate this and treat it as unconfirmed data-artifact risk, not a finding.

3. Repeat correspondent across the chain — UNCLEAR / NOT VERIFIABLE
The only correspondence data point I obtained is the prosecution attorney of record: Michael B. Holmes, c/o ManyWorlds, Inc., Brenham, TX (per the patent's front page as indexed by FreePatentsOnline). That is one appearance, on a different field, and single appearances are explicitly not a finding under your rule. The recording correspondent for the 2026-04-15 link — the field that would show whether a repeat-player NPE attorney ran this transfer — was not retrievable, and with only one post-issuance link there is no recurrence to test. Cannot be assessed; do not read this as confirming or clearing the chain.

4. Cascading transfers — NOT PRESENT (on the retrievable record)
One post-issuance transfer only (ManyWorlds → Many Worlds 2T Innovations). No chained LLCs, no <24-month cascade. This is a single-step divestiture, not a daisy chain.

5. Pre-litigation transfer — PRESENT (strongly)
The transfer is recorded 2026-04-15; the first suit naming this patent, Many Worlds 2T Innovations LLC v. Spotify Technology S.A. et al., EDTX 2:26-cv-00303, carries an April 2026 filing date (docket sources show 04/16/26). A transfer landing within roughly a day-to-a-few-weeks of the first complaint is the classic clean-standing / assertion-enablement arrangement. Two caveats I owe you: (a) I retrieved the recording date, not the assignment execution date — an earlier execution date would soften the timing; and (b) I could not independently re-verify the exact complaint filing date from the docket image itself. Even on the loosest reading, the transfer falls inside your 6-month window.

6. Bankruptcy fire-sale — NOT PRESENT
No Chapter 7/11 record for ManyWorlds, Inc., and no sale-in-bankruptcy docket found. The April 2026 transfer is a private conveyance, not a court-supervised asset sale. Not established.

7. Privateering — NOT PRESENT
Privateering requires an operating company sponsoring suits against its competitors. Here the operating company (ManyWorlds) is the seller/exit party, not the beneficiary of the litigation, and there is no SEC disclosure (ManyWorlds appears to have been private) or Patent Progress/EFF coverage tying it to a competitor attack. Not established.

8. Defensive aggregator — NOT PRESENT
The chain terminates at an asserting plaintiff, not at RPX, AST, LOT, Unified Patents, or OIN. Inverse signal does not apply. Not present.

Verdict

NPE — high confidence

Two strong signals are present on the retrievable record. First, transfer-to-asserter: on 2026-04-15 all rights moved from operating assignee ManyWorlds, Inc. to Many Worlds 2T Innovations LLC, an entity whose name encodes the accused product itself — Spotify's "Two Tower (2T) recommendation engine" (Compl. ¶55, EDTX 2:26-cv-00303) — with no evidence the LLC sells anything. Second, pre-litigation transfer: that April 2026 recording precedes the first complaint naming this patent (EDTX 2:26-cv-00303, filed April 2026) and the later OpenAI complaint (EDTX 2:26-cv-00774, filed 2026-08-31) that re-asserts this patent among five. The chain is a single clean step — a deliberate feature, not a cascade.

Where my confidence is bounded: I could not retrieve reel/frame numbers, the 2026 execution date, the recording correspondent, or the LLC's registered-agent address. The recurring-correspondent test (your Signal 3) is unassessable, and the shell-entity prongs resting on state-formation records are unverified. Those gaps do not change the verdict, because the transfer-to-asserter and pre-suit-timing findings stand on their own — but they are exactly the fields to capture on a live Assignment Center pull.

Verify at: https://assignment.uspto.gov/patent/index.html (search patent number 8676742) and https://assignmentcenter.uspto.gov/ — retrieve reel/frame, execution date, and, critically, the correspondent of record on the 2026-04-15 entry. Cross-check the same correspondent against the other Many Worlds 2T Innovations patents (8,843,433; 10,699,202; 12,307,388; 12,299,603; 8,655,829; 9,165,259; 9,454,729) to convert Signal 3 from unclear to a finding.

