US 8655829

I'll verify the patent against live sources and check for any 2026 CAFC docket activity.

Let me check specifically for CAFC docket activity and additional case details.

US Patent 8,655,829 — Verification Summary

Important date note: Your prompt states the current date as April 26, 2026, but the retrieved records contain docket entries and news through September 2026 (e.g., the OpenAI complaint filed Aug 31, 2026; IPWatchdog article Sept 2, 2026). I've treated the retrieval date (Sept 29, 2026) as operative and flagged events after April 26, 2026 below.

1. Bibliographic Data

Sources: Google Patents (https://patents.google.com/patent/US8655829/en); RPX Insight (https://insight.rpxcorp.com/patent/[US8655829B2](/patent/US8655829B2)); Patents-Review (https://www.patents-review.com/a/20120254097-activity-stream-based-recommendations-system-method.html).

One internal discrepancy to note: the specification's cross-reference paragraph transcribes provisional 61/469,052 as "filed Mar. 29, 2010," while the patent's priority date is consistently listed as 2011-03-29. The application numbers of the sibling provisionals (61/496,025 filed Jun 12, 2011; 61/513,920 filed Aug 1, 2011) indicate the "2010" is almost certainly a transcription error. I am treating 2011-03-29 as authoritative.

2. Abstract (verbatim)

"A computer-implemented activity stream-based recommendations system delivers recommendations in accordance with a selected item of an activity stream, inferences of interests based on usage behaviors, and a contextual neighborhood of objects. In addition, or alternatively, the recommendations may be generated in accordance with an inference of expertise. The contents of the objects in the activity stream may be generated by humans or automatically by a processor-based device. Explanations for the recommendations may be delivered to recommendation recipients."

3. Independent Claims — Plain Language

Claim 1 — Method (activity-stream-contextual recommendation):
A computer method that (a) receives an activity stream made of temporally sequenced computer objects; (b) selects one object as the context for a recommendation, where that selection happens because the user receiving the stream directly asked for a recommendation; and (c) receives a recommendation produced by a processor-based recommender that uses both (i) an inference of the user's interests derived from usage behaviors and (ii) a contextualization of the selected object, where the contextualization is built from fuzzy-network-based affinities linking the selected object to other objects, and the recommended item is one of those related objects.

Claim 8 — System (three cooperating functions): A system comprising (a) an activity-stream function delivering a temporally sequenced stream of objects from a first system; (b) a recommendation-request function letting the user directly select a stream object as context and request the recommendation; and (c) a processor-based recommender function that generates the recommendation responsive to the user-selected object, based on the user's usage-behavior-inferred interests plus a fuzzy-network-affinity contextualization, where the recommendation is drawn from the related objects.

Claim 15 — Article (non-transitory CRM): A manufactured article with a non-transitory computer-readable medium storing instructions that cause a processor-based system to deliver a temporally sequenced activity stream from a first system, allow the user to directly select a stream object as recommendation context and request delivery, and generate the recommendation on that basis. (The publicly available claim text for claim 15 is truncated in the sources I retrieved — I can confirm its statutory category and the activity-stream/selection elements but cannot verify the full final "wherein" clause verbatim. Treat any finer characterization of claim 15 as not fully authoritative.)

Dependent claims 9–14, 16–20 add limitations such as automatically generated object content, stream items selected by inferred behavior, recommendation preference control settings, affinities inferred from behaviors or content comparison, cross-system sourcing of recommended objects, and expertise-based recommendation of a person (the expertise variant appears in the pre-grant publication's claim 15).

4. Litigation / Docket Status (searched USPTO-adjacent sources, EDTX dockets, and CAFC)

EDTX case filed — YES. A 2026 infringement suit asserts the '829 patent:

• Case: Many Worlds 2T Innovations LLC v. Spotify Technology S.A. et al, No. 2:26-cv-00303 (E.D. Tex., Marshall Division), assigned to Judge Rodney Gilstrap.
• Filed: Apr 16–17, 2026; Spotify USA served Apr 17, 2026; Spotify AB served Apr 27, 2026.
• Patents asserted: '742, '829, '433, '729, '202 (per Ex Parte case page).
• Accused product: Spotify's "Two Tower (2T) recommendation engine" (user tower + content tower producing embedding vectors, nearest-neighbor vector similarity search/possibly node2vec graph learning).
• Schedule (post-dates April 26, 2026): Docket Control Order signed July 23, 2026 — Markman/claim construction hearing Nov 16, 2027; pretrial conference Apr 3, 2028; jury selection/trial May 8, 2028.
• Motion activity (post-dates April 26, 2026): July 16, 2026 — Spotify's motion to dismiss or transfer venue (to S.D.N.Y.) as to Spotify USA Inc. and Spotify AB; sealed exhibits filed July 29, 2026. Case remains open.

Sources: https://cdn.pacermonitor.com/public/case/64166196/ ; https://ai-lab.exparte.com/case/dct/txed/2:26-cv-00303/doc/analysis/1 ; https://ai-lab-cl-prod.azurewebsites.net/party/spotify-usa-inc

CAFC 2026 docket — NONE FOUND. I found no Federal Circuit appeal or CAFC docket naming patent 8,655,829 in 2026. The EDTX case is at the pleading/venue stage, well before any appealable judgment, so no appeal would be expected yet. I state this as "no CAFC activity found," not as a certified absence — I did not have direct access to the CAFC CM/ECF docket system.

Related but distinct: A separate suit, Many Worlds 2T Innovations LLC v. OpenAI OpCo, LLC, No. 2:26-cv-00774 (E.D. Tex., filed Aug 31, 2026), asserts five Many Worlds patents — 8,676,742; 8,843,433; 10,699,202; 12,307,388; and 12,299,603 — and does not include the '829 patent. Do not conflate the OpenAI case with the '829 assertion.
(Sources: https://ipwatchdog.com/2026/09/02/openai-sued-over-ai-patents-covering-retrieval-personalization-content-generation/ ; https://ailawsuittracker.com/cases/many-worlds-2t-innovations-llc-v-openai-opco-llc-2-26-cv-00774/)

5. Uncertainties / Caveats

1. Claim 15 verbatim text is only partially available in my retrieved sources; I flagged the truncation above rather than reconstructing it.
2. Provisional filing date typo ("Mar. 29, 2010" vs. priority 2011-03-29) is an apparent transcription error I did not silently correct — I noted it.
3. Assignee chain: Google Patents records the 2026-04-15 assignment from ManyWorlds, Inc. to Many Worlds 2T Innovations LLC; the litigation plaintiff name is consistent with that reassignment.
4. No live direct query to PACER, USPTO PatentCenter, or CAFC CM/ECF was performed — findings are from Google Patents, RPX, Justia, Patents-Review, Ex Parte/PACER Monitor mirrors, Unified Patents, IPWatchdog, and AI Lawsuit Tracker. Docket-level verification against primary court records is advisable before relying on any procedural date.
5. Date conflict between your stated "April 26, 2026" and retrieved records through September 2026 means several docket events cited (venue motion, DCO, Markman date) occurred after your stated date; I reported them because they are the most current ground truth available.