I'll research US patent 8676742 and its prior art citations. Let me start with searches on the patent's citation records.

I found the cited references list. Let me gather the complete citation data from multiple sources.

Let me get the complete citation list and the patent's claims, which I'll need to assess anticipation.

I've reached my search budget. Below is a complete, honest write-up of what I found — including where the record is incomplete and where I cannot responsibly give a definitive answer.

Prior-Art Analysis — US Patent No. 8,676,742

1. Patent identification (literal match)

Litigation context (not prior art, but relevant): the family is in active assertion. A Unified Patents entry lists Texas Eastern District Court case 2:26-cv-00303, and a docket tracker lists Many Worlds 2T Innovations LLC v. OpenAI OpCo, LLC, 2:26-cv-00774, asserting this patent (with 8,843,433; 10,699,202; 12,307,388; 12,299,603).

Literal-ID caution: I interpreted "8676742" strictly. Search hits for the bare string "8676742" also surfaced (a) a Brazilian journal DOI (10.20396/zet.v32i00.8676742) and (b) a number inside an Israeli government PDF — these are unrelated and I have excluded them from the analysis. I did not auto-correct the number.

2. Important limitations you must know before relying on this

I want to be explicit rather than fabricate:

1. I could not retrieve the issued claim set of US 8,676,742. The authoritative text supplied was truncated before the claims (it ends mid-discussion of MTAV construction). The databases I reached returned the abstract, summary, and description, not the "What is claimed is" section. Therefore I cannot responsibly produce a claim-number-by-claim-number § 102 anticipation chart. Any such chart would be guesswork. What I give below is a feature-level relevance assessment, clearly labeled provisional.
2. The "References Cited" list I recovered is drawn from secondary databases (Justia and FreePatentsOnline), not directly from the patent's printed front page or USPTO PatentCenter. The lists were truncated in my results (both cut off mid-alphabet at "8…" / "2009…"). I could not recover the Foreign Patent Documents or the Non-Patent Literature ("Other Publications") sections at all.
3. Patent citations ≠ prior art. Many references appearing on a US patent's face are cited as background, or are the applicant's own earlier work. Under pre-AIA § 102 (which governs here — priority 2011-03-29 and filing 2011-10-07, both before the AIA's Mar. 16, 2013 first-to-file date), a reference is only § 102(a)/(e) art if it is "by others." The several Flinn et al. references are the same inventive entity and therefore generally are not § 102(a)/(e) art — they would only bite under § 102(b) if published more than one year before 2011-03-29.

3. What this patent appears to claim (scope, inferred)

From the summary/definitions (claims unavailable), the invention is directed to: transforming one or more non-fuzzy originating computer structures (folders, views, subject-tag taxonomies, social/activity streams) into an integrated fuzzy network / fuzzy content network ("adaptive contextualization"); generating affinity vectors (TTAV, CCAV, CTAV, MTAV, MMAV, MTEV) and relationship indicators (0–1 affinities); performing knowledge and expertise discovery against that structure; mapping/normalizing standardized tags to port MTAVs/MTEVs across systems; and generating personalized recommendations with enhanced beneficial serendipity, including recommending across multiple source systems with a score-harmonization function. Independent claims most likely recite a computer-implemented method/system performing this transformation + recommendation, with dependent claims on the affinity-vector math, standardized MTAV, and cross-system harmonization.

4. Prior-art references (US patent documents cited on the face)

4a. Same-inventor / same-family references (Flinn et al.) — most topically relevant, but likely NOT § 102(a)/(e) art

These are the closest in subject matter (fuzzy networks, affinity vectors, adaptive recommendations). Because they share the inventive entity, they generally are not "by others"; they matter only as § 102(b) art if published before 2011-03-29 (one-year bar) — those with dates after 2011-03-29 are not § 102 art at all.