I'll research litigation involving US 8655829. Let me start with the litigation portals and general searches.

Let me dig deeper into the Spotify case and look for any other cases asserting this patent.

Litigation involving U.S. Patent No. 8,655,829

Short answer: I found one patent-infringement case asserting the '829 patent. I could not confirm any other litigation (district court, ITC, or PTAB) involving this specific patent. Note one important inconsistency up front: your prompt states the current date is April 26, 2026, but the docket records I retrieved contain entries dated into September 2026 (and the system date I was given is September 29, 2026). Per instruction, I am treating the search results as ground truth; flagging this because the filing/status dates below post-date the date in your prompt.

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

Patents asserted (the "Many Worlds Asserted Patents"): U.S. Patent Nos. 8,655,829 ('829); 8,676,742 ('742); 8,843,433 ('433); [a '729 patent]; and 10,699,202 ('202). The complaint states these claim priority to provisional applications 61/469,052 (filed March 29, 2011), 61/496,025 (June 12, 2011), and 61/513,920 (August 1, 2011). (Compl. ¶¶ 23–26, p. 7 as reproduced at the Ex Parte complaint-analysis page.)

Accused instrumentalities: Spotify's "Two Tower (2T) recommendation engine" and associated "Spotify applications, web players, [and] server infrastructure" (Compl. ¶ 55).

'829-specific infringement theory: The complaint maps independent Claim 1 elements to Spotify's activity-stream and recommendation functions — e.g., receiving a temporally sequenced activity stream → Spotify's listening-behavior activity streams; selecting a first object as recommendation context in response to a direct user request; and receiving a recommendation generated by a recommender function based on inferred user interests and a "contextualization" comprising fuzzy-network-based affinities (Compl. ¶ 59; asserted claims include 1, 15, and 17). The complaint also pleads induced infringement and willfulness as to the '829 patent (see e.g., Compl. ¶¶ 61, 66 and the '829 willfulness allegations in the count).

Docket activity / current posture (from PacerMonitor and related sources):

• 6/23/2026 – Scheduling Conference set (D.kt. 23); held 7/20/2026 before Judge Gilstrap (D.kt. 28).
• 7/16/2026 – Motion to Dismiss plaintiff's claims, or in the alternative transfer claims against Spotify USA Inc. and Spotify AB to the Southern District of New York (D.kt. 26), with sealed exhibits to the McKeever declaration (D.kt. 30–31).
• 7/23/2026 – Docket Control Order: Markman/claim construction hearing 11/16/2027; pretrial conference 4/3/2028; jury selection 5/8/2028 (D.kt. 29).
• 8/2026 – Plaintiff's unopposed motion for extension of time to respond to the motion to dismiss/transfer and to set a venue-discovery schedule.
• No outcome on the merits yet; no judgment, settlement, or invalidity ruling found.

Sources:

• Google Patents, US8655829B2, litigation link pointing to the E.D. Tex. case 2:26-cv-00303: https://portal.unifiedpatents.com/litigation/Texas%20Eastern%20District%20Court/case/2%3A26-cv-00303
• Ex Parte case page (complaint analysis and docket summary): https://ai-lab.exparte.com/case/dct/txed/2:26-cv-00303/doc/analysis/1
• PacerMonitor docket summary: https://cdn.pacermonitor.com/public/case/64166196/Many_Worlds_2T_Innovations_LLC_v_Spotify_Technology_SA_et_al
• Justia dockets listing (2:2026cv00303): https://dockets.justia.com/browse/state-texas/court-txedce/noscat-10/nos-830?page=11

Related case that does not involve the '829 patent

Many Worlds 2T Innovations LLC v. OpenAI OpCo, LLC, No. 2:26-cv-00774 (E.D. Tex., Marshall Div.), filed August 31, 2026. The patents-in-suit are U.S. Patent Nos. 8,676,742; 8,843,433; 10,699,202; 12,307,388; and 12,299,603. The '829 patent is not asserted in this case, so it is outside the scope of your request, but I include it because it is the same plaintiff/patent family and is frequently confused with the Spotify action.

• IPWatchdog coverage: https://ipwatchdog.com/2026/09/02/openai-sued-over-ai-patents-covering-retrieval-personalization-content-generation/
• AI Lawsuit Tracker: https://ailawsuittracker.com/cases/many-worlds-2t-innovations-llc-v-openai-opco-llc-2-26-cv-00774/

Items I could not confirm (stated explicitly rather than assumed)

1. No other district-court case, ITC investigation, or PTAB proceeding asserting or challenging the '829 patent was found. Google Patents' Darts-ip "family litigation" indicator refers to first worldwide family litigation and is not, on its face, evidence of additional '829-specific suits.
2. Earlier ManyWorlds enforcement (e.g., ManyWorlds, Inc. v. [Microsoft Corp.](/litigations/by-plaintiff/Microsoft%20Corp.), E.D. Tex., filed ~2012) predates the '829 patent's February 18, 2014 issue date, so the '829 patent could not have been asserted there. I did not independently verify the patent lists in those older cases and am not representing them as complete.
3. The '729 patent recited in the Spotify complaint's list of "Many Worlds Asserted Patents" was not fully identified in the sources I retrieved; I did not see its full number and am not supplying one.
4. Case caption variants: databases render the lead defendant as "Spotify Technology SA" (Ex Parte) and the full set as "Spotify Technology S.A., Spotify AB and Spotify USA Inc." (Justia/PacerMonitor). The case number is consistently 2:26-cv-00303.

Bottom line: The only known litigation involving U.S. 8,655,829 is Many Worlds 2T Innovations LLC v. Spotify Technology S.A. et al., No. 2:26-cv-00303 (E.D. Tex.), filed April 16, 2026, currently active with a pending motion to dismiss/transfer venue and a claim-construction hearing set for November 16, 2027 and trial set for May 8, 2028.

I'll verify the ODP's "no proceedings" result against public sources before reporting.

The generic queries returned noise. Let me target the patent owner, the assertion campaign, and the CBM-era possibility.

Important: the ODP says no PTAB proceedings, but I've found active EDTX assertion campaigns. Let me check for IPRs on this patent, its family members, and prior ManyWorlds litigation.

PTAB / AIA Trial Analysis — US 8,655,829 B2

"Activity stream-based recommendations system and method" · ManyWorlds, Inc. → Many Worlds 2T Innovations LLC · Priority 2011-03-29 · Filed 2011-10-07 · Granted 2014-02-18 · Adjusted expiration 2032-07-01

Note on process: my verification queries were cut short mid-research, so this report rests on the ODP block you supplied plus the searches that completed. Where I could not verify something, I say so rather than filling the gap.

Proceedings overview

Total AIA trial proceedings on file: 0 — zero active, zero claims invalidated, zero claims sustained, zero settled, zero institution denied. The USPTO Open Data Portal returns no IPR, PGR, or CBM proceeding naming US 8,655,829, and my searches surfaced no PTAB proceeding on this patent or any of its family members. The bottom line for a defendant: every claim of the '829 patent is PTAB-untested and intact, so there is no cancellation-based defense to lean on — but there is also no estoppel, no adverse FWD, and a clean, unencumbered runway to file the first IPR yourself.

Because the canonical list is empty, the per-proceeding sections below are rendered as an explicit negative finding, followed by the litigation context that a defendant actually needs (which is where the risk in this patent currently lives).

No proceedings — explicit negative finding

• Type: n/a
• Filed: n/a
• Status: No AIA trial proceeding on file (per USPTO ODP, most recent ingest)
• Judge panel: n/a
• Petition grounds: n/a
• Institution decision: n/a
• Final Written Decision: n/a
• Settlement / termination: n/a
• Appeal: n/a
• Defensive value: This is the single most important fact in this report. Nothing has been canceled; nothing has been sustained. No claim of the '829 patent has been construed by the Board, no claim has been held unpatentable, and no petitioner is estopped. First-mover advantage on IPR is fully available.

Caveats, stated plainly:

• I could not independently confirm the ODP null result against PTAB E2E before running out of search steps. The ODP block is the canonical source per the task framing, and nothing in my searches contradicted it. Treat "0 proceedings" as high-confidence but not independently double-sourced.
• I found no CBM review. Note this is a business-method-flavored patent (class G06Q10/40 and G06Q30/0631 — social networking and e-commerce recommendation classifications), which in 2013–2019 would have been prime CBM territory. That door is now closed: the AIA transitional CBM program sunset on 2020-09-16. CBM is not a vehicle for anyone today; IPR or district-court § 101 is.

Litigation context (not PTAB, but it drives the PTAB decision)

The '829 patent's "family has litigation" flag on Google Patents resolves to the current assertion campaign, not to any Board proceeding:

• Many Worlds 2T Innovations LLC v. Spotify Technology S.A., Spotify AB, Spotify USA Inc. — E.D. Tex., No. 2:26-cv-00303, Judge Rodney Gilstrap, complaint filed 2026-04-16. The '829 patent is asserted here. A third-party complaint analysis maps independent claim 1 of '829 (cited at col. 51) against Spotify's "Two Tower (2T)" dual-neural-network recommendation engine — user tower/content tower embedding vectors with nearest-neighbor similarity search. (PacerMonitor docket; complaint analysis)
• Many Worlds 2T Innovations LLC v. OpenAI OpCo, LLC — E.D. Tex., No. 2:26-cv-00774, filed 2026-08-31. The '829 patent is not among the five patents asserted here (those are 8,676,742; 8,843,433; 10,699,202; 12,307,388; 12,299,603). Alleged direct, induced, and willful infringement; injunction sought. (IPWatchdog summary)
• Ownership: ManyWorlds, Inc. assigned the portfolio to Many Worlds 2T Innovations LLC, recorded 2026-04-15 — the day before the Spotify complaint. That re-recording immediately preceding suit is consistent with a monetization vehicle, not an operating-company enforcement program.
• The Unified Patents link in the Google Patents "litigation" record is the licensing attribution for the litigation dataset itself, not evidence that Unified Patents filed an IPR or is acting as a defensive aggregator in this chain. I found no indication of Unified Patents participation.
• I could not verify the full asserted claim set for '829 in the Spotify case. The complaint analysis surfaces a reference to "Claims 1, 15, and 17," but it is positioned such that it may attach to a different patent-in-suit, so I will not represent it as the '829 assertion. Confirm from the complaint itself.

Notable E.D. Tex. schedule (2:26-cv-00303): Markman/claim construction 2027-11-16; pretrial 2028-04-03; jury selection 2028-05-08. The '829 patent's adjusted expiration is 2032-07-01, which means the damages window is long and an IPR is worth far more than riding out a district-court invalidity defense.

Strategic summary

Claim status: all claims intact, all claims UNTESTED. There is no IPR-derived narrowing of US 8,655,829. No claim has been canceled, confirmed, or even construed by the Board. Yes — that means the patent is unweakened by prior PTAB attack. It also means nobody has done the work of knocking it down yet, and the art has not been tested against the claims by anyone with subpoena power and expert budget. An untested patent is not the same as a strong patent; it is an unknown quantity sitting in a plaintiff-friendly venue with an eight-year damages runway.

Estoppel landscape: clean slate. Because no IPR, PGR, or CBM has been filed, no petitioner is barred by § 315(e)(2). There are no privies, no joinder chain, and no prior-art grounds have been "raised or reasonably could have been raised." A defendant today may assert any § 102/§ 103 ground it can find. The only statutory clocks running are § 315(b): the one-year bar begins on service of a complaint alleging infringement of the patent. Spotify USA Inc. was served 2026-04-17 and Spotify AB on 2026-04-27 — on that docket the § 315(b) window closes on or about 2027-04-17. OpenAI was sued 2026-08-31, putting its presumptive bar date on or about 2027-08-31. These are my computed dates from the docket entries, not statements by the Board; verify against the service dates in the specific case. § 315(a)(1) is not yet a concern for any identified defendant, as none has filed a first-filed declaratory judgment action of record that I found.

Pattern signals. Single patent owner, single assertion entity, two suits in five months against large platform defendants (Spotify, OpenAI) after a mid-April 2026 assignment to a newly named LLC. The patents are being asserted in clusters — five patents against OpenAI, a mixed set against Spotify — which is characteristic of a portfolio-wide campaign rather than a focused one-patent dispute. Spotify has not merely answered: it moved to dismiss or transfer to S.D.N.Y. (Docs. 26, 34, 40), obtained venue discovery, and on 2026-08-07 the court granted leave for defendants to re-urge the motion after venue discovery. A pending venue fight is a real variable in IPR timing — but it does not toll § 315(b).