4b. Co-pending sibling applications (same family, filed 2012) — NOT prior art

US 2012/0254074, /0254075, /0254079, /0254095, /0254096, /0254097, /0254098, /0254099 (all Oct. 4, 2012, Flinn et al.) and US 2012/0290518 (Nov. 15, 2012, Flinn et al.). These are contemporaneously filed family members, not prior art.

4c. Third-party US patents cited (potential genuine § 102 art, 1994–2008)

4d. Third-party US published applications cited

4e. Foreign Patent Documents & Non-Patent Literature

Not recovered. My searches did not return the patent's "Foreign Patent Documents" or "Other Publications" sections. I will not invent entries for them. If you need these, they must be pulled from the printed front page / USPTO PatentCenter for 13/268,007.

5. Bottom line on § 102 anticipation

• I could not identify a single reference that cleanly anticipates the independent claim(s), because (a) I do not have the issued claim text, and (b) most of the closest references are the applicant's own earlier patents, which are generally excluded from pre-AIA § 102(a)/(e) as not "by others."
• The externally-sourced references with the strongest § 102(b) potential against the recommendation/affinity aspects are US 6,912,505 (Linden, item-to-item collaborative filtering), US 5,867,799 (Lang), US 6,438,579 (Hosken), and, for behavior/expertise inference, US 6,922,672 (Hailpern) and US 7,188,153 (Lunt). These plausibly bear on claims drawn to computing item-to-item affinities and producing personalized recommendations, but not on the likely-distinctive features (non-fuzzy→fuzzy structural transformation, integrated cross-system contextualization, standardized-tag MTAV portability, cross-system score harmonization).
• The Flinn/ManyWorlds patents issued before 2011-03-29 (e.g., US 6,795,826; 7,493,294; 7,526,458/459/464; 7,539,652; 7,606,772) are the most material to the fuzzy-network subject matter and are real § 102(b) candidates if the examiner/infringer treats them as printed publications more than one year before the priority date — despite being same-entity.

To produce a defensible claim-by-claim § 102 chart, you need two things I could not obtain: (1) the verbatim issued claims of US 8,676,742, and (2) the complete face-of-patent citation list including Foreign Patent Documents and NPL. Once you supply the claims, I can map each reference above to specific claim elements and flag which limitations are met or missing.

I'll research the prior art referenced on this patent before building the § 103 analysis. Note upfront: the authoritative text supplied to me for this page does not actually include a Prior Art / "References Cited" section — it ends mid-description. So I need to retrieve the cited references, and I'll be explicit about what I can and cannot confirm.

The prior-art section did not come through in the authoritative text. Let me retrieve the actual "Referenced Cited" list for this patent.

I have a partial reference list. Let me pin down the actual "References Cited" and "Cited By" entries, plus the family's own overlapping disclosures.

Obviousness Analysis — US 8,676,742 under 35 U.S.C. § 103

Analyst: Senior US patent analyst
Date of analysis: Per my operating clock, 2026-09-29. ⚠️ The task prompt states "Current Date: April 26, 2026" while the system date is 2026-09-29. I flagged this same conflict in the previously generated sections, and I flag it again here rather than silently reconciling it. Nothing in this § 103 analysis turns on the difference.

0. Verification caveats you must read first

The premise cannot be fully satisfied. The task says to "use the results from the Prior Art section of this page." The authoritative Google Patents text supplied to me contains no Prior Art / "References Cited" section — it terminates mid‑description ("…topic object saves are counted in addition to content object saves"). The only "prior art" content present is Google's machine‑generated keyword line ("recommendation, user, content, computer, topic") and the "prior art date 2011‑03‑29," which is a date field, not a reference list.

Accordingly, I rebuilt the corpus from retrieved third‑party records. Every reference below is flagged with how it was sourced and whether I read it. I did not fabricate citation status, examiner treatment, or file‑history content.