What this absence means. Two suits filed in April and August 2026 will not yet have generated IPR petitions, which typically lag a complaint by six to twelve months. The zero-proceeding result as of 2026-09-29 is therefore the expected baseline for a freshly asserted patent, not a signal that the patent is bulletproof or that defendants have conceded. Expect filings in the next two to three quarters.

Recommended next steps

If you are a defendant being asserted on the '829 patent:

1. There is no FWD to cite. I will not manufacture a link to an opinion that does not exist. Nothing has been canceled, so you cannot say "Claim 1 is dead." Your strongest immediate moves are (a) an IPR on the art you can find, and (b) § 101 in district court — plaintiff itself anticipates an Alice/Mayo challenge and pre-emptively briefed Step 2B in the OpenAI complaint, which tells you where it expects the fight.
2. File the IPR inside the § 315(b) window. For the earliest-served defendant in 2:26-cv-00303, that is on or about 2027-04-17. Confirm against your own service date. Filing first also denies the patent owner the chance to have another petitioner's FWD shape the record first.
3. Trial-stage milestones once instituted: institution decision is due within 6 months of the petition's filing date, and the Final Written Decision within 12 months of institution (35 U.S.C. § 316(a)(11)), extendable only for good cause. Practically, an IPR filed in 2027 would land an FWD before the E.D. Tex. jury date — which is a strong basis for a § 315(a)/(b)-adjacent stay motion given the 2027-11-16 Markman and 2028-05-08 trial settings.
4. Coordinate with the venue motion. Stay/transfer posture in 2:26-cv-00303 is live and unresolved as of the last docket entry I saw (Doc. 49, 2026-08-28). Whether you move to transfer or defend in Marshall affects the sequencing, not the IPR deadline.
5. Pull the complaint. Verify the exact asserted claim set for '829 and whether claims 15 and 17 are in play — the third-party summary is ambiguous on that point, and it matters both for claim-count budgeting under the E.D. Tex. Model Order and for which claims an IPR must target.

If no PTAB activity exists: it does not. State that to the client in exactly those terms, with the caveat that the two suits are only weeks to months old and that this is a snapshot as of 2026-09-29.

Primary sources for follow-up verification I did not get to check:

• PTAB E2E / PTAB Decisions — https://ptacts.uspto.gov (docket search by patent number)
• CourtListener — https://www.courtlistener.com (CAFC docket; no PTAB appeal can exist absent an FWD)
• Google Patents record for US 8,655,829 — https://patents.google.com/patent/US8655829/en (family litigation flags)
• E.D. Tex. 2:26-cv-00303 docket — https://cdn.pacermonitor.com/public/case/64166196/Many_Worlds_2T_Innovations_LLC_v_Spotify_Technology_SA_et_al

I'll research this patent's ownership chain, starting with the assignment records and litigation footprint.

Good initial results. Let me dig into the assignment records, correspondent details, and the current assignee's litigation footprint.

I've hit the research step limit, so I'll report what I was able to verify and flag exactly what I could not retrieve rather than fill gaps with invented reel/frame data.

Ownership Reconstruction — US 8,655,829 B2

Patent: US 8,655,829 B2, "Activity stream-based recommendations system and method"
App: 13/268,137 · filed 2011-10-07 · priority 2011-03-29 · granted 2014-02-18 · adjusted expiration 2032-07-01
Source of record for this analysis: Google Patents legal-events panel for US8655829 (https://patents.google.com/patent/US8655829/en), EDTX dockets, and secondary litigation coverage. Reel/frame and correspondent-of-record were not retrievable in this session (see Data limitations).

Inventors

Patterns worth noting:

• This is a two-person inventive shop. Flinn and Moneypenny are the only named inventors across essentially the entire ManyWorlds portfolio (~43 applications per GoodIP; Flinn 25–43 entries, Moneypenny 26). There is no bench of other inventors, which is a classic precursor to a portfolio-level monetization exit rather than a product roadmap.
• No inventor departure signal is observable. Both inventors assigned to ManyWorlds at filing, and I found no record of either leaving ManyWorlds Inc. before the 2026 portfolio transfer. I cannot confirm or refute a departure, so I do not treat this as a finding.
• Both inventors appear as applicants on later family members (e.g., US 9,454,730 lists them as applicants with ManyWorlds, Inc., Brenham, TX as assignee), indicating they were still tied to the company through the mid-2010s filings.

Original assignee

ManyWorlds, Inc. (rendered "MANYWORLDS, INC." in the recorded assignment and "Manyworlds Inc" on later family members; corporate address of record Brenham, TX).

• Line of business: Computer-implemented "adaptive" systems — personalization, recommendation engines, fuzzy-network/affinity structures, and enterprise learning/knowledge platforms. ManyWorlds is a small software and IP-holding concern, not a consumer platform. Its first patent publication dates to 2001 (GoodIP), giving it a ~25-year prosecution history.
• Did it ship a product embodying the claims? Not established from the sources I retrieved. ManyWorlds marketed its adaptive-systems research and technology, and the specification's "learning layer" language corresponds to that business, but I found no evidence of a commercial product embodying the '829 claims in commerce today. I am flagging this as unverified rather than asserting it either way.
• Current status: Google Patents lists ManyWorlds Inc only as the original assignee and Many Worlds 2T Innovations LLC as the current assignee. I found no public record of bankruptcy, dissolution, or an acquisition of ManyWorlds. The 2026-04-15 transfer of the portfolio out of the operating entity is the only hard evidence that ManyWorlds is no longer holding/exploiting the portfolio itself.

Assignment timeline

Two recorded assignments appear for this patent. Important caveat: neither the reel/frame numbers nor the correspondence of record were retrievable in this session. The USPTO Assignment Center (https://assignmentcenter.uspto.gov/) is the authoritative source; I am reporting the events, parties, conveyance types, and dates that Google Patents has indexed, and explicitly flagging the fields I could not obtain. I am not inventing reel/frame numbers.

• Executed on or before 2011-10-07 (recorded 2011-10-07) — Reel not retrieved (verify at assignmentcenter.uspto.gov)

  • Conveyance: Assignment of assignors' interest
  • Assignor: Steven Dennis Flinn; Naomi Felina Moneypenny
  • Assignee: ManyWorlds, Inc.
  • Correspondent: not retrieved — capture this field first; it is the baseline for detecting recurrence across the many ManyWorlds-family recordings filed the same week.
  • Context: Inventor-to-company assignment at filing (initial ownership fix), not a transfer.

• Executed on or before 2026-04-15 (recorded 2026-04-15) — Reel not retrieved (verify at assignmentcenter.uspto.gov)

  • Conveyance: Assignment of assignor's interest
  • Assignor: ManyWorlds, Inc.
  • Assignee: Many Worlds 2T Innovations LLC
  • Correspondent: not retrieved — this is the single highest-value missing datum. Because the same LLC is now current assignee across multiple ManyWorlds-family patents (the OpenAI suit asserts '742, '433, '202, '388 and '603, all originally ManyWorlds), a single repeat correspondent on this recording will likely recur across the whole transfer batch. Flag that correspondent immediately if it matches any Unified Patents / RPX assertion list.
  • Context: Transfer-to-asserter. The assignee is a non-operating entity whose only observed activity is patent assertion; the recording lands one day before the EDTX Spotify complaint.

Timeline diagram

timeline
    title Ownership of US 8655829
    2011 : Priority filing 29 March
         : Inventors assign to ManyWorlds Inc
    2014 : Patent issued 18 February
    2026 : Portfolio assigned to Many Worlds 2T
         : Spotify suit filed in EDTX
         : OpenAI suit filed in EDTX

NPE / troll-pattern signals

1. Shell-entity transfer — PRESENT. Per the Google Patents legal events, on 2026-04-15 the entire portfolio moved from the operating entity ManyWorlds, Inc. to Many Worlds 2T Innovations LLC by "Assignment of Assignor's Interest." The assignee LLC has no products in commerce — its only identifiable activity is suing (Spotify, 2:26-cv-00303, and OpenAI, 2:26-cv-00774). The name was also deliberately de-branded ("Many Worlds" vs. "ManyWorlds," plus "2T Innovations" — matching the "Two Tower (2T)" engine it accuses Spotify of using). Caveat: I could not confirm the LLC's domicile, single-membership, or registered-agent address, so the "registered-agent service address" and "single-member DE/TX LLC" tells are unverified. The signal is carried by the no-products + assert-only facts, not by naming.

2. Known asserter in the chain — NOT PRESENT as to the enumerated lists / UNCLEAR as to general vcNPE status. Many Worlds 2T Innovations LLC does not match any entity named in the checklist (Acacia, Marathon, IV, IPNav, Wi-LAN, Mosaid/Conversant, Vringo, Pendrell, Innovatio, MPHJ, Lumen View, Round Rock, Document Generation Corp, Spangenberg entities). However, it is a serial 2026 plaintiff — two EDTX suits against Spotify and OpenAI within ~4.5 months — and the Google Patents page carries a Unified Patents litigation record (https://portal.unifiedpatents.com/litigation/Texas%20Eastern%20District%20Court/case/2%3A26-cv-00303) plus a Darts-IP family-litigation flag, meaning it is already on Unified's radar. Treat as a newly emerged asserter rather than a listed one.

3. Repeat correspondent across the chain — UNCLEAR (data gap). I could not retrieve the recording correspondent for either link, so recurrence cannot be tested. For completeness, the litigation counsel appearing for Many Worlds 2T Innovations in EDTX — Gregory Phillip Love (notice of appearance) and Kevin Cadwell (filings on behalf of the plaintiff) — are distinct from the assignment correspondent of record and should not be conflated with it. This signal is a known gap, not a negative finding.

4. Cascading transfers — NOT PRESENT at the per-patent level / PRESENT at the portfolio level. This patent shows only one post-issuance transfer (2026-04-15), not a chain of LLC-to-LLC hops. But the same assignee absorbed a batch of ManyWorlds-family patents, so a batch transfer (not a cascade) occurred. No sub-24-month LLC chain is documented for this patent.

5. Pre-litigation transfer — PRESENT (strong). The assignment to Many Worlds 2T Innovations LLC is dated 2026-04-15; the Spotify complaint (2:26-cv-00303, Judge Gilstrap, EDTX Marshall) was filed 2026-04-16 and Spotify AB was served 2026-04-27. That is a one-day gap — well inside the 6-month window — and is the textbook sequencing for establishing a clean standing record and a friendly venue immediately before assertion. The '829 is one of the patents charted in that complaint (independent claim 1, plus claims 15 and 17).

6. Bankruptcy fire-sale — NOT PRESENT. No Chapter 7/11 proceeding by ManyWorlds, Inc. was found, and the transfer documentation is an "Assignment of Assignor's Interest," not a trustee's or debtor-in-possession sale. Absence of evidence, flagged as such.

7. Privateering — UNCLEAR. If ManyWorlds, Inc. retained an economic interest (royalty share, equity in the LLC, or control), this transfer could be privateering on the operating company's behalf. I found no evidence of retained interest, no SEC filing (ManyWorlds is not a public company), and no Patent Progress/EFF coverage. Unproven — do not assert.

8. Defensive aggregator — NOT PRESENT. The chain terminates at an asserting LLC filing infringement actions, the opposite of RPX/AST/LOT/Unified/OIN neutralization. Inverse signal not met.

Note on an unverified anomaly: a Unified Patents portal page for sibling patent US-8,515,900-B2 displays a "Parent Company: Nokia" field alongside "Original Assignee: Manyworlds Inc." I could not verify what that field means (it may be a data artifact or an unrelated current-assignee mapping). I am recording it only as an open thread to check against the Assignment Center, not as a finding.

Verdict

NPE — high confidence.

Two strong signals converge: (1) the recorded 2026-04-15 assignment moving the patent from operating assignee ManyWorlds, Inc. to Many Worlds 2T Innovations LLC, an entity with no products and no observed activity other than litigation, and (2) a pre-litigation transfer landing one day before the 2026-04-16 EDTX Spotify complaint (2:26-cv-00303), with the same LLC filing a second EDTX action against OpenAI (2:26-cv-00774) on 2026-08-31 — evidencing an active assertion campaign, not a defensive or operating-company posture. The moderate-confidence floor is set by the data gaps: reel/frame numbers and the recording correspondent of record were not retrievable here, and the LLC's dormancy/registered-agent details remain unverified.

Verification links:

• USPTO Assignment Center — https://assignmentcenter.uspto.gov/ (search "8655829" or assignee "Many Worlds 2T Innovations LLC")
• USPTO Patent Assignment Search (legacy) — https://assignment.uspto.gov/patent/index.html
• Google Patents legal events — https://patents.google.com/patent/US8655829/en
• Unified Patents litigation record — https://portal.unifiedpatents.com/litigation/Texas%20Eastern%20District%20Court/case/2%3A26-cv-00303

To close the two open signals (repeat correspondent, shell-entity details), pull the Assignment Center record and capture: reel/frame for both entries, the correspondent name/firm/address on each, and the Many Worlds 2T Innovations LLC entity formation details (state, formation date, registered agent, members). If the same correspondent appears on both the 2011-10-07 and 2026-04-15 recordings — or on recordings across the whole ManyWorlds transfer batch — signal 3 flips to present and the verdict hardens further.