A point of legal significance I can ground in the retrieved text: the '742 description internally references figures up to FIG. 45 ("FIG. 45 depicts various hardware topologies…", framework 2000 in FIG. 42, etc.) while its own "Brief Description of the Drawings" lists only FIG. 1–13. That is consistent with the '742 specification text being carried over from the earlier 45‑figure ManyWorlds family document. This matters enormously for § 103: the substance of the '742's own disclosure was published by the same inventor/assignee in 2006–2007, more than one year before the 2011‑03‑29 priority date, and is therefore § 102(b) prior art against claims that are not entitled to that earlier chain.

1. Legal framework applied

• Pre‑AIA § 102/§ 103 governs. Application 13/268,007 was filed 2011‑10‑07; AIA first‑inventor‑to‑file §§ 102/103 apply only to applications filed on or after 2013‑03‑16.
• Priority: 2011‑03‑29 (listed as both priority date and prior‑art date; FreePatentsOnline indicates benefit of provisional 61/469,052, "Methods and Systems of Expertise Discovery" — the day‑of‑month is truncated in my source). Critically, the '742 record shows no benefit claim to the 2003/2004/2006 ManyWorlds applications. The claims therefore rise or fall on a ~March 2011 date.
• Consequence: documents published or patented more than one year before 2011‑03‑29 — i.e., before ~2010‑03‑29 — are § 102(b) art. Under pre‑AIA § 103(c), § 102(e)/(f)/(g) art that is commonly owned is disqualified only as to those subsections; § 102(b) art is not disqualified, even when the patentee owns it. A patentee's own earlier patent or published application is fully available as § 103 art.
• Standard: Graham v. John Deere; KSR Int'l v. Teleflex (motivation may be a design need, market pressure, or "a finite number of identified, predictable solutions"); In re Kahn / MPEP 2143 (articulated reasoning with rational underpinning; predictable combination of known elements is sufficient).

2. The prior-art corpus I can actually cite

2a. ManyWorlds / Flinn family — the heart of the case

These documents appear in the '742's reference area and/or share the '742's specification lineage. All dates are as retrieved.

2b. Third‑party art appearing in the reference area

2c. Art the '742 specification itself admits and distinguishes (usable as admitted prior art)

The '742 text expressly describes and distinguishes US 5,790,426 (Robinson, collaborative filtering) — "limited to inferences associated with one type of usage behavior, the direct rating of informational items only"; US 6,285,999 (Page, node ranking in a linked database) — "limited to non‑fuzzy networks"; and US 5,875,446 (Brown et al., hierarchical grouping/ranking with weighted relationships) — "does not enable delivery of non‑hierarchical structures."

Note the asymmetry these admissions create: the patent concedes that weighted‑relationship graph retrieval with ranked context (Brown) and behavior‑based recommendation (Robinson) are old; it distinguishes them only on breadth of behaviors and structural plasticity. That is a narrow distinction, and it is the ground on which the ManyWorlds family's own later publications close the gap.

2d. Documents I will not treat as prior art

The '742's co‑family documents — US 8,600,926 (Integrated interest and expertise‑based discovery), US 8,645,312 (Expertise discovery methods and systems), US 8,843,433 (Integrated search and adaptive discovery), and the 2012/0254xxx sibling publications, all sharing the 2011‑03‑29 priority — are not separate prior art for a § 103 combination against a claim entitled to that priority. I use them only to corroborate what the family specification discloses, never as combining references. (They appear in Google's "Cited By (10)" list for family member US 2012/0254075 — see patents.google.com/patent/US20120254075A1 — which confirms they are family, not art.)

3. Claim‑by‑claim § 103 analysis

Independent claim 1 (method)

Claim 1 resolves to three elements: (i) a contextual scope function establishing a contextual neighborhood within a fuzzy network‑based structure; (ii) a discovery function generating a recommendation according to the contextual neighborhood and an inference from a plurality of usage behaviors; (iii) the neighborhood being based on a selected object.