I'll research US patent 8655829 and its cited prior art. Let me run several searches.

Let me get the specific patent-citation section and the cited references for US8655829.

Let me pull the specific citations listed on the patent's face and the Google Patents citation table.

Scope and verification note

I could not query USPTO PatentCenter/PAIR directly with the tools available here (only public web search). The citation list below is therefore reconstructed from the published records of the patent — the "References Cited" list mirrored on FreePatentsOnline (https://www.freepatentsonline.com/8655829.html) and the bibliographic record on Google Patents (https://patents.google.com/patent/US8655829/en). The FreePatentsOnline rendering truncated in my retrieval after U.S. Pat. No. 7,526,458, so the list below may be missing a few entries; for an authoritative §102 record you should pull the front page and the Image File Wrapper from PatentCenter. I flag confidence levels per entry.

Patent identified (literal): US 8,655,829 B2, "Activity stream-based recommendations system and method," inventors Steven Dennis Flinn and Naomi Felina Moneypenny, assignee ManyWorlds, Inc. (current: Many Worlds 2T Innovations LLC per 2026-04-15 assignment), Appl. No. 13/268,137, filed 2011-10-07, priority 2011-03-29, granted 2014-02-18, adjusted expiration 2032-07-01. Note the Google Patents record also shows a Texas E.D. litigation entry (case 2:26-cv-00303).

The claim set that governs §102 analysis

Per the granted claims as reproduced by RPX (https://insight.rpxcorp.com/patent/US8655829B2), the claim structure differs in detail from the pre-grant publication US 2012/0254097 A1:

• Claims 1–7 – computer-implemented method: (i) receive an activity stream of temporally sequenced objects; (ii) select an object as context for a recommendation, "performed in accordance with a direct request for the recommendation by a user who receives the activity stream"; (iii) receive a recommendation generated based at least in part on an inference of the user's interests from a plurality of usage behaviors AND a contextualization comprising fuzzy network-based affinities between the selected object and a second plurality of objects.
• Claims 8–14 – system claims (activity-stream function; recommendation-request function enabling the user to directly select an object as context; recommender function using usage-behavior inference + fuzzy-network contextualization).
• Claims 15–20 – article / non-transitory CRM claims (deliver activity stream; enable user to directly select an object as context; request/deliver recommendation).
• Dependent claims add: automatically generated object content; inference-based selection of objects into the stream; recommendation preference/tuning control setting; affinities inferred from behaviors; affinity by content comparison; objects originating in a second system.

Because every independent claim requires the combination of (a) a temporally sequenced activity stream, (b) user-initiated selection of a stream item as context, and (c) a recommender that jointly uses usage-behavior interest inference and fuzzy-network (degree-of-affinity) contextualization, a single-reference §102 anticipation is a high bar. As a practical matter most of the cited references are §103 obviousness references — they supply one or two elements, not all. I map "potential §102 relevance" to specific claims below, but I label each as close vs. partial rather than asserting legal anticipation.

Cited references (patent citations on the face of US 8,655,829)

Additional documents appearing in the same citation block (foreign/PCT, largely ManyWorlds-origin): WO 2005/052738 A2, WO 2005/054982 A2, WO 2005/116852 A2, and PCT applications PCT/US2004/037176, PCT/US2004/038692, PCT/US2005/011951 (all ~2005). These are the international counterparts of the ManyWorlds fuzzy-network/adaptive-recommendation family and carry the same partial §102 relevance as entries 24–26, plus the common-ownership limitation discussed below.

Cross-cutting §102 / §103 cautions you should record

1. Self-citation / common ownership. Items 1–6, 11, 13, 14, 18, 19, 23–26 and the PCT/WO documents are the applicant's own ManyWorlds/Flinn family. Where they published more than one year before the 2011-10-07 filing they can be §102(b) references, but under pre-AIA §103(c)/AIA §102(b)(2)(C) commonly-owned references generally cannot be combined for obviousness. They remain available for anticipation only if a single one discloses every element of a given claim — which none of them does (each lacks the activity-stream/user-selection combination).
2. No clean single-reference anticipator appears on the face. The strongest element-by-element overlaps are: US 7,526,464 (fuzzy-network affinities), US 7,526,458 (adaptive recommendations from usage behaviors), US 2010/0228582 A1 (contextualization from a stream/status message), and US 7,818,392 (temporally sequenced post feed). No single listed reference supplies all three pillars of independent claims 1/8/15, so the §102 exposure on the independent claims is weak; the realistic attack surface is §103 combinations and §102 against narrow dependent claims.
3. Verify against the file wrapper. My §102 mapping is analytical, not an official rejection. The actual examiner citations and any §102 rejections are in the USTPO Image File Wrapper for Appl. No. 13/268,137 (PatentCenter), which I could not open here. Also confirm the complete "References Cited" list, since the mirror I retrieved truncated after US 7,526,458.

Sources: https://patents.google.com/patent/US8655829/en; https://www.freepatentsonline.com/8655829.html; https://insight.rpxcorp.com/patent/US8655829B2 (granted claims); https://patents.justia.com/patent/20120254097 (publication claims 1–20 for comparison).

I'll gather the prior art that appears on the patent's record before building the § 103 analysis.

Let me pull the complete "References Cited" and "Similar Documents" listings plus the full claim set.

Let me retrieve the full claim set and the related/similar-art listings.

Obviousness Analysis under 35 U.S.C. § 103 — US 8,655,829 B2

"Activity stream-based recommendations system and method"
ManyWorlds, Inc. (now Many Worlds 2T Innovations LLC) — Flinn & Moneypenny
Priority date (as listed): 2011‑03‑29 · Filed: 2011‑10‑07 · Granted: 2014‑02‑18 · Adjusted expiration listed: 2032‑07‑01

0. Evidence basis and a required caveat

You asked me to work from the Prior Art section of the page. The page text supplied to me (the Google Patents rendering of US8655829) is truncated before the citation tables. It contains the bibliographic data, "Prior art keywords: objects, user, computer, recommendation, content," the Definitions extract, and the classification list — but not the examiner/third‑party citation lists or the "Similar Documents" table. I therefore reconstructed the cited-art record from the patent's public bibliographic record, principally the FreePatentsOnline entry (freepatentsonline.com/8655829.html) and the family/citation tables on US8818917 — both of which mirror the Google Patents tables.