Result: claim 1 is, on its face, a predictable combination of two teachings from a single family of documents — (a) fuzzy‑network degree‑of‑separation subsetting and (b) behavior‑inferred, context‑keyed recommendation. Because both teachings sit in the same specification, motivation is at its strongest: the combination is not merely suggested by the art, it is literally disclosed together.

Independent claim 8 (system; scope "set by a user")

Adds a contextual neighborhood scoping function that establishes the scope at a level set by a user.

• 2007/0156614 discloses that "the preference inferencing algorithm may be tuned by the individual user… The user may also set explicit recommendation tuning controls," and that weighting between expert and personal preferences is user‑adjustable.
• The FREES selection operator is explicitly parameterized by a degree‑of‑separation value ("0.5," "2.4," "2.5") — a settable scope level.
• Motivation: a user‑adjustable relevance radius is the natural interface for the document's own stated trade‑off between in‑context (focused) and out‑of‑context (serendipitous) recommendations — a trade‑off the '742 specification itself describes.

Independent claim 15 (system; affinity vector + expertise inference)

Adds: an affinity vector of values between a user and the neighborhood objects, generated from behavioral inferences, at least one inference being an inference of a level of expertise.

• The '742's own MTAV/MTEV disclosure is the family's disclosure. 2007/0156614 discusses designated subject‑matter experts ("Community subsets, such as subject matter experts, may be designated. Expert opinions… may be encoded… Expert views may be directly inferred"), expert‑vs‑personal weighting, and user community affinity structures.
• Expertise‑inference people matching is independently disclosed in Flinn US 7,739,231 (Mutual commit people matching process) and in the family's expertise‑discovery siblings; for the expertise element specifically, US 7,558,748 (purpose‑based adaptive rendering) and US 2009/0248599 (behavior prediction) provide non‑ManyWorlds support if an examiner wants art outside the patentee's own portfolio.
• Motivation: an affinity vector over a neighborhood is the standard data structure for the recommendation operation already disclosed; adding an expertise dimension is a finite, predictable design choice with an evident benefit (routing content vs. routing people), squarely within KSR.

Dependents of note

• Claim 2 (scope set by the recommendation recipient) — supported by the user tuning controls above.
• Claim 6 / claim 16 (expertise inference; over a plurality of behavior categories) — Flinn 7,739,231 + Table 1 categories of the 2007/0156614 family.
• Claim 7 (recommendation responsive to a search request) — the family spec: "Adaptive recommendations 250 may also be generated in response to direct user requests or queries, including search requests"; the family also published Integrated Search and Adaptive Discovery art.
• Claim 11 (selected object chosen based on recipient access) — current‑object context, above.
• Claim 20 (transforming function generating the fuzzy network from originating structures) — the family spec's structural transformation of flat files/hierarchies/relational/tagging structures, with AECR, SFR, CSFR, FWSO metrics and the mixed folder/view pseudo‑code; 2007/0156614 discloses fuzzy‑network operators and syndication/combination.