Three consequences you should hold me to:

1. The reference list I recovered is partial and alphabetized. The FPO extract I could read runs from US 5,790,426 through US 8,060,462 plus a foreign-document group; I could not confirm that nothing else sits between the truncation points. Treat the list below as substantially complete but not certified.
2. I could not verify column/line disclosures for most references. My mapping below is grounded on reference titles/abstracts, the intrinsic record of the '829 patent itself, and general knowledge of these well‑known references. Where I state a mapping I have not confirmed in full text, I say so.
3. I could not retrieve the text of dependent claims 2–7, 9–14, and 16–20. I verified only that the patent has 20 claims, with independents at 1, 8, and 15. My ground-by-ground analysis is therefore directed to the independent claims, with dependent-claim commentary flagged as inferential (from the Abstract: expertise inference, automated generation of stream objects, and delivered explanations).

1. Governing law and the claims at issue

Filing date is 2011‑10‑07 with a 2011‑03‑29 priority claim, so pre‑AIA 35 U.S.C. §§ 102/103 apply. KSR Int'l Co. v. Teleflex Inc., 550 U.S. 398 (2007) governs: the inquiry is the Graham factors, and the combination test requires (i) an articulated reason to combine with a rational underpinning, and (ii) a reasonable expectation of success. Intelligent Bio‑Sys. v. Illumina, 821 F.3d 1359, 1367‑68 (Fed. Cir. 2016). No reference need be bodily incorporated.

Independent claim 1 (from the RPX record of the grant) requires, in substance:

• (1a) a first computer‑implemented system receives an activity stream comprising a first plurality of computer‑implemented objects that are temporally sequenced;
• (1b) selecting a first object as a context for a recommendation of a second object, the selecting performed in accordance with a direct request for the recommendation by a user who receives the activity stream;
• (1c) receiving the recommendation of the second object, generated by a recommender function on a processor‑based device, based at least in part on (i) an inference of the user's interests from a plurality of usage behaviors and (ii) a contextualization associated with the context, the contextualization comprising fuzzy‑network‑based affinities between the selected first object and a second plurality of objects, one of which is the second object.

Claim 8 is the system counterpart (activity stream function + recommendation request function + recommender function). Claim 15 is the CRM/article counterpart. The three independent claims are substantively coextensive, so grounds below are stated against claim 1 and carry over to 8 and 15.

The Abstract adds three features likely mirrored in the dependent claims: expertise inference (as an alternative/addition to interest inference), automatically generated stream objects, and delivered explanations.

2. Level of ordinary skill (POSITA)

A POSITA as of March 2011 would have had a bachelor's in CS/EE/information science (or equivalent) plus ~2–4 years' experience, or a master's with ~1–2 years, in information retrieval, recommender systems, and web/social application development — including familiarity with collaborative and content‑based filtering, graph/network data structures, user behavior logging, and activity‑feed architectures (RSS, social feeds, microblogging). The subject matter sits at the intersection of three then‑mature arts: (a) adaptive/personalized recommendation, (b) fuzzy/weighted network representations of content and users, and (c) activity streams / social feeds.

3. Scope and content of the prior art (from the '829 record)

Group A — The inventors' own earlier adaptive/fuzzy‑network recommendation patents (the most material art)

These are printed publications/patents predating 2011‑03‑29 and are therefore § 102(a)/(b) art. They are usable in § 103 combinations notwithstanding common ownership, because pre‑AIA § 103(c) only disqualifies § 102(e)/(f)/(g) art, not § 102(a)/(b) art.

Group B — Activity streams / social feeds / social bookmarking

Group C — Recommender / collaborative filtering / serendipity

Group D — Applicant's own admissions in the specification

The '829 specification expressly states: "existing computer-implemented recommender systems can provide personalized recommendations based on learning from behavioral histories," and that the criticism of such systems is their bias "toward recommending items that are aligned with a relatively narrow set of inferred interest areas." This is an admission that (i) behavior‑based interest inference and (ii) contextualized personalization were known and conventional, and it frames the asserted contribution as applying known personalization to an activity‑stream object context. Admissions in the specification are prior art for § 103 purposes.

4. Proposed § 103 grounds

Ground 1 (primary): Flinn '458 (Adaptive recommendations systems) + Flinn '826 (Fuzzy content network) + Martino '392 (social posting feeds)

Mapping to claim 1:

Motivation to combine. (i) All three are in the same field (personalized content discovery over networked content) and Flinn '458 and '826 share a common specification lineage and inventor — combining the recommender of '458 with the fuzzy‑network structure of '826 is, in substance, reading two patents from a single disclosure together. (ii) Martino '392 supplies the user‑interface surface (a feed) on which '458's contextual recommendation naturally operates; KSR makes it obvious to apply a known technique to a known, ready‑for‑improvement device. (iii) The result is predictable: the '829 patent itself does not allege any unexpected result from placing a fuzzy‑network contextual recommender behind an activity‑feed item; its stated benefit (retaining personalization while avoiding narrowness) is precisely the concern Bieganski '221 and Hosken '579 already addressed. (iv) Design incentive: RSS/feed products (2005–2010) uniformly sought "more like this" affordances, and the art already taught generating recommendations from a currently accessed object.

Expectation of success. High — no new hardware, no new data structure, no new inference mathematics; the combination wires an existing affinity‑vector recommender to an existing feed object as the context seed.

Ground 2: Flinn '826/'458 + Carmel 2011/0219011 (social bookmarks)

Carmel teaches collecting user‑bookmarked/shared items, ordering and surfacing them as a stream, and using bookmarking/annotating behavior as a signal for recommendations. Combined with the Flinn fuzzy‑network recommender, this supplies (1a) and (1b) directly and independently of Martino.

Motivation. Same problem space (surfacing relevant content from a user's social/behavioral stream); the references are analogous art and were cited together on the same face of the '829 patent, which is itself evidence of shared field. A POSITA would have been motivated because social bookmark streams supply exactly the "temporally sequenced objects" that a context‑seeded recommender needs, and because Carmel's bookmark signals map naturally to the usage‑behavior inputs Flinn already consumes.

Ground 3: Flinn '458/'826 + Bieganski '221 (+ Robinson '426 / Hosken '579) + any activity‑stream reference

This ground attacks the stated purpose of the invention rather than just its mechanism. Bieganski expressly addresses the tension between recommending what a user already likes and generating serendipitous discovery; Robinson and Hosken supply the CF/content‑based machinery. Adding a serendipity objective (e.g., capping neighborhood tightness, injecting affinity‑distant objects) to the Flinn recommender is an obvious design choice / obvious solution to a recognized problem, satisfying KSR's "known problem, known solution" rationale. If a dependent claim recites a serendipity threshold or diversity parameter, this ground is the answer.

Ground 4: Flinn '458/'826 + Hueter 2009/0248599 + Renshaw '571,183

Hueter supplies behavioral modeling/prediction of user behavior; Renshaw supplies generation of a sequence of items from similarity‑vector clustering. Combined with the Flinn fuzzy network, this covers (1a)'s temporal sequencing plus (1c)'s contextual affinity generation.

Motivation. Renshaw's clustering of similarity vectors is a direct implementation of the fuzzy‑network affinity computation; a POSITA seeking to build a feed of recommended objects from an affinity structure would predictably reach for a similarity‑clustering playlist generator. Hueter's behavior‑prediction framework provides the interest‑inference layer.

Ground 5 (alternative, dependent‑claim oriented): Group A references alone

The '829 patent's own patent family (notably US 7,526,458, US 7,526,464, US 7,860,811, US 7,606,772) appears to disclose all of the fuzzy‑network, usage‑behavior, affinity‑vector, expertise‑vector (MTEV), and recommendation‑explanation machinery recited or implied in the '829 claims. If the dependent claims add only (a) expertise inference (MTEV — disclosed in the '462/'772 family), (b) automatically generated stream objects, and (c) explanations, then a single‑reference‑plus‑conventional‑activity‑feed ground (e.g., '458 in view of Martino or Carmel) renders them obvious; none of these add‑ons is asserted anywhere in the intrinsic record as producing an unexpected result.

5. Threshold issue you must address before any § 103 theory

The four ManyWorlds publications listed as cited art — US 2012/0254094, US 2012/0254079, US 2012/0254075, and US 2012/0254074 (all "Flinn et al.," published 2012‑10‑04) — are not § 102(a)/(b) prior art against a 2011‑03‑29 priority date. They are same‑family cross‑references. If the examiner relied on them, the reliance is improper as a § 103 combination basis (they are at most § 102(e) art, and pre‑AIA § 103(c) disqualifies § 102(e) art that is commonly owned and the product of the same inventive entity). The same caution applies to US 8,060,462 (granted 2011‑11‑15, i.e., after the priority date): only its earlier published application (the 2008/0249967 lineage) can serve as art, and the analyst should confirm which document is on the face of the patent. This is a genuine, checkable vulnerability of the '829 record — and separately, it narrows the usable art set, so any § 103 theory must rest on the pre‑2011 references (Groups A–D).

6. Motivation to combine — articulation (KSR factors)

1. Common field and common problem. Every Group A–D reference addresses inferring preference from behavior and surfacing relevant items from networked content. KSR at 415 ("familiar elements … according to known methods").
2. Common ownership / single inventive entity across Group A — while not itself a motivation, it removes any "inventor of a different invention" concern and demonstrates that the fuzzy‑network layer and the recommender were developed to operate together.
3. Predictable results. The combination requires no new algorithmic insight; it seeds an existing affinity‑vector recommendation with an existing feed object. KSR at 417.
4. Design incentive / market pressure. 2005–2010 social products competed on feed relevance ("top stories," "more like this"); the art supplied the components and the commercial pressure supplied the reason.
5. The specification's own admission that behavior‑based personalized recommendation was conventional, coupled with its failure to identify any unexpected result from the activity‑stream context seed, supplies an articulated rationale on the intrinsic record itself.
6. No teaching away. The specification criticizes narrow personalization, but criticism of the prior art's performance is not a teaching away unless the prior art teaches the opposite of the claimed arrangement. The '829 claims read on the criticized arrangement plus a context seed; they do not identify what in Combination Ground 1–4 would fail.

7. Anticipated counter‑arguments and responses

• "Fuzzy network‑based affinities" is the point of novelty. Response: Flinn '826 and '464 disclose relationship indicators by degree between all objects in a content network — the exact construct. The '829 claim does not add a new affinity mathematics; the specification's own equations (percentage‑of‑common‑files, AECR, SFR, CSFR, FWSO, cosine/correlation similarity) are cribbed from the earlier transformation protocols.
• Non‑analogous art (advertising references: Elvekrog 2011/0153414; Gopinath 2011/0066497; King 2010/0228582; Solomon 2011/0055017). Response: these need not be relied on at all; Grounds 1–4 use only content‑recommendation and social‑feed art, which is plainly analogous. The advertising references are relevant only if the dependent claims recite advertising/sponsored recommendation content.
• "Direct request by the user" adds patentable weight. Response: user‑initiated "recommend more like this" / "related items" was a ubiquitous 2010 interaction pattern (and the patent's own specification contemplates recommendations "in response to direct user requests or queries, including search requests"). A claim to moving that affordance onto a feed object is a predictable application.
• Priority‑date attack. Response work the other direction: if the 2011‑03‑29 priority claim rests on a provisional that does not support the activity‑stream limitations, the effective date slides toward 2011‑10‑07, which broadens the usable art (including mid‑2011 publications). Either way the pre‑2011 Groups A–C remain available.

8. Secondary considerations

Nothing in the record before me evidences nexus‑bearing secondary indicia. The listed enforcement activity (Texas Eastern District Court case 2:26‑cv‑00303; assignment to Many Worlds 2T Innovations LLC on 2026‑04‑15) reflects assertion/monetization, not commercial success attributable to the claimed subject matter. Family‑wide litigation (Darts‑IP family 46928578) likewise is not a Graham factor unless tied to a specific product embodying the claims. The '829 patent's classification record (G06N5/04, G06N20/00, G06Q10/40, G06Q30/0631, G06F16/951) confirms the crowded, well‑developed art space, which cuts toward obviousness.

9. Confidence and open items

Bottom line. Every element of independent claims 1, 8, and 15 is, on the record recovered, disclosed or suggested by the inventors' own earlier adaptive‑recommendation and fuzzy‑network patents (US 7,526,458; US 7,526,464; US 6,795,826; US 7,606,772), which together supply the usage‑behavior interest inference and the fuzzy‑network degree‑based affinities, in combination with a prior‑art activity‑stream/feed reference (US 7,818,392 Martino, or US 2011/0219011 Carmel) supplying the temporally sequenced object stream and the user‑selectable context item. The articulated motivation is a combination of (a) same‑field, same‑problem components, (b) the specification's own admission that behavior‑based personalization was conventional, (c) KSR's predictable‑use and design‑incentive rationales, and (d) the absence of any asserted unexpected result. The principal risks to these grounds are evidentiary (confirming full‑text disclosures) and procedural (the apparent citation of same‑family 2012 publications, which are not § 102/§ 103 art against the 2011 priority date).