4. Motivation to combine — the enumerated rationales

1. Same field, same problem, same solution space. Every core reference addresses adaptive recommendation over a degree‑weighted information graph. KSR permits combination where the references are analogous art and the problem is common.
2. Common ownership/inventorship and literal incorporation. The ManyWorlds documents cross‑reference and incorporate one another by reference ("as if set forth in its entirety"), and the '742's description text tracks the earlier family document. A POSITA reading the '742's own specification is looking at the combination.
3. Design need / market pressure. The '742 specification identifies the motivating problem itself: personalization "inhibit[s] beneficial serendipity." Scoping a recommendation to a neighborhood of adjustable radius is the obvious control knob for trading relevance against serendipity — a "finite number of identified, predictable solutions."
4. Predictable results, no teaching away. Combining degree‑of‑separation subsetting with behavior‑inferred recommendation yields nothing more than the expected result: recommendations drawn from a bounded graph region relevant to the user's current object. The specification asserts a "beneficial serendipity" benefit but offers no comparative data, no unexpected‑results evidence, and no showing that the combination produces anything beyond the sum of its parts.
5. Evidence the art is actually used this way by examiners. In USPTO Appeal No. 2018‑007541 (FD2018007541), the Examiner rejected claims under § 103 relying on "the combined teaching of Rapaport I, Juan, and Flinn," citing Flinn's adaptive‑recommendation function and the privacy control 1152a of FIG. 7. That is a different case and I cannot confirm which Flinn patent is involved, but it demonstrates that Flinn's recommendation disclosures are treated as § 103‑citable art for social‑recommendation claim limitations. (Source: e-foia.uspto.gov/Foia/RetrievePdf?system=BPAI&flNm=fd2018007541-07-02-2020-0.)

5. A stronger threshold theory than § 103, and the patentee's best counters

Flag a § 102 problem. If the '742's description text corresponds to the 2006–2007 ManyWorlds publications (as the internal FIG. 42–45 references suggest), then 2007/0156614 and 2007/0112712 are § 102(b) publications of the same subject matter, and claim 1 — whose elements I mapped to those publications in § 3 above — may be anticipated, not merely obvious. An anticipation theory is cleaner and avoids any motivation‑to‑combine dispute. This should be developed before the § 103 case is argued.

Patentee's best non‑obviousness arguments, and the rebuttals:

• "No reference discloses a contextual scope function." — Rebutted by the FREES selection operator (2007/0156614, "selection operator 822… selected… based on degrees of separation"), unless the patentee persuades a court to read "contextual scope function" more narrowly than the specification supports. Claim‑construction risk here is real but cuts against the patentee, since the specification uses "neighborhood" interchangeably with degree‑of‑separation subsetting.
• "Claim 20's transforming function is the invention." — But claim 20 is a dependent; independent claims 1, 8, 15 do not require transformation. The transformation disclosure is itself in the 2006–2007 art.
• "Claim 15's MTEV is novel." — Expertise inference is disclosed in Flinn 7,739,231 and family expert‑weighting disclosures. The remaining question is whether a reference discloses an affinity vector in which at least one value is derived from an expertise inference; the patentee would attack the linkage, not the existence of expertise inference.
• Objective indicia. I found no evidence in the retrieved record of unexpected results, industry praise, licensing attributable to these claims, or copying — and the specification offers none.

6. Bottom line

The claims are broad and read naturally onto generic graph‑neighborhood retrieval fused with behavioral inference — which is consistent with the reported litigation reading claim 1 onto a generic two‑tower embedding/nearest‑neighbor engine. The distinctive risk for the patent is that its own patentee's 2006–2007 publications appear to disclose the substance of the specification on which the 2011 claims rest.

7. What I could not verify (please treat as open)

1. The actual "References Cited" list of record and whether any of the documents above were cited on the face of the patent or applied by the Examiner. My list is assembled from a third‑party aggregator whose list header was not visible in the retrieved snippet.
2. The '742 file history — no office actions, reasons for allowance, or examiner distinctions were retrieved. If the Examiner allowed over Flinn art on the contextual‑scope element, that reasoning matters.
3. Filing dates of the 2011 publications (Funada 2011/0125700, Carmel 2011/0219011) needed to establish § 102(a)/(e) status.
4. Line‑level verification of MTAV/MTEV, AECR, SFR, CSFR, and FWSO disclosures in the granted Flinn patents (7,526,458/459/464) versus the published 2007/0156614 — I read the latter, not the former.
5. The literal title of provisional 61/469,052 and its exact filing day (truncated in my source), and whether the '742 claims any benefit chain to the 2003–2006 ManyWorlds applications. This determines whether the 2007 publications are § 102(b) art or co‑pending family material